1 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * tree.h (build_index_vector): Declare.
6 * tree.c (build_index_vector): New function.
7 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
8 of units as polynomial, forcibly converting it to a constant if
9 vectorizable_reduction has already enforced the condition.
10 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
11 to create a {1,2,3,...} vector.
12 (vectorizable_reduction): Treat the number of units as polynomial.
13 Choose vectype_in based on the largest scalar element size rather
14 than the smallest number of units. Enforce the restrictions
17 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18 Alan Hayward <alan.hayward@arm.com>
19 David Sherwood <david.sherwood@arm.com>
21 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
22 number of units as polynomial.
24 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29 * target.def (autovectorize_vector_sizes): Return the vector sizes
30 by pointer, using vector_sizes rather than a bitmask.
31 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
32 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
33 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
35 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
36 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
37 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
38 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
39 * omp-general.c (omp_max_vf): Likewise.
40 * omp-low.c (omp_clause_aligned_alignment): Likewise.
41 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
42 * tree-vect-loop.c (vect_analyze_loop): Likewise.
43 * tree-vect-slp.c (vect_slp_bb): Likewise.
44 * doc/tm.texi: Regenerate.
45 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
47 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
48 the vector size as a poly_uint64 rather than an unsigned int.
49 (current_vector_size): Change from an unsigned int to a poly_uint64.
50 (get_vectype_for_scalar_type): Update accordingly.
51 * tree.h (build_truth_vector_type): Take the size and number of
52 units as a poly_uint64 rather than an unsigned int.
53 (build_vector_type): Add a temporary overload that takes
54 the number of units as a poly_uint64 rather than an unsigned int.
55 * tree.c (make_vector_type): Likewise.
56 (build_truth_vector_type): Take the number of units as a poly_uint64
57 rather than an unsigned int.
59 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
60 Alan Hayward <alan.hayward@arm.com>
61 David Sherwood <david.sherwood@arm.com>
63 * target.def (get_mask_mode): Take the number of units and length
64 as poly_uint64s rather than unsigned ints.
65 * targhooks.h (default_get_mask_mode): Update accordingly.
66 * targhooks.c (default_get_mask_mode): Likewise.
67 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
68 * doc/tm.texi: Regenerate.
70 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
75 * omp-general.c (omp_max_vf): Likewise.
76 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
77 (expand_omp_simd): Handle polynomial safelen.
78 * omp-low.c (omplow_simd_context): Add a default constructor.
79 (omplow_simd_context::max_vf): Change from int to poly_uint64.
80 (lower_rec_simd_input_clauses): Update accordingly.
81 (lower_rec_input_clauses): Likewise.
83 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * tree-vectorizer.h (vect_nunits_for_cost): New function.
88 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
89 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
90 (vect_analyze_slp_cost): Likewise.
91 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
92 (vect_model_load_cost): Likewise.
94 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
95 Alan Hayward <alan.hayward@arm.com>
96 David Sherwood <david.sherwood@arm.com>
98 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
99 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
100 from an unsigned int * to a poly_uint64_pod *.
101 (calculate_unrolling_factor): New function.
102 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
109 from an unsigned int to a poly_uint64.
110 (_loop_vec_info::slp_unrolling_factor): Likewise.
111 (_loop_vec_info::vectorization_factor): Change from an int
113 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
114 (vect_get_num_vectors): New function.
115 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
116 (vect_get_num_copies): Use vect_get_num_vectors.
117 (vect_analyze_data_ref_dependences): Change max_vf from an int *
118 to an unsigned int *.
119 (vect_analyze_data_refs): Change min_vf from an int * to a
121 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
122 than an unsigned HOST_WIDE_INT.
123 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
124 (vect_analyze_data_ref_dependence): Change max_vf from an int *
125 to an unsigned int *.
126 (vect_analyze_data_ref_dependences): Likewise.
127 (vect_compute_data_ref_alignment): Handle polynomial vf.
128 (vect_enhance_data_refs_alignment): Likewise.
129 (vect_prune_runtime_alias_test_list): Likewise.
130 (vect_shift_permute_load_chain): Likewise.
131 (vect_supportable_dr_alignment): Likewise.
132 (dependence_distance_ge_vf): Take the vectorization factor as a
133 poly_uint64 rather than an unsigned HOST_WIDE_INT.
134 (vect_analyze_data_refs): Change min_vf from an int * to a
136 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
137 vfm1 as a poly_uint64 rather than an int. Make the same change
138 for the returned bound_scalar.
139 (vect_gen_vector_loop_niters): Handle polynomial vf.
140 (vect_do_peeling): Likewise. Update call to
141 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
142 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
144 * tree-vect-loop.c (vect_determine_vectorization_factor)
145 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
146 (vect_get_known_peeling_cost): Likewise.
147 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
148 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
149 (vect_transform_loop): Likewise. Use the lowest possible VF when
150 updating the upper bounds of the loop.
151 (vect_min_worthwhile_factor): Make static. Return an unsigned int
153 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
154 polynomial unroll factors.
155 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
156 (vect_make_slp_decision): Likewise.
157 (vect_supported_load_permutation_p): Likewise, and polynomial
159 (vect_analyze_slp_cost): Handle polynomial vf.
160 (vect_slp_analyze_node_operations): Likewise.
161 (vect_slp_analyze_bb_1): Likewise.
162 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
163 than an unsigned HOST_WIDE_INT.
164 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
165 (vectorizable_load): Handle polynomial vf.
166 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
168 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
170 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
171 Alan Hayward <alan.hayward@arm.com>
172 David Sherwood <david.sherwood@arm.com>
174 * match.pd: Handle bit operations involving three constants
175 and try to fold one pair.
177 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
179 * tree-vect-loop-manip.c: Include gimple-fold.h.
180 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
181 niters_maybe_zero parameters. Handle other cases besides a step of 1.
182 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
183 Add a path that uses a step of VF instead of 1, but disable it
185 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
186 and niters_no_overflow parameters. Update calls to
187 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
188 Create a new SSA name if the latter choses to use a ste other
189 than zero, and return it via niters_vector_mult_vf_var.
190 * tree-vect-loop.c (vect_transform_loop): Update calls to
191 vect_do_peeling, vect_gen_vector_loop_niters and
192 slpeel_make_loop_iterate_ntimes.
193 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
194 (vect_gen_vector_loop_niters): Update declarations after above changes.
196 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
198 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
199 128-bit round to integer instructions.
200 (ceil<mode>2): Likewise.
201 (btrunc<mode>2): Likewise.
202 (round<mode>2): Likewise.
204 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
206 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
207 unaligned VSX load/store on P8/P9.
208 (expand_block_clear): Allow the use of unaligned VSX
211 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
213 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
215 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
216 swap associated with both a load and a store.
218 2018-01-02 Andrew Waterman <andrew@sifive.com>
220 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
221 * config/riscv/riscv.md (clear_cache): Use it.
223 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
225 * web.c: Remove out-of-date comment.
227 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
229 * expr.c (fixup_args_size_notes): Check that any existing
230 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
231 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
232 (emit_single_push_insn): ...here.
234 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
236 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
237 (const_vector_encoded_nelts): New function.
238 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
239 (const_vector_int_elt, const_vector_elt): Declare.
240 * emit-rtl.c (const_vector_int_elt_1): New function.
241 (const_vector_elt): Likewise.
242 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
245 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
247 * expr.c: Include rtx-vector-builder.h.
248 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
249 directly on the tree encoding.
250 (const_vector_from_tree): Likewise.
251 * optabs.c: Include rtx-vector-builder.h.
252 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
253 sequence of "u" values.
254 * vec-perm-indices.c: Include rtx-vector-builder.h.
255 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
256 directly on the vec_perm_indices encoding.
258 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
260 * doc/rtl.texi (const_vector): Describe new encoding scheme.
261 * Makefile.in (OBJS): Add rtx-vector-builder.o.
262 * rtx-vector-builder.h: New file.
263 * rtx-vector-builder.c: Likewise.
264 * rtl.h (rtx_def::u2): Add a const_vector field.
265 (CONST_VECTOR_NPATTERNS): New macro.
266 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
267 (CONST_VECTOR_DUPLICATE_P): Likewise.
268 (CONST_VECTOR_STEPPED_P): Likewise.
269 (CONST_VECTOR_ENCODED_ELT): Likewise.
270 (const_vec_duplicate_p): Check for a duplicated vector encoding.
271 (unwrap_const_vec_duplicate): Likewise.
272 (const_vec_series_p): Check for a non-duplicated vector encoding.
273 Say that the function only returns true for integer vectors.
274 * emit-rtl.c: Include rtx-vector-builder.h.
275 (gen_const_vec_duplicate_1): Delete.
276 (gen_const_vector): Call gen_const_vec_duplicate instead of
277 gen_const_vec_duplicate_1.
278 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
279 (gen_const_vec_duplicate): Use rtx_vector_builder.
280 (gen_const_vec_series): Likewise.
281 (gen_rtx_CONST_VECTOR): Likewise.
282 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
283 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
284 Build a new vector rather than modifying a CONST_VECTOR in-place.
285 (handle_special_swappables): Update call accordingly.
286 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
287 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
288 Build a new vector rather than modifying a CONST_VECTOR in-place.
289 (handle_special_swappables): Update call accordingly.
291 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
293 * simplify-rtx.c (simplify_const_binary_operation): Use
294 CONST_VECTOR_ELT instead of XVECEXP.
296 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
298 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
299 the selector elements to be different from the data elements
300 if the selector is a VECTOR_CST.
301 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
302 ssizetype for the selector.
304 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
306 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
307 before testing each element individually.
308 * tree-vect-generic.c (lower_vec_perm): Likewise.
310 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
312 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
313 * selftest-run-tests.c (selftest::run_tests): Call it.
314 * vector-builder.h (vector_builder::operator ==): New function.
315 (vector_builder::operator !=): Likewise.
316 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
317 (vec_perm_indices::all_from_input_p): New function.
318 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
319 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
320 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
321 instead of reading the VECTOR_CST directly. Detect whether both
322 vector inputs are the same before constructing the vec_perm_indices,
323 and update the number of inputs argument accordingly. Use the
324 utility functions added above. Only construct sel2 if we need to.
326 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
328 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
329 the broadcast of the low byte.
330 (expand_mult_highpart): Use an explicit encoding for the permutes.
331 * optabs-query.c (can_mult_highpart_p): Likewise.
332 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
333 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
334 (vectorizable_bswap): Likewise.
335 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
336 explicit encoding for the power-of-2 permutes.
337 (vect_permute_store_chain): Likewise.
338 (vect_grouped_load_supported): Likewise.
339 (vect_permute_load_chain): Likewise.
341 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
343 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
344 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
345 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
346 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
347 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
348 (vect_gen_perm_mask_any): Likewise.
350 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
352 * int-vector-builder.h: New file.
353 * vec-perm-indices.h: Include int-vector-builder.h.
354 (vec_perm_indices): Redefine as an int_vector_builder.
355 (auto_vec_perm_indices): Delete.
356 (vec_perm_builder): Redefine as a stand-alone class.
357 (vec_perm_indices::vec_perm_indices): New function.
358 (vec_perm_indices::clamp): Likewise.
359 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
360 (vec_perm_indices::new_vector): New function.
361 (vec_perm_indices::new_expanded_vector): Update for new
362 vec_perm_indices class.
363 (vec_perm_indices::rotate_inputs): New function.
364 (vec_perm_indices::all_in_range_p): Operate directly on the
365 encoded form, without computing elided elements.
366 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
367 encoding. Update for new vec_perm_indices class.
368 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
369 the given vec_perm_builder.
370 (expand_vec_perm_var): Update vec_perm_builder constructor.
371 (expand_mult_highpart): Use vec_perm_builder instead of
372 auto_vec_perm_indices.
373 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
374 vec_perm_indices instead of auto_vec_perm_indices. Use a single
375 or double series encoding as appropriate.
376 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
377 vec_perm_indices instead of auto_vec_perm_indices.
378 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
379 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
380 (vect_permute_store_chain): Likewise.
381 (vect_grouped_load_supported): Likewise.
382 (vect_permute_load_chain): Likewise.
383 (vect_shift_permute_load_chain): Likewise.
384 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
385 (vect_transform_slp_perm_load): Likewise.
386 (vect_schedule_slp_instance): Likewise.
387 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
388 (vectorizable_mask_load_store): Likewise.
389 (vectorizable_bswap): Likewise.
390 (vectorizable_store): Likewise.
391 (vectorizable_load): Likewise.
392 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
393 vec_perm_indices instead of auto_vec_perm_indices. Use
394 tree_to_vec_perm_builder to read the vector from a tree.
395 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
396 vec_perm_builder instead of a vec_perm_indices.
397 (have_whole_vector_shift): Use vec_perm_builder and
398 vec_perm_indices instead of auto_vec_perm_indices. Leave the
399 truncation to calc_vec_perm_mask_for_shift.
400 (vect_create_epilog_for_reduction): Likewise.
401 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
402 from auto_vec_perm_indices to vec_perm_indices.
403 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
404 instead of changing individual elements.
405 (aarch64_vectorize_vec_perm_const): Use new_vector to install
406 the vector in d.perm.
407 * config/arm/arm.c (expand_vec_perm_d::perm): Change
408 from auto_vec_perm_indices to vec_perm_indices.
409 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
410 instead of changing individual elements.
411 (arm_vectorize_vec_perm_const): Use new_vector to install
412 the vector in d.perm.
413 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
414 Update vec_perm_builder constructor.
415 (rs6000_expand_interleave): Likewise.
416 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
417 (rs6000_expand_interleave): Likewise.
419 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
421 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
422 to qimode could truncate the indices.
423 * optabs.c (expand_vec_perm_var): Likewise.
425 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
427 * Makefile.in (OBJS): Add vec-perm-indices.o.
428 * vec-perm-indices.h: New file.
429 * vec-perm-indices.c: Likewise.
430 * target.h (vec_perm_indices): Replace with a forward class
432 (auto_vec_perm_indices): Move to vec-perm-indices.h.
433 * optabs.h: Include vec-perm-indices.h.
434 (expand_vec_perm): Delete.
435 (selector_fits_mode_p, expand_vec_perm_var): Declare.
436 (expand_vec_perm_const): Declare.
437 * target.def (vec_perm_const_ok): Replace with...
438 (vec_perm_const): ...this new hook.
439 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
440 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
441 * doc/tm.texi: Regenerate.
442 * optabs.def (vec_perm_const): Delete.
443 * doc/md.texi (vec_perm_const): Likewise.
444 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
445 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
446 expand_vec_perm for constant permutation vectors. Assert that
447 the mode of variable permutation vectors is the integer equivalent
448 of the mode that is being permuted.
449 * optabs-query.h (selector_fits_mode_p): Declare.
450 * optabs-query.c: Include vec-perm-indices.h.
451 (selector_fits_mode_p): New function.
452 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
453 is defined, instead of checking whether the vec_perm_const_optab
454 exists. Use targetm.vectorize.vec_perm_const instead of
455 targetm.vectorize.vec_perm_const_ok. Check whether the indices
456 fit in the vector mode before using a variable permute.
457 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
458 vec_perm_indices instead of an rtx.
459 (expand_vec_perm): Replace with...
460 (expand_vec_perm_const): ...this new function. Take the selector
461 as a vec_perm_indices rather than an rtx. Also take the mode of
462 the selector. Update call to shift_amt_for_vec_perm_mask.
463 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
464 Use vec_perm_indices::new_expanded_vector to expand the original
465 selector into bytes. Check whether the indices fit in the vector
466 mode before using a variable permute.
467 (expand_vec_perm_var): Make global.
468 (expand_mult_highpart): Use expand_vec_perm_const.
469 * fold-const.c: Includes vec-perm-indices.h.
470 * tree-ssa-forwprop.c: Likewise.
471 * tree-vect-data-refs.c: Likewise.
472 * tree-vect-generic.c: Likewise.
473 * tree-vect-loop.c: Likewise.
474 * tree-vect-slp.c: Likewise.
475 * tree-vect-stmts.c: Likewise.
476 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
478 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
479 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
480 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
481 (aarch64_vectorize_vec_perm_const): ...this new function.
482 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
483 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
484 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
485 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
486 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
487 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
488 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
490 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
491 check for NEON modes.
492 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
493 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
494 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
495 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
497 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
498 the old VEC_PERM_CONST conditions.
499 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
500 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
501 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
502 (ia64_vectorize_vec_perm_const_ok): Merge into...
503 (ia64_vectorize_vec_perm_const): ...this new function.
504 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
505 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
506 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
507 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
508 * config/mips/mips.c (mips_expand_vec_perm_const)
509 (mips_vectorize_vec_perm_const_ok): Merge into...
510 (mips_vectorize_vec_perm_const): ...this new function.
511 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
512 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
513 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
514 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
515 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
516 (rs6000_expand_vec_perm_const): Delete.
517 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
519 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
520 (altivec_expand_vec_perm_const_le): Take each operand individually.
521 Operate on constant selectors rather than rtxes.
522 (altivec_expand_vec_perm_const): Likewise. Update call to
523 altivec_expand_vec_perm_const_le.
524 (rs6000_expand_vec_perm_const): Delete.
525 (rs6000_vectorize_vec_perm_const_ok): Delete.
526 (rs6000_vectorize_vec_perm_const): New function.
527 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
528 an element count and rtx array.
529 (rs6000_expand_extract_even): Update call accordingly.
530 (rs6000_expand_interleave): Likewise.
531 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
532 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
533 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
534 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
535 (rs6000_expand_vec_perm_const): Delete.
536 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
537 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
538 (altivec_expand_vec_perm_const_le): Take each operand individually.
539 Operate on constant selectors rather than rtxes.
540 (altivec_expand_vec_perm_const): Likewise. Update call to
541 altivec_expand_vec_perm_const_le.
542 (rs6000_expand_vec_perm_const): Delete.
543 (rs6000_vectorize_vec_perm_const_ok): Delete.
544 (rs6000_vectorize_vec_perm_const): New function. Remove stray
545 reference to the SPE evmerge intructions.
546 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
547 an element count and rtx array.
548 (rs6000_expand_extract_even): Update call accordingly.
549 (rs6000_expand_interleave): Likewise.
550 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
551 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
553 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
555 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
557 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
558 vector mode and that that mode matches the mode of the data
560 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
561 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
562 directly using expand_vec_perm_1 when forcing selectors into
564 (expand_vec_perm_var): New function, split out from expand_vec_perm.
566 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
568 * optabs-query.h (can_vec_perm_p): Delete.
569 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
570 * optabs-query.c (can_vec_perm_p): Split into...
571 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
572 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
573 particular selector is valid.
574 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
575 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
576 (vect_grouped_load_supported): Likewise.
577 (vect_shift_permute_load_chain): Likewise.
578 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
579 (vect_transform_slp_perm_load): Likewise.
580 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
581 (vectorizable_bswap): Likewise.
582 (vect_gen_perm_mask_checked): Likewise.
583 * fold-const.c (fold_ternary_loc): Likewise. Don't take
584 implementations of variable permutation vectors into account
585 when deciding which selector to use.
586 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
587 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
588 with a false third argument.
589 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
590 to test whether the constant selector is valid and can_vec_perm_var_p
591 to test whether a variable selector is valid.
593 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
595 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
596 * optabs-query.c (can_vec_perm_p): Likewise.
597 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
598 instead of vec_perm_indices.
599 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
600 (vect_gen_perm_mask_checked): Likewise,
601 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
602 (vect_gen_perm_mask_checked): Likewise,
604 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
606 * optabs-query.h (qimode_for_vec_perm): Declare.
607 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
608 (qimode_for_vec_perm): ...this new function.
609 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
611 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
613 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
614 does not have a conditional at the top.
616 2018-01-02 Richard Biener <rguenther@suse.de>
618 * ipa-inline.c (big_speedup_p): Fix expression.
620 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
623 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
626 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
630 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
631 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
632 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
633 cond_taken_branch_cost 3->4.
635 2018-01-01 Jakub Jelinek <jakub@redhat.com>
637 PR tree-optimization/83581
638 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
639 TODO_cleanup_cfg if any changes have been made.
642 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
643 convert_modes if target mode has the right side, but different mode
647 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
648 last argument when extracting from CONCAT. If either from_real or
649 from_imag is NULL, use expansion through memory. If result is not
650 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
651 the parts directly to inner mode, if even that fails, use expansion
655 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
656 check for bswap in mode rather than HImode and use that in expand_unop
659 2017-12-31 Jakub Jelinek <jakub@redhat.com>
662 * config/i386/i386.c (ix86_attribute_table): Remove excess
663 initializer for "shared" attribute.
665 2017-12-30 Tom de Vries <tom@codesourcery.com>
668 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
670 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
671 offload_funcs entries that no longer have a corresponding cgraph_node.
672 Mark the remaining ones as DECL_PRESERVE_P.
673 (output_lto): Call prune_offload_funcs.
675 2017-12-30 Jakub Jelinek <jakub@redhat.com>
677 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
678 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
679 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
680 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
681 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
682 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
683 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
684 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
685 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
686 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
687 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
688 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
689 vaesenc_<mode>, vpclmulqdq_<mode>,
690 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
692 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
694 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
695 _Float<N> and _Float<N>X variants for rounding built-in
697 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
698 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
699 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
700 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
701 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
702 * builtins.c (mathfn_built_in_2): Likewise.
703 * internal-fn.def (CEIL): Likewise.
705 (NEARBYINT): Likewise.
709 * convert.c (convert_to_integer_1): Likewise.
710 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
711 (integer_valued_real_call_p): Likewise.
712 * fold-const-call.c (fold_const_call_ss): Likewise.
713 * gencfn-macros.c (print_case_cfn): Change CFN and operator
714 printers to take a const char * suffix instead of a bool.
715 (print_define_operator_list): Likewise.
716 (fltall_suffixes): New list of suffixes, that include the
717 traditional suffixes as well as all of the _Float<N> and
719 (main): For _Float<N> and _Float<N>X functions, emit both
720 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
721 has the _Float<N> and _Float<N>X case names or operators. The
722 <name>_ALL variant has both the traditional and the
723 _Float<N>/_Float<N>X case names or operators.
724 * match.pd (COPYSIGN optimizations): Provide optimizations for
725 _Float<N> and _Float<N>X types where possible.
726 (MIN/MAX optimizations): Likewise.
727 (sqrt optimizations): Likewise.
728 (rounding optimizations): Likewise.
730 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
732 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
733 gen_int_mode rather than GEN_INT.
734 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
736 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
738 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
739 instead of CONSTANT_P.
740 (gen_vec_series): Likewise.
741 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
743 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
745 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
748 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
750 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
751 (aarch64_ld1x2<VDC:mode>): Likewise.
752 (aarch64_simd_ld1<mode>_x2): Likewise.
753 (aarch64_simd_ld1<mode>_x2): Likewise.
754 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
755 (vld1_s8_x2): Likewise.
756 (vld1_u16_x2): Likewise.
757 (vld1_s16_x2): Likewise.
758 (vld1_u32_x2): Likewise.
759 (vld1_s32_x2): Likewise.
760 (vld1_u64_x2): Likewise.
761 (vld1_s64_x2): Likewise.
762 (vld1_f16_x2): Likewise.
763 (vld1_f32_x2): Likewise.
764 (vld1_f64_x2): Likewise.
765 (vld1_p8_x2): Likewise.
766 (vld1_p16_x2): Likewise.
767 (vld1_p64_x2): Likewise.
768 (vld1q_u8_x2): Likewise.
769 (vld1q_s8_x2): Likewise.
770 (vld1q_u16_x2): Likewise.
771 (vld1q_s16_x2): Likewise.
772 (vld1q_u32_x2): Likewise.
773 (vld1q_s32_x2): Likewise.
774 (vld1q_u64_x2): Likewise.
775 (vld1q_s64_x2): Likewise.
776 (vld1q_f16_x2): Likewise.
777 (vld1q_f32_x2): Likewise.
778 (vld1q_f64_x2): Likewise.
779 (vld1q_p8_x2): Likewise.
780 (vld1q_p16_x2): Likewise.
781 (vld1q_p64_x2): Likewise.
783 2017-12-27 Martin Liska <mliska@suse.cz>
785 PR tree-optimization/83552
786 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
787 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
789 2017-12-27 Tom de Vries <tom@codesourcery.com>
791 * config/nvptx/nvptx.c (nvptx_option_override): Disable
792 -gstatement-frontiers.
794 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
796 PR rtl-optimization/83513
797 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
798 before priority comparison.
800 2017-12-25 Jakub Jelinek <jakub@redhat.com>
803 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
804 ix86_isa_flags2 to ix86_isa_flags.
805 * config/i386/i386-c.c (ix86_target_macros_internal): Test
806 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
807 isa_flags rather than isa_flags2.
808 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
809 and -mavx512bitalg from isa2_opts to isa_opts.
810 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
811 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
812 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
814 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
815 OPTION_MASK_ISA_AVX512F ored with another option similarly to
816 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
817 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
818 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
819 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
820 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
821 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
822 builtins from bdesc_args2 to bdesc_args section.
823 (__builtin_ia32_compressstoreuqi512_mask,
824 __builtin_ia32_compressstoreuhi512_mask,
825 __builtin_ia32_compressstoreuqi256_mask,
826 __builtin_ia32_expandloadqi512_mask,
827 __builtin_ia32_expandloadqi512_maskz,
828 __builtin_ia32_expandloadhi512_mask,
829 __builtin_ia32_expandloadhi512_maskz,
830 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
831 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
832 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
833 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
834 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
835 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
836 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
837 __builtin_ia32_vpshldv_v32hi_maskz,
838 __builtin_ia32_vpopcountb_v64qi_mask,
839 __builtin_ia32_vpopcountw_v32hi_mask,
840 __builtin_ia32_vpshufbitqmb512_mask,
841 __builtin_ia32_vpshufbitqmb256_mask): Add
842 " | OPTION_MASK_ISA_AVX512BW".
843 (__builtin_ia32_expandloadqi256_mask,
844 __builtin_ia32_expandloadqi256_maskz,
845 __builtin_ia32_vpopcountb_v32qi_mask): Add
846 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
847 (__builtin_ia32_expandloadhi256_mask,
848 __builtin_ia32_expandloadhi256_maskz,
849 __builtin_ia32_expandloadqi128_mask,
850 __builtin_ia32_expandloadqi128_maskz,
851 __builtin_ia32_expandloadhi128_mask,
852 __builtin_ia32_expandloadhi128_maskz,
853 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
854 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
855 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
856 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
857 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
858 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
859 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
860 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
861 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
862 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
863 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
864 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
865 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
866 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
867 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
868 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
869 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
870 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
871 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
872 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
873 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
874 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
875 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
876 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
877 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
878 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
879 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
880 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
881 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
882 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
883 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
884 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
885 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
886 __builtin_ia32_vpopcountw_v8hi_mask): Add
887 " | OPTION_MASK_ISA_AVX512VL".
888 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
889 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
890 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
891 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
892 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
893 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
894 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
895 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
896 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
897 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
898 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
899 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
900 _mm_mask_bitshuffle_epi64_mask): Likewise.
901 * common/config/i386/i386-common.c
902 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
903 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
904 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
905 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
906 OPTION_MASK_ISA_AVX512BITALG_UNSET.
907 (OPTION_MASK_ISA2_AVX512F_UNSET,
908 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
909 (ix86_handle_option): For -mno-general-regs-only, clear from
910 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
911 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
912 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
913 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
914 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
915 for moving from ix86_isa_flags2 to ix86_isa_flags.
917 2017-12-23 Jakub Jelinek <jakub@redhat.com>
920 * fold-const.c (struct contains_label_data): New type.
921 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
922 inside of a SWITCH_BODY seen during the walk.
923 (contains_label_p): Use walk_tree instead of
924 walk_tree_without_duplicates, prepare data for contains_label_1 and
927 2017-12-22 Jakub Jelinek <jakub@redhat.com>
930 * tree-iterator.c (alloc_stmt_list): Start with cleared
931 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
935 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
936 enable avx512f explicitly in #pragma GCC target.
937 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
938 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
939 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
940 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
941 __builtin_ia32_vpdpbusds_v8si_mask,
942 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
943 __builtin_ia32_vpdpbusds_v4si_mask,
944 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
945 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
946 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
947 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
948 __builtin_ia32_vpdpwssds_v8si_mask,
949 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
950 __builtin_ia32_vpdpwssds_v4si_mask,
951 __builtin_ia32_vpdpwssds_v4si_maskz): Use
952 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
953 just OPTION_MASK_ISA_AVX512VNNI.
955 2017-12-22 Martin Jambor <mjambor@suse.cz>
958 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
961 2017-12-22 Julia Koval <julia.koval@intel.com>
962 Sebastian Peryt <sebastian.peryt@intel.com>
964 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
965 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
966 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
967 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
968 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
969 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
970 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
971 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
972 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
973 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
974 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
975 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
976 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
977 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
978 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
979 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
980 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
981 _mm256_maskz_popcnt_epi64): New intrinsics.
982 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
983 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
984 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
985 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
986 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
987 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
988 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
989 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
990 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
991 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
992 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
993 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
994 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
995 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
996 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
997 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
998 __builtin_ia32_vpshufbitqmb256_mask,
999 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1000 * config/i386/i386-c.c (__AVX512BITALG__): New.
1001 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1002 (ix86_valid_target_attribute_inner_p): Ditto.
1003 (ix86_expand_args_builtin): Handle new types.
1004 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1005 * config/i386/i386.opt: Add -mavx512bitalg.
1006 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1007 avx512bitalgintrin.h.
1008 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1009 (vpopcount<mode><mask_name>): Add more types.
1010 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1011 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1013 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1015 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1016 Or in OPTION_MASK_ISA_AVX512F_SET.
1017 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1018 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1019 (ix86_handle_option): Adjust for
1020 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1021 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1023 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1024 isa_flag instead of isa_flag2.
1025 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1026 isa_opts2 to isa_opts.
1027 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1030 2017-12-22 Mike Stump <mikestump@comcast.net>
1031 Eric Botcazou <ebotcazou@adacore.com>
1033 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1035 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1038 * config/i386/i386.c (ix86_function_arg_boundary): Return
1039 PARM_BOUNDARY for TYPE_EMPTY_P types.
1042 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1043 if navail is >= dir.len.
1045 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1047 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1050 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1052 PR rtl-optimization/80747
1053 PR rtl-optimization/83512
1054 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1055 succ edge from ENTRY, copy partition from e->dest to the newly
1057 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1058 ENTRY, use BB_PARTITION of its successor block as current_partition.
1059 Don't copy partition when splitting succ edge from ENTRY.
1061 PR tree-optimization/83523
1062 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1063 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1064 (convert_mult_to_fma): Likewise.
1066 PR tree-optimization/83521
1067 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1068 gimple_build_assign without code on result of
1069 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1070 a VIEW_CONVERT_EXPR.
1072 2017-12-21 Andrew Pinski <apinski@cavium.com>
1073 Steve Ellcey <sellcey@cavium.com>
1075 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1076 multi-arch for ilp32.
1078 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1081 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1082 constraints to operand 2.
1083 (*ashl<mode>3_mask_1): Ditto.
1084 (*<shift_insn><mode>3_mask): Ditto.
1085 (*<shift_insn><mode>3_mask_1): Ditto.
1086 (*<rotate_insn><mode>3_mask): Ditto.
1087 (*<rotate_insn><mode>3_mask_1): Ditto.
1089 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1091 * reorg.c (make_return_insns): Reemit each insn with its own location.
1093 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1095 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1096 generation for cases where splatting a value is not useful.
1097 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1098 across a vec_duplicate and a paradoxical subreg forming a vector
1099 mode to a vec_concat.
1101 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1103 * combine.c (simplify_set): Do not transform subregs to zero_extends
1104 if the destination is not a scalar int mode.
1106 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1109 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1110 shared zero if the input has overflowed.
1112 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1114 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1115 support for Cortex-A55 and Cortex-A75.
1117 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1119 * common/config/arm/arm-common.c (compare_opt_names): Add function
1120 comment. Use strcmp instead of manual loop.
1122 2017-12-21 Martin Liska <mliska@suse.cz>
1124 PR gcov-profile/83509
1125 * gcov-dump.c (dump_gcov_file): Do not read info about
1126 support_unexecuted_blocks for gcda files.
1128 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1130 PR rtl-optimization/82973
1131 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1132 (valid_for_const_vector_p): ... this.
1133 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1134 (valid_for_const_vector_p): ... this. Adjust function comment.
1135 (gen_vec_duplicate): Adjust caller.
1136 * optabs.c (expand_vector_broadcast): Likewise.
1137 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1138 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1140 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1141 instead of GET_CODE == CONST_FIXED.
1142 (simplify_subreg): Use CONST_FIXED_P macro instead of
1143 GET_CODE == CONST_FIXED.
1146 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1147 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1148 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1149 (ix86_option_override_internal): Adjust for
1150 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1151 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1152 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1153 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1154 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1155 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1156 def_builtin2 for CET builtins.
1157 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1158 up formatting in IX86_BUILTIN_RDPID code.
1159 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1160 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1161 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1162 ix86_isa_flags2 to ix86_isa_flags.
1163 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1165 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1166 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1167 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1169 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1170 Or in OPTION_MASK_ISA_AVX512F_SET.
1171 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1172 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1173 (ix86_handle_option): Adjust for
1174 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1175 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1177 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1178 Alan Hayward <alan.hayward@arm.com>
1179 David Sherwood <david.sherwood@arm.com>
1181 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1182 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1183 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1184 Track polynomial offsets.
1186 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1187 Alan Hayward <alan.hayward@arm.com>
1188 David Sherwood <david.sherwood@arm.com>
1190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1191 Treat drb->init as a poly_int. Fail if its misalignment wrt
1192 vector_alignment isn't known.
1194 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1195 Alan Hayward <alan.hayward@arm.com>
1196 David Sherwood <david.sherwood@arm.com>
1198 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1200 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1201 (vect_loop_versioning): Take the loop versioning threshold as a
1203 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1204 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1205 versioning_threshold.
1206 (vect_analyze_loop_2): Compute the loop versioning threshold
1207 whenever loop versioning is needed, and store it in the new
1208 field rather than combining it with the cost model threshold.
1209 (vect_transform_loop): Update call to vect_loop_versioning.
1210 Try to combine the loop versioning and cost thresholds here.
1212 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1213 Alan Hayward <alan.hayward@arm.com>
1214 David Sherwood <david.sherwood@arm.com>
1216 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1217 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1218 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1219 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1220 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1221 (group_compare_offset): Update accordingly.
1222 (split_small_address_groups_p): Likewise.
1223 (record_use): Take addr_offset as a poly_uint64 rather than
1224 an unsigned HOST_WIDE_INT.
1225 (strip_offset): Return the offset as a poly_uint64 rather than
1226 an unsigned HOST_WIDE_INT.
1227 (record_group_use, split_address_groups): Track polynomial offsets.
1228 (add_iv_candidate_for_use): Likewise.
1229 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1230 than a HOST_WIDE_INT.
1231 (strip_offset_1): Return the offset as a poly_int64 rather than
1234 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1235 Alan Hayward <alan.hayward@arm.com>
1236 David Sherwood <david.sherwood@arm.com>
1238 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1239 rather than a HOST_WIDE_INT.
1240 * tree.c (get_binfo_at_offset): Likewise.
1242 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1247 rather than a HOST_WIDE_INT.
1248 * tree-sra.c (build_ref_for_offset): Likewise.
1250 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1251 Alan Hayward <alan.hayward@arm.com>
1252 David Sherwood <david.sherwood@arm.com>
1254 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1256 * tree.c (mem_ref_offset): Likewise.
1257 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1258 * builtins.c (get_object_alignment_2): Likewise.
1259 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1260 * gimple-fold.c (get_base_constructor): Likewise.
1261 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1262 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1264 * ipa-polymorphic-call.c
1265 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1266 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1267 (get_ancestor_addr_info): Likewise.
1268 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1269 * match.pd: Likewise.
1270 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1271 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1272 * tree-eh.c (tree_could_trap_p): Likewise.
1273 * tree-object-size.c (addr_object_size): Likewise.
1274 * tree-ssa-address.c (copy_ref_info): Likewise.
1275 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1276 (indirect_refs_may_alias_p): Likewise.
1277 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1278 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1279 (non_rewritable_mem_ref_base): Likewise.
1280 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1281 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1282 * varasm.c (decode_addr_const): Likewise.
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 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1289 offsets for MEM_REFs.
1291 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1292 Alan Hayward <alan.hayward@arm.com>
1293 David Sherwood <david.sherwood@arm.com>
1295 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1296 as a poly_uint64 rather than an unsigned int.
1297 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1299 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1300 Alan Hayward <alan.hayward@arm.com>
1301 David Sherwood <david.sherwood@arm.com>
1303 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1305 (decode_addr_const): Update accordingly.
1307 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1308 Alan Hayward <alan.hayward@arm.com>
1309 David Sherwood <david.sherwood@arm.com>
1311 * tree.h (bit_field_size, bit_field_offset): New functions.
1312 * hsa-gen.c (gen_hsa_addr): Use them.
1313 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1314 (simplify_vector_constructor): Likewise.
1315 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1316 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1317 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1318 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1321 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1322 Alan Hayward <alan.hayward@arm.com>
1323 David Sherwood <david.sherwood@arm.com>
1325 * expr.h (emit_group_load, emit_group_load_into_temps)
1326 (emit_group_store): Take the size as a poly_int64 rather than an int.
1327 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1328 (emit_group_load_into_temp, emit_group_store): Likewise.
1330 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1335 an unsigned int to a poly_uint64.
1336 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1337 sizes as poly_uint64s rather than unsigned ints.
1338 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1341 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1342 Alan Hayward <alan.hayward@arm.com>
1343 David Sherwood <david.sherwood@arm.com>
1345 * lra-constraints.c (emit_inc): Change inc_amount from an int
1348 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1349 Alan Hayward <alan.hayward@arm.com>
1350 David Sherwood <david.sherwood@arm.com>
1352 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1354 (add_stack_var, stack_var_cmp, partition_stack_vars)
1355 (dump_stack_var_partition): Update accordingly.
1356 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1357 than a HOST_WIDE_INT.
1358 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1359 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1360 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1361 rather than a HOST_WIDE_INT.
1363 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * target.def (return_pops_args): Treat both the input and output
1368 sizes as poly_int64s rather than HOST_WIDE_INTS.
1369 * targhooks.h (default_return_pops_args): Update accordingly.
1370 * targhooks.c (default_return_pops_args): Likewise.
1371 * doc/tm.texi: Regenerate.
1372 * emit-rtl.h (incoming_args): Change pops_args, size and
1373 outgoing_args_size from int to poly_int64_pod.
1374 * function.h (expr_status): Change x_pending_stack_adjust and
1375 x_stack_pointer_delta from int to poly_int64.
1376 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1377 (ARGS_SIZE_RTX): Update accordingly.
1378 * calls.c (highest_outgoing_arg_in_use): Change from int to
1380 (stack_usage_watermark, stored_args_watermark): New variables.
1381 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1382 (emit_call_1): Change the stack_size and rounded_stack_size
1383 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1385 (save_fixed_argument_area): Check stack_usage_watermark.
1386 (initialize_argument_information): Change old_pending_adj from
1387 a HOST_WIDE_INT * to a poly_int64_pod *.
1388 (compute_argument_block_size): Return the size as a poly_int64
1390 (finalize_must_preallocate): Track polynomial argument sizes.
1391 (compute_argument_addresses): Likewise.
1392 (internal_arg_pointer_based_exp): Track polynomial offsets.
1393 (mem_overlaps_already_clobbered_arg_p): Rename to...
1394 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1395 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1396 Check stored_args_used_watermark.
1397 (load_register_parameters): Update accordingly.
1398 (check_sibcall_argument_overlap_1): Likewise.
1399 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1400 args size as a poly_int64 rather than an int. Return a bool
1401 indicating whether the optimization was possible and return
1402 the new adjustment by reference.
1403 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1404 Update stored_args_watermark.
1405 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1406 (expand_call): Likewise. Maintain stack_usage_watermark and
1407 stored_args_watermark. Update calls to
1408 combine_pending_stack_adjustment_and_call.
1409 (emit_library_call_value_1): Handle polynomial argument sizes.
1410 Call stack_region_maybe_used_p and mark_stack_region_used.
1411 Maintain stack_usage_watermark.
1412 (store_one_arg): Likewise. Update call to
1413 mem_overlaps_already_clobbered_arg_p.
1414 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1416 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1417 * config/microblaze/microblaze.c (microblaze_function_prologue):
1419 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1420 TARGET_RETURN_POPS_ARGS interface.
1421 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1423 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1424 * config/i386/i386.c (ix86_return_pops_args): Update for new
1425 TARGET_RETURN_POPS_ARGS interface.
1426 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1427 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1428 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1429 TARGET_RETURN_POPS_ARGS interface.
1430 * config/vax/vax.c (vax_return_pops_args): Likewise.
1431 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1432 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1433 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1434 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1435 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1436 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1437 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1438 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1439 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1440 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1441 and x_stack_pointer_delta from int to poly_int64.
1442 * dojump.c (do_pending_stack_adjust): Update accordingly.
1443 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1444 stack_pointer_deltas.
1445 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1446 (pad_to_arg_alignment): Track polynomial offsets.
1447 (assign_parm_find_stack_rtl): Likewise.
1448 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1449 * toplev.c (output_stack_usage): Update reference to
1450 current_function_pushed_stack_size.
1452 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1453 Alan Hayward <alan.hayward@arm.com>
1454 David Sherwood <david.sherwood@arm.com>
1456 * function.c (in_arg_offset, var_offset, dynamic_offset)
1457 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1458 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1459 rather than a HOST_WIDE_INT.
1460 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1461 (instantiate_virtual_regs_in_insn): Likewise.
1463 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1464 Alan Hayward <alan.hayward@arm.com>
1465 David Sherwood <david.sherwood@arm.com>
1467 * rtl.h (get_args_size, add_args_size_note): New functions.
1468 (find_args_size_adjust): Return a poly_int64 rather than a
1470 (fixup_args_size_notes): Likewise. Make the same change to the
1471 end_args_size parameter.
1472 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1473 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1474 * calls.c (emit_call_1): Likewise.
1475 * explow.c (adjust_stack_1): Likewise.
1476 * cfgcleanup.c (old_insns_match_p): Update use of
1477 find_args_size_adjust.
1478 * combine.c (distribute_notes): Track polynomial arg sizes.
1479 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1480 end_true_args_size, beg_delay_args_size and end_delay_args_size
1481 from HOST_WIDE_INT to poly_int64.
1482 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1483 than a HOST_WIDE_INT.
1484 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1485 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1486 polynomial arg sizes.
1487 * emit-rtl.c (try_split): Use get_args_size.
1488 * recog.c (peep2_attempt): Likewise.
1489 * reload1.c (reload_as_needed): Likewise.
1490 * expr.c (find_args_size_adjust): Return the adjustment as a
1491 poly_int64 rather than a HOST_WIDE_INT.
1492 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1493 to a poly_int64 and change the return type in the same way.
1494 (emit_single_push_insn): Track polynomial arg sizes.
1496 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1497 Alan Hayward <alan.hayward@arm.com>
1498 David Sherwood <david.sherwood@arm.com>
1500 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1501 from HOST_WIDE_INT to poly_int64.
1502 * expr.c (push_block, emit_push_insn): Likewise.
1504 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1505 Alan Hayward <alan.hayward@arm.com>
1506 David Sherwood <david.sherwood@arm.com>
1508 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1510 (get_frame_size): Return the size as a poly_int64 rather than a
1512 (frame_offset_overflow): Take the offset as a poly_int64 rather
1513 than a HOST_WIDE_INT.
1514 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1515 (assign_stack_temp): Likewise for the size.
1516 * function.c (get_frame_size): Return a poly_int64 rather than
1518 (frame_offset_overflow): Take the offset as a poly_int64 rather
1519 than a HOST_WIDE_INT.
1520 (try_fit_stack_local): Take the start, length and size as poly_int64s
1521 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1522 rather than a HOST_WIDE_INT.
1523 (add_frame_space): Take the start and end as poly_int64s rather than
1525 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1526 (assign_stack_temp): Likewise for the size.
1527 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1529 (find_temp_slot_from_address): Handle polynomial offsets.
1530 (combine_temp_slots): Likewise.
1531 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1533 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1534 poly_int64 rather than a HOST_WIDE_INT.
1535 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1536 than a HOST_WIDE_INT.
1537 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1538 polynomial frame offsets.
1539 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1540 as a poly_int64 rather than an int.
1541 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1542 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1543 * config/v850/v850.c (compute_frame_size): Likewise.
1544 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1545 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1546 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1547 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1548 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1549 rather than a HOST_WIDE_INT.
1550 * explow.c (get_dynamic_stack_base): Likewise.
1551 * final.c (final_start_function): Use the constant lower bound
1552 of the frame size for -Wframe-larger-than.
1553 * ira.c (do_reload): Adjust for new get_frame_size return type.
1554 * lra.c (lra): Likewise.
1555 * reload1.c (reload): Likewise.
1556 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1557 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1558 * rtlanal.c (get_initial_register_offset): Return the offset as
1559 a poly_int64 rather than a HOST_WIDE_INT.
1561 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1562 Alan Hayward <alan.hayward@arm.com>
1563 David Sherwood <david.sherwood@arm.com>
1565 * reload1.c (elim_table): Change initial_offset, offset and
1566 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1567 (offsets_at): Change the target array's element type from
1568 HOST_WIDE_INT to poly_int64_pod.
1569 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1570 (elimination_costs_in_insn, update_eliminable_offsets)
1571 (verify_initial_elim_offsets, set_offsets_for_label)
1572 (init_eliminable_invariants): Update after above changes.
1574 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1575 Alan Hayward <alan.hayward@arm.com>
1576 David Sherwood <david.sherwood@arm.com>
1578 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1579 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1580 (decomposition): Change start and end from HOST_WIDE_INT
1582 (decompose, immune_p): Update accordingly.
1583 (find_inc_amount): Return a poly_int64 rather than an int.
1584 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1587 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1588 Alan Hayward <alan.hayward@arm.com>
1589 David Sherwood <david.sherwood@arm.com>
1591 * tree.h (get_inner_reference): Return the bitsize and bitpos
1592 as poly_int64_pods rather than HOST_WIDE_INT.
1593 * fold-const.h (ptr_difference_const): Return the pointer difference
1594 as a poly_int64_pod rather than a HOST_WIDE_INT.
1595 * expr.c (get_inner_reference): Return the bitsize and bitpos
1596 as poly_int64_pods rather than HOST_WIDE_INT.
1597 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1599 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1600 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1601 (optimize_bit_field_compare): Update call to get_inner_reference.
1602 (decode_field_reference): Likewise.
1603 (fold_unary_loc): Track polynomial offsets and sizes.
1604 (split_address_to_core_and_offset): Return the bitpos as a
1605 poly_int64_pod rather than a HOST_WIDE_INT.
1606 (ptr_difference_const): Likewise for the pointer difference.
1607 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1608 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1609 * dbxout.c (dbxout_expand_expr): Likewise.
1610 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1611 (loc_list_from_tree_1, fortran_common): Likewise.
1612 * gimple-laddress.c (pass_laddress::execute): Likewise.
1613 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1614 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1615 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1616 * tree-affine.c (tree_to_aff_combination): Likewise.
1617 (get_inner_reference_aff): Likewise.
1618 * tree-data-ref.c (split_constant_offset_1): Likewise.
1619 (dr_analyze_innermost): Likewise.
1620 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1621 * tree-sra.c (ipa_sra_check_caller): Likewise.
1622 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1623 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1624 (instrument_bool_enum_load, instrument_object_size): Likewise.
1625 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1626 to get_inner_reference.
1627 * hsa-gen.c (gen_hsa_addr): Likewise.
1628 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1629 * tsan.c (instrument_expr): Likewise.
1630 * match.pd: Update call to ptr_difference_const.
1632 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayward <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * fold-const.c (fold_comparison): Track sizes and offsets as
1637 poly_int64s rather than HOST_WIDE_INTs when folding address
1640 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1641 Alan Hayward <alan.hayward@arm.com>
1642 David Sherwood <david.sherwood@arm.com>
1644 * expr.h (get_bit_range): Return the bitstart and bitend as
1645 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1646 as a poly_int64 rather than a HOST_WIDE_INT.
1647 * expr.c (get_bit_range): Likewise.
1648 (expand_assignment): Update call accordingly.
1649 * fold-const.c (optimize_bit_field_compare): Likewise.
1651 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1652 Alan Hayward <alan.hayward@arm.com>
1653 David Sherwood <david.sherwood@arm.com>
1655 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1656 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1657 as an optional final parameter.
1658 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1659 (aarch64_legitimate_address_p): Likewise.
1660 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1661 instead of an rtx code.
1662 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1663 (aarch64_legitimate_address_hook_p): Likewise.
1664 (aarch64_print_ldpstp_address): Likewise.
1665 (aarch64_print_operand_address): Likewise.
1666 (aarch64_address_cost): Likewise.
1667 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1668 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1669 (aarch64_mem_pair_lanes_operand): Likewise.
1671 2017-12-20 Richard Biener <rguenther@suse.de>
1673 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1674 update_stmt_if_modified.
1676 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1678 PR tree-optimization/83491
1679 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1680 before walking uses. Improve coding style and comments.
1682 2017-12-20 Tom de Vries <tom@codesourcery.com>
1684 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1686 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1689 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1691 (pass_ipa_free_fn_summary): Add small_p private data member,
1692 initialize to false in the ctor.
1693 (pass_ipa_free_fn_summary::clone,
1694 pass_ipa_free_fn_summary::set_pass_param,
1695 pass_ipa_free_fn_summary::gate): New methods.
1696 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1697 | TODO_dump_symtab if small_p.
1698 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1699 entry and add another instance of the pass with false parm after
1701 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1702 ipa_free_fn_summary here.
1704 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1706 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1708 2017-12-20 Martin Sebor <msebor@redhat.com>
1711 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1712 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1713 (expand_builtin_strncmp): Same.
1715 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1718 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1719 allowed debug stmts before labels.
1720 (expand_gimple_basic_block): Likewise.
1721 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1722 * gimple-iterator.h (gsi_after_labels): Likewise.
1723 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1724 rename reused variable, and simplify using gsi_move_before.
1725 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1726 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1727 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1728 (gimple_verify_flow_info, gimple_block_label): Likewise.
1729 (make_blocks): Move debug markers after adjacent labels.
1730 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1731 allowed debug insns outside blocks.
1732 * df-scan.c (df_insn_delete): Likewise.
1733 * lra-constraints.c (update_ebb_live_info): Likewise.
1734 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1735 (vt_initialize, delete_vta_debug_insns): Likewise.
1736 (reemit_marker_as_note): Drop BB parm. Adjust callers.
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 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1743 (round_up_to_byte_boundary): Likewise.
1744 * expr.h (get_bit_range): Add temporary shim.
1745 * gimple-ssa-store-merging.c (store_operand_info): Change the
1746 bitsize, bitpos, bitregion_start and bitregion_end fields from
1747 unsigned HOST_WIDE_INT to poly_uint64.
1748 (merged_store_group): Likewise load_align_base.
1749 (compatible_load_p, compatible_load_p): Update accordingly.
1750 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1751 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1752 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1753 bitregion_start and bitregion_end as poly_uint64s rather than
1754 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1755 (handled_load): Take the bitsize, bitpos,
1756 bitregion_start and bitregion_end as poly_uint64s rather than
1757 unsigned HOST_WIDE_INTs.
1758 (pass_store_merging::process_store): Update call to
1759 mem_valid_for_store_merging.
1761 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * builtins.c (get_object_alignment_2): Track polynomial offsets
1766 and sizes. Update the alignment handling.
1768 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1769 Alan Hayward <alan.hayward@arm.com>
1770 David Sherwood <david.sherwood@arm.com>
1772 * tree.h (get_inner_reference): Add a version that returns the
1773 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1774 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1775 the case in which bitpos is not associated with the first byte.
1777 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1778 Alan Hayward <alan.hayward@arm.com>
1779 David Sherwood <david.sherwood@arm.com>
1781 * tree-affine.h (get_inner_reference_aff): Return the size as a
1783 * tree-affine.c (get_inner_reference_aff): Likewise.
1784 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1785 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1787 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1788 Alan Hayward <alan.hayward@arm.com>
1789 David Sherwood <david.sherwood@arm.com>
1791 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1792 HOST_WIDE_INT rather than a poly_int64.
1794 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1795 Alan Hayward <alan.hayward@arm.com>
1796 David Sherwood <david.sherwood@arm.com>
1798 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1799 HOST_WIDE_INT to poly_int64_pod.
1800 (perform_symbolic_merge): Update accordingly.
1801 (bswap_replace): Likewise.
1803 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1804 Alan Hayward <alan.hayward@arm.com>
1805 David Sherwood <david.sherwood@arm.com>
1807 * tree-affine.h (aff_tree::offset): Change from widest_int
1809 (wide_int_ext_for_comb): Delete.
1810 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1811 constants as poly_widest_int rather than widest_int.
1812 (aff_combination_constant_multiple_p): Return the multiplier
1813 as a poly_widest_int.
1814 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1816 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1817 version static and add an overload for poly_widest_int.
1818 (aff_combination_const, aff_combination_add_cst)
1819 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1820 the constants as poly_widest_int rather than widest_int.
1821 (tree_to_aff_combination): Generalize INTEGER_CST case to
1823 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1824 (aff_combination_add_product, aff_combination_mult): Handle
1826 (aff_combination_constant_multiple_p): Return the multiplier
1827 as a poly_widest_int.
1828 * tree-predcom.c (determine_offset): Return the offset as a
1830 (split_data_refs_to_components, suitable_component_p): Update
1832 (valid_initializer_p): Update call to
1833 aff_combination_constant_multiple_p.
1834 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1835 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1836 as a poly_int64 rather than a HOST_WIDE_INT.
1837 (get_address_cost): Handle polynomial offsets.
1838 (iv_elimination_compare_lt): Likewise.
1839 (rewrite_use_nonlinear_expr): Likewise.
1841 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1846 as a poly_int64_pod rather than a HOST_WIDE_INT.
1847 (get_addr_base_and_unit_offset): Likewise.
1848 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1849 (get_addr_base_and_unit_offset): Likewise.
1850 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1851 to poly_int64 in example.
1852 * fold-const.c (fold_binary_loc): Update call to
1853 get_addr_base_and_unit_offset.
1854 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1855 (maybe_canonicalize_mem_ref_addr): Likewise.
1856 (gimple_fold_stmt_to_constant_1): Likewise.
1857 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1859 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1860 * match.pd: Likewise.
1861 * omp-low.c (lower_omp_target): Likewise.
1862 * tree-sra.c (build_ref_for_offset): Likewise.
1863 (build_debug_ref_for_model): Likewise.
1864 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1865 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1866 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1867 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1868 (constant_pointer_difference): Likewise.
1869 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1870 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1871 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1872 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1873 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1874 (set_ssa_val_to): Likewise.
1875 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1876 (maybe_diag_stxncpy_trunc): Likewise.
1877 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1878 * tree.c (build_simple_mem_ref_loc): Likewise.
1879 (array_at_struct_end_p): Likewise.
1881 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1882 Alan Hayward <alan.hayward@arm.com>
1883 David Sherwood <david.sherwood@arm.com>
1885 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1886 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1887 (get_ref_base_and_extent_hwi): Declare.
1888 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1889 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1890 (get_ref_base_and_extent_hwi): New function.
1891 * cfgexpand.c (expand_debug_expr): Update call to
1892 get_ref_base_and_extent.
1893 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1894 * gimple-fold.c (get_base_constructor): Return the offset as a
1895 poly_int64_pod rather than a HOST_WIDE_INT.
1896 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1897 * ipa-polymorphic-call.c
1898 (ipa_polymorphic_call_context::set_by_invariant)
1899 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1900 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1901 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1902 rather than get_ref_base_and_extent.
1903 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1904 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1905 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1907 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1908 call to get_ref_base_and_extent.
1909 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1910 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1911 (stmt_kills_ref_p): Likewise.
1912 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1913 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1915 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1916 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1917 when calling native_encode_expr.
1918 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1919 call to get_ref_base_and_extent.
1920 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1921 get_ref_base_and_extent.
1922 * var-tracking.c (track_expr_p): Likewise.
1924 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1925 Alan Hayward <alan.hayward@arm.com>
1926 David Sherwood <david.sherwood@arm.com>
1928 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1929 HOST_WIDE_INT to poly_int64_pod.
1930 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1931 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
1938 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1939 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1940 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1941 from HOST_WIDE_INT to poly_int64_pod.
1942 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1943 (copy_cfa): New function.
1944 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1945 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1946 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1947 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1948 (initial_return_save): Treat offsets as poly_ints.
1949 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1950 offset is nonconstant.
1951 (reg_save): Take the offset as a poly_int64. Fall back to
1952 DW_CFA_expression for nonconstant offsets.
1953 (queue_reg_save): Take the offset as a poly_int64.
1954 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1956 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1957 Alan Hayward <alan.hayward@arm.com>
1958 David Sherwood <david.sherwood@arm.com>
1960 * rtl.h (operand_subword, operand_subword_force): Take the offset
1961 as a poly_uint64 an unsigned int.
1962 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1964 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1965 Alan Hayward <alan.hayward@arm.com>
1966 David Sherwood <david.sherwood@arm.com>
1968 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1969 'p' format code. Use INT_LIST rather than SUBREG as the example of
1970 a code with an XINT and an XEXP. Remove the implication that
1971 accessing an rtx field using XINT is expected to work.
1972 * rtl.def (SUBREG): Change format from "ei" to "ep".
1973 * rtl.h (rtunion::rt_subreg): New field.
1974 (XCSUBREG): New macro.
1975 (SUBREG_BYTE): Use it.
1976 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1977 Update constructor accordingly.
1978 (subreg_shape::operator ==): Update accordingly.
1979 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1980 than an unsigned int.
1981 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1982 a poly_uint64 rather than an unsigned int.
1983 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1984 than an unsigned int.
1985 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1986 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1987 an unsigned int. Take the sizes as poly_uint64s.
1988 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1989 an unsigned int. Take the shift as a poly_uint64 rather than
1991 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1992 as a poly_uint64 rather than an unsigned int.
1993 (simplify_subreg_regno): Likewise.
1994 (byte_lowpart_offset): Return the memory offset as a poly_int64
1996 (subreg_memory_offset): Likewise. Take the subreg offset as a
1997 poly_uint64 rather than an unsigned int.
1998 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1999 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2000 poly_uint64 rather than an unsigned int.
2001 * rtl.c (rtx_format): Describe 'p' in comment.
2002 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2003 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2004 offset as a poly_uint64 rather than an unsigned int.
2005 (byte_lowpart_offset): Return the memory offset as a poly_int64
2007 (subreg_memory_offset): Likewise. Take the subreg offset as a
2008 poly_uint64 rather than an unsigned int.
2009 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2010 mode sizes as poly_uint64s rather than unsigned ints. Return a
2011 poly_uint64 rather than an unsigned int.
2012 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2013 (copy_insn_1): Handle 'p'.
2014 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2015 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2016 an unsigned int. Return the shift in the same way.
2017 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2019 (subreg_size_offset_from_lsb): Take the sizes and shift as
2020 poly_uint64s rather than unsigned ints. Return the offset as
2022 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2023 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2025 * rtlhash.c (add_rtx): Handle 'p'.
2026 * genemit.c (gen_exp): Likewise.
2027 * gengenrtl.c (type_from_format, gendef): Likewise.
2028 * gensupport.c (subst_pattern_match, get_alternatives_number)
2029 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2030 (subst_dup): Likewise.
2031 * gengtype.c (adjust_field_rtx_def): Likewise.
2032 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2033 (match_pattern_2): Likewise.
2034 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2035 (rtx_test::subreg_field): New function.
2036 (operator ==, safe_to_hoist_p, transition_parameter_type)
2037 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2038 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2039 * genpeep.c (match_rtx): Likewise.
2040 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2041 (rtx_writer::print_rtx_operand): Handle 'p'.
2042 (print_value): Handle SUBREG.
2043 * read-rtl.c (apply_int_iterator): Likewise.
2044 (rtx_reader::read_rtx_operand): Handle 'p'.
2045 * alias.c (rtx_equal_for_memref_p): Likewise.
2046 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2047 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2049 * calls.c (expand_call): Likewise.
2050 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2051 (make_extraction, gen_lowpart_for_combine): Likewise.
2052 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2054 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2055 rather than an unsigned int. Treat subreg offsets as poly_ints.
2056 (exp_equiv_p): Handle 'p'.
2057 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2058 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2059 * dse.c (find_shift_sequence): Likewise.
2060 * dwarf2out.c (rtl_for_decl_location): Likewise.
2061 * expmed.c (extract_low_bits): Likewise.
2062 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2063 (expand_expr_real_2): Likewise.
2064 * final.c (alter_subreg): Likewise.
2065 (leaf_renumber_regs_insn): Handle 'p'.
2066 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2067 Treat subreg offsets as poly_ints.
2068 * fwprop.c (forward_propagate_and_simplify): Likewise.
2069 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2070 * ira.c (get_subreg_tracking_sizes): Likewise.
2071 * ira-conflicts.c (go_through_subreg): Likewise.
2072 * ira-lives.c (process_single_reg_class_operands): Likewise.
2073 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2074 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2075 as a poly_uint64 rather than an unsigned int.
2076 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2077 subreg offsets as poly_ints.
2078 * lra-constraints.c (operands_match_p): Handle 'p'.
2079 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2080 * lra-spills.c (assign_mem_slot): Likewise.
2081 * postreload.c (move2add_valid_value_p): Likewise.
2082 * recog.c (general_operand, indirect_operand): Likewise.
2083 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2084 (copyprop_hardreg_forward_1): Likewise.
2085 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2086 (record_subregs_of_mode): Likewise.
2087 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2088 * reload.c (operands_match_p): Handle 'p'.
2089 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2090 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2091 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2092 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2093 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2094 (simplify_const_poly_int_tests<N>::run): Likewise.
2095 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2096 a poly_uint64 rather than an unsigned int.
2097 * valtrack.c (debug_lowpart_subreg): Likewise.
2098 * var-tracking.c (var_lowpart): Likewise.
2099 (loc_cmp): Handle 'p'.
2101 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2102 Alan Hayward <alan.hayward@arm.com>
2103 David Sherwood <david.sherwood@arm.com>
2105 * ira.c (get_subreg_tracking_sizes): New function.
2106 (init_live_subregs): Take an integer size rather than a register.
2107 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2108 to init_live_subregs.
2110 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2111 Alan Hayward <alan.hayward@arm.com>
2112 David Sherwood <david.sherwood@arm.com>
2114 * expr.c (store_constructor_field): Change bitsize from a
2115 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2116 HOST_WIDE_INT to a poly_int64.
2117 (store_constructor): Change size from a HOST_WIDE_INT to
2119 (store_field): Likewise bitsize and bitpos.
2121 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2122 Alan Hayward <alan.hayward@arm.com>
2123 David Sherwood <david.sherwood@arm.com>
2125 * expmed.h (store_bit_field): Change bitregion_start and
2126 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2127 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2128 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2129 (store_fixed_bit_field, store_split_bit_field): Likewise.
2130 * expr.c (store_constructor_field, store_field): Likewise.
2131 (optimize_bitfield_assignment_op): Likewise. Make the same change
2132 to bitsize and bitpos.
2133 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2134 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2135 same change in the constructor arguments.
2136 (get_best_mode): Change bitregion_start and bitregion_end from
2137 unsigned HOST_WIDE_INT to poly_uint64.
2138 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2139 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2141 (bit_field_mode_iterator::next_mode): Update for new types
2142 of m_bitregion_start and m_bitregion_end.
2143 (get_best_mode): Change bitregion_start and bitregion_end from
2144 unsigned HOST_WIDE_INT to poly_uint64.
2146 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2147 Alan Hayward <alan.hayward@arm.com>
2148 David Sherwood <david.sherwood@arm.com>
2150 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2151 accepts poly_uint64 offsets.
2152 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2153 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2154 * expmed.c (lowpart_bit_field_p): Likewise.
2155 (extract_bit_field_as_subreg): New function, split out from...
2156 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2157 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2158 extractions, check that BITSIZE matches the size of the extracted
2159 value and that BITNUM is an exact multiple of that size.
2160 If all else fails, try forcing the value into memory if
2161 BITNUM is variable, and adjusting the address so that the
2162 offset is constant. Split the part that can only handle constant
2163 bitsize and bitnum out into...
2164 (extract_integral_bit_field): ...this new function.
2165 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2166 rather than unsigned HOST_WIDE_INTs.
2168 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2169 Alan Hayward <alan.hayward@arm.com>
2170 David Sherwood <david.sherwood@arm.com>
2172 * expmed.h (store_bit_field): Take bitsize and bitnum as
2173 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2174 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2175 that returns the byte size.
2176 (store_bit_field_1): Take bitsize and bitnum as
2177 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2178 to simple_mem_bitfield_p. Split the part that can only handle
2179 constant bitsize and bitnum out into...
2180 (store_integral_bit_field): ...this new function.
2181 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2182 than unsigned HOST_WIDE_INTs.
2183 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2185 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2186 Alan Hayward <alan.hayward@arm.com>
2187 David Sherwood <david.sherwood@arm.com>
2189 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2190 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2192 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2193 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2194 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2195 offset as a poly_int64 rather than an int.
2196 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2197 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2198 * lra-constraints.c (equiv_address_substitution): Track offsets
2200 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2201 (curr_insn_transform): Handle the new form of sp_offset.
2202 * lra-eliminations.c (lra_elim_table): Change previous_offset
2203 and offset from HOST_WIDE_INT to poly_int64.
2204 (print_elim_table, update_reg_eliminate): Update accordingly.
2205 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2206 (get_elimination): Update accordingly.
2207 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2208 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2209 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2210 poly_int64 offsets generally.
2211 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2212 (mark_not_eliminable, init_elimination): Update accordingly.
2213 (remove_reg_equal_offset_note): Return a bool and pass the new
2214 offset back by pointer as a poly_int64.
2215 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2216 rather than a HOST_WIDE_INT.
2217 (do_remat): Track offsets poly_int64s.
2218 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2220 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2221 Alan Hayward <alan.hayward@arm.com>
2222 David Sherwood <david.sherwood@arm.com>
2224 * rtl.h (mem_attrs): Add a default constructor. Change size and
2225 offset from HOST_WIDE_INT to poly_int64.
2226 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2227 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2228 (widen_memory_access): Take the sizes and offsets as poly_int64s
2229 rather than HOST_WIDE_INTs.
2230 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2231 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2233 (adjust_offset_for_component_ref): Change the offset from a
2234 HOST_WIDE_INT to a poly_int64.
2235 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2236 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2237 * dce.c (find_call_stack_args): Likewise.
2238 * dse.c (record_store): Likewise.
2239 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2240 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2241 * read-rtl-function.c (test_loading_mem): Likewise.
2242 * rtlanal.c (may_trap_p_1): Likewise.
2243 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2244 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2245 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2246 (mem_attrs::mem_attrs): New function.
2247 (set_mem_attributes_minus_bitpos): Change bitpos from a
2248 HOST_WIDE_INT to poly_int64.
2249 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2250 (clear_mem_offset, clear_mem_size, change_address)
2251 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2252 initialize mem_attrs.
2253 (set_mem_offset, set_mem_size, adjust_address_1)
2254 (adjust_automodify_address_1, offset_address, widen_memory_access):
2255 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2257 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2258 Alan Hayward <alan.hayward@arm.com>
2259 David Sherwood <david.sherwood@arm.com>
2261 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2262 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2263 rather than 0 to represent an unknown size. Assert that the size
2264 is known when the mode isn't BLKmode.
2265 (may_trap_p_1): Use -1 for unknown sizes.
2266 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2268 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2269 Alan Hayward <alan.hayward@arm.com>
2270 David Sherwood <david.sherwood@arm.com>
2272 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2273 to poly_int64. Update commentary for positions_needed.large.
2274 (read_info_type): Change offset and width from HOST_WIDE_INT
2276 (set_usage_bits): Likewise.
2277 (canon_address): Return the offset as a poly_int64 rather than
2278 a HOST_WIDE_INT. Use strip_offset_and_add.
2279 (set_all_positions_unneeded, any_positions_needed_p): Use
2280 positions_needed.large to track stores with non-constant widths.
2281 (all_positions_needed_p): Likewise. Take the offset and width
2282 as poly_int64s rather than ints. Assert that rhs is nonnull.
2283 (record_store): Cope with non-constant offsets and widths.
2284 Nullify the rhs of an earlier store if we can't tell which bytes
2286 (find_shift_sequence): Take the access_size and shift as poly_int64s
2288 (get_stored_val): Take the read_offset and read_width as poly_int64s
2289 rather than HOST_WIDE_INTs.
2290 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2291 non-constant offsets and widths.
2293 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * inchash.h (inchash::hash::add_poly_int): New function.
2298 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2299 Use poly_int64 rather than HOST_WIDE_INT.
2300 (ao_ref::max_size_known_p): New function.
2301 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2302 rather than HOST_WIDE_INT.
2303 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2304 to temporaries until its interface is adjusted to match.
2305 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2306 (aliasing_component_refs_p, decl_refs_may_alias_p)
2307 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2308 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2309 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2311 * alias.c (ao_ref_from_mem): Likewise.
2312 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2313 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2314 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2315 (maybe_trim_complex_store, maybe_trim_constructor_store)
2316 (live_bytes_read, dse_classify_store): Likewise.
2317 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2318 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2319 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2320 (vn_reference_lookup_3): Likewise.
2321 * tree-ssa-uninit.c (warn_uninitialized_vars): 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 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2328 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2329 instead of ranges_overlap_p.
2331 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2332 Alan Hayward <alan.hayward@arm.com>
2333 David Sherwood <david.sherwood@arm.com>
2335 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2336 sizes as poly_int64s rather than HOST_WIDE_INTs.
2338 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2339 Alan Hayward <alan.hayward@arm.com>
2340 David Sherwood <david.sherwood@arm.com>
2342 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2343 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2344 * gimple-fold.c (fold_ctor_reference): Likewise.
2346 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2347 Alan Hayward <alan.hayward@arm.com>
2348 David Sherwood <david.sherwood@arm.com>
2350 * target.def (dwarf_poly_indeterminate_value): New hook.
2351 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2352 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2353 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2354 * doc/tm.texi: Regenerate.
2355 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2356 offset as a poly_int64.
2357 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2358 offset as a poly_int64.
2359 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2360 Take the offset as a poly_int64.
2361 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2362 (frame_pointer_fb_offset): Change to a poly_int64.
2363 (int_loc_descriptor): Take the offset as a poly_int64. Use
2364 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2365 (based_loc_descr): Take the offset as a poly_int64.
2366 Use strip_offset_and_add to handle (plus X (const)).
2367 Use new_reg_loc_descr instead of an open-coded version of the
2368 previous implementation.
2369 (mem_loc_descriptor): Handle CONST_POLY_INT.
2370 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2371 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2373 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2374 Alan Hayward <alan.hayward@arm.com>
2375 David Sherwood <david.sherwood@arm.com>
2377 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2378 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2379 * inchash.h (inchash::hash::add_poly_hwi): New function.
2380 * gengtype.c (main): Register poly_int64.
2381 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2382 offset as a poly_int.
2383 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2384 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2385 offset as a poly_int64.
2386 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2387 * print-rtl.c (print_poly_int): New function.
2388 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2390 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2392 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2393 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2394 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2395 Enforce track_offset_p for parts of a PARALLEL.
2396 (vt_add_function_parameter): Use const_offset for the final
2397 offset to track. Use get_tracked_reg_offset for the parts
2400 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2401 Alan Hayward <alan.hayward@arm.com>
2402 David Sherwood <david.sherwood@arm.com>
2404 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2405 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2406 * doc/tm.texi: Regenerate.
2407 * hooks.h (hook_bool_uint_uint_true): Delete.
2408 (hook_bool_puint64_puint64_true): Declare.
2409 * hooks.c (hook_bool_uint_uint_true): Delete.
2410 (hook_bool_puint64_puint64_true): New function.
2411 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2412 instead of unsigned ints.
2413 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2414 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2416 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2417 Alan Hayward <alan.hayward@arm.com>
2418 David Sherwood <david.sherwood@arm.com>
2420 * optabs.h (expand_operand): Add an int_value field.
2421 (create_expand_operand): Add an int_value parameter and use it
2422 to initialize the new expand_operand field.
2423 (create_integer_operand): Replace with a declaration of a function
2424 that accepts poly_int64s. Move the implementation to...
2425 * optabs.c (create_integer_operand): ...here.
2426 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2427 the mode preserves the value of int_value, instead of calling
2428 const_int_operand on the rtx. Use gen_int_mode to generate
2431 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2432 Alan Hayward <alan.hayward@arm.com>
2433 David Sherwood <david.sherwood@arm.com>
2435 * dumpfile.h (dump_dec): Declare.
2436 * dumpfile.c (dump_dec): New function.
2437 * pretty-print.h (pp_wide_integer): Turn into a function and
2438 declare a poly_int version.
2439 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2441 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2442 Alan Hayward <alan.hayward@arm.com>
2443 David Sherwood <david.sherwood@arm.com>
2445 * doc/generic.texi (POLY_INT_CST): Document.
2446 * tree.def (POLY_INT_CST): New tree code.
2447 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2448 * tree-core.h (tree_poly_int_cst): New struct.
2449 (tree_node): Add a poly_int_cst field.
2450 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2451 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2452 instead of a wide_int_ref.
2453 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2455 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2456 instead of an unsigned HOST_WIDE_INT.
2457 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2458 (ptrdiff_tree_p): Declare.
2459 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2460 extern inline implementations if the target doesn't use POLY_INT_CST.
2461 (poly_int_tree_p): New function.
2462 (wi::unextended_tree): New class.
2463 (wi::int_traits <unextended_tree>): New override.
2464 (wi::extended_tree): Add a default constructor.
2465 (wi::extended_tree::get_tree): New function.
2466 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2467 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2468 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2469 (wi::tree_to_poly_wide_ref): New typedefs.
2470 (wi::ints_for): Provide overloads for extended_tree and
2472 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2473 (wi::to_wide): New functions.
2474 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2475 * tree.c (poly_int_cst_hasher): New struct.
2476 (poly_int_cst_hash_table): New variable.
2477 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2478 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2480 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2481 (init_ttree): Initialize poly_int_cst_hash_table.
2482 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2483 a poly_int64 instead of a HOST_WIDE_INT.
2484 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2485 instead of an unsigned HOST_WIDE_INT.
2486 (wide_int_to_tree): Rename to...
2487 (wide_int_to_tree_1): ...this.
2488 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2489 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2490 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2491 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2492 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2493 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2495 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2496 (streamer_read_tree_body): Likewise.
2497 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2498 (streamer_write_tree_body): Likewise.
2499 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2500 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2501 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2502 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2503 * gimple-expr.h (is_gimple_constant): Likewise.
2504 * gimplify.c (maybe_with_size_expr): Likewise.
2505 * print-tree.c (print_node): Likewise.
2506 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2507 * tree-pretty-print.c (dump_generic_node): Likewise.
2508 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2509 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2510 * tree-vrp.c (compare_values_warnv): Likewise.
2511 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2512 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2513 (force_expr_to_var_cost): Likewise.
2514 * tree-ssa-loop.c (for_each_index): Likewise.
2515 * fold-const.h (build_invariant_address, size_int_kind): Take a
2516 poly_int64 instead of a HOST_WIDE_INT.
2517 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2518 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2520 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2521 (int_const_binop_2): New function, split out from...
2522 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2523 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2524 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2525 * gimple-ssa-strength-reduction.c (slsr_process_add)
2526 (slsr_process_mul): Check for INTEGER_CSTs before using them
2528 * stor-layout.c (bits_from_bytes): New function.
2529 (bit_from_pos): Use it.
2530 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2531 by BITS_PER_UNIT to get the TYPE_SIZE.
2532 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2533 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2535 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * doc/rtl.texi (const_poly_int): Document. Also document the
2540 rtl sharing behavior.
2541 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2542 * rtl.h (const_poly_int_def): New struct.
2543 (rtx_def::u): Add a cpi field.
2544 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2545 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2546 (wi::rtx_to_poly_wide_ref): New typedef
2547 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2548 (poly_int_rtx_p): New functions.
2549 (trunc_int_for_mode): Declare a poly_int64 version.
2550 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2551 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2553 (strip_offset): Declare.
2554 (strip_offset_and_add): New function.
2555 * rtl.def (CONST_POLY_INT): New rtx code.
2556 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2557 (shared_const_p): Use poly_int_rtx_p.
2558 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2560 (gen_int_shift_amount): Likewise.
2561 * emit-rtl.c (const_poly_int_hasher): New class.
2562 (const_poly_int_htab): New variable.
2563 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2564 (const_poly_int_hasher::hash): New function.
2565 (const_poly_int_hasher::equal): Likewise.
2566 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2567 (immed_wide_int_const): Rename to...
2568 (immed_wide_int_const_1): ...this and make static.
2569 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2570 instead of a wide_int_ref.
2571 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2572 (gen_lowpart_common): Handle CONST_POLY_INT.
2573 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2574 * cselib.c (cselib_hash_rtx): Likewise.
2575 * dwarf2out.c (const_ok_for_output_1): Likewise.
2576 * expr.c (convert_modes): Likewise.
2577 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2578 * rtlhash.c (add_rtx): Likewise.
2579 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2580 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2581 Handle existing CONST_POLY_INT rtxes.
2582 * expmed.h (expand_shift): Take a poly_int64 instead of a
2584 * expmed.c (expand_shift): Likewise.
2585 * rtlanal.c (strip_offset): New function.
2586 (commutative_operand_precedence): Give CONST_POLY_INT the same
2587 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2589 * rtl-tests.c (const_poly_int_tests): New struct.
2590 (rtl_tests_c_tests): Use it.
2591 * simplify-rtx.c (simplify_const_unary_operation): Handle
2593 (simplify_const_binary_operation): Likewise.
2594 (simplify_binary_operation_1): Fold additions of symbolic constants
2595 and CONST_POLY_INTs.
2596 (simplify_subreg): Handle extensions and truncations of
2598 (simplify_const_poly_int_tests): New struct.
2599 (simplify_rtx_c_tests): Use it.
2600 * wide-int.h (storage_ref): Add default constructor.
2601 (wide_int_ref_storage): Likewise.
2602 (trailing_wide_ints): Use GTY((user)).
2603 (trailing_wide_ints::operator[]): Add a const version.
2604 (trailing_wide_ints::get_precision): New function.
2605 (trailing_wide_ints::extra_size): Likewise.
2607 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2608 Alan Hayward <alan.hayward@arm.com>
2609 David Sherwood <david.sherwood@arm.com>
2611 * emit-rtl.h (gen_int_shift_amount): Declare.
2612 * emit-rtl.c (gen_int_shift_amount): New function.
2613 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2615 * calls.c (shift_return_value): Likewise.
2616 * cse.c (fold_rtx): Likewise.
2617 * dse.c (find_shift_sequence): Likewise.
2618 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2619 (expand_shift, expand_smod_pow2): Likewise.
2620 * lower-subreg.c (shift_cost): Likewise.
2621 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2622 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2623 (expand_vec_perm_var): Likewise.
2624 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2625 (simplify_binary_operation_1): Likewise.
2626 * combine.c (try_combine, find_split_point, force_int_to_mode)
2627 (simplify_shift_const_1, simplify_shift_const): Likewise.
2628 (change_zero_ext): Likewise. Use simplify_gen_binary.
2630 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2632 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2634 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2636 * doc/invoke.texi (ARM Options): Document accepted extension options
2637 for -march=armv8.3-a.
2639 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2642 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2643 or --with-float=softfp, set the default CPU to arm10e.
2645 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2647 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2648 * config/visium/predicates.md (const_shift_operand): Likewise.
2649 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2650 (visium_legitimize_reload_address): Likewise.
2652 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2655 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2657 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2660 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2662 2017-12-20 Julia Koval <julia.koval@intel.com>
2664 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2665 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2666 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2667 * config.gcc: Include vpclmulqdqintrin.h.
2668 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2669 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2671 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2672 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2673 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2674 * config/i386/i386.c (isa2_opts): Add -mcx16.
2675 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2676 (ix86_option_override_internal): Move mcx16 to flags2.
2677 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2678 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2679 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2680 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2681 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2682 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2683 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2684 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2685 * doc/invoke.texi: Add -mvpclmulqdq.
2687 2017-12-20 Tom de Vries <tom@codesourcery.com>
2690 * config/i386/i386.c (ix86_static_chain): Move
2691 DECL_STATIC_CHAIN test ...
2692 * calls.c (rtx_for_static_chain): ... here. New function.
2693 * calls.h (rtx_for_static_chain): Declare.
2694 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2695 instead of targetm.calls.static_chain.
2696 * df-scan.c (df_get_entry_block_def_set): Same.
2698 2017-12-19 Tom de Vries <tom@codesourcery.com>
2700 PR tree-optimization/83493
2701 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2704 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2706 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2707 inform with hardcoded english plural handling.
2709 2017-12-18 Jeff Law <law@redhat.com>
2711 PR tree-optimization/83477
2712 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2713 a non-virtual PHI, always push a new range.
2715 2017-12-19 Martin Sebor <msebor@redhat.com>
2718 * builtins.c (compute_objsize): Handle non-constant offsets.
2720 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2722 PR tree-optimization/83444
2723 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2724 character load case, if get_stridx on MEM_REF's operand doesn't
2725 look usable, retry with get_addr_stridx.
2727 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2730 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2731 (variable_tracking_main_1): Keep markers even when VTA fails.
2734 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2735 even if there are markers before it.
2736 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2738 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2740 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2741 typical order conditions.
2742 * sel-sched.c (extract_new_fences_from): Likewise.
2743 * config/visium/constraints.md (J, K, L): Likewise.
2744 * config/visium/predicates.md (const_shift_operand): Likewise.
2745 * config/visium/visium.c (visium_legitimize_address,
2746 visium_legitimize_reload_address): Likewise.
2747 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2748 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2749 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2750 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2751 SET_DUMP_DETAIL): Likewise.
2752 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2753 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2754 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2755 avr_set_core_architecture, avr_set_current_function,
2756 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2757 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2758 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2759 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2760 avr_map_decompose, avr_fold_builtin): Likewise.
2761 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2762 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2763 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2764 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2765 * config/m32c/m32c.c (m32c_conditional_register_usage,
2766 m32c_address_cost): Likewise.
2767 * config/m32c/predicates.md (shiftcount_operand,
2768 longshiftcount_operand): Likewise.
2769 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2770 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2771 can_use_cdx_ldstw): Likewise.
2772 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2773 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2775 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2776 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2777 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2778 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2779 * config/vax/vax.c (adjacent_operands_p): Likewise.
2780 * config/ft32/constraints.md (L, b, KA): Likewise.
2781 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2783 * cfgexpand.c (expand_stack_alignment): Likewise.
2784 * gcse.c (insert_expr_in_table): Likewise.
2785 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2786 * cgraphunit.c (cgraph_node::expand): Likewise.
2787 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2788 * emit-rtl.c (add_insn): Likewise.
2789 * input.c (dump_location_info): Likewise.
2790 * passes.c (NEXT_PASS): Likewise.
2791 * read-rtl-function.c (parse_note_insn_name,
2792 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2794 * sched-rgn.c (sched_rgn_init): Likewise.
2795 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2796 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2797 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2799 * explow.c (eliminate_constant_term): Likewise.
2800 * final.c (leaf_renumber_regs_insn): Likewise.
2801 * cfgrtl.c (print_rtl_with_bb): Likewise.
2802 * genhooks.c (emit_init_macros): Likewise.
2803 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2804 * tree-data-ref.c (conflict_fn): Likewise.
2805 * selftest.c (assert_streq): Likewise.
2806 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2807 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2808 fold_binary_loc, multiple_of_p): Likewise.
2809 * reload.c (push_reload, find_equiv_reg): Likewise.
2810 * et-forest.c (et_nca, et_below): Likewise.
2811 * dbxout.c (dbxout_symbol_location): Likewise.
2812 * reorg.c (relax_delay_slots): Likewise.
2813 * dojump.c (do_compare_rtx_and_jump): Likewise.
2814 * gengtype-parse.c (type): Likewise.
2815 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2816 simplify_const_relational_operation): Likewise.
2817 * reload1.c (do_output_reload): Likewise.
2818 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2819 * gengtype.c (type_for_name): Likewise.
2820 * gimple-ssa-sprintf.c (format_directive): Likewise.
2822 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2825 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2826 accessing it. Adjust comment.
2828 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2831 * predict.c (zero_one_minusone): New function.
2832 (apply_return_prediction): Avoid return prediction for functions
2833 returning only -1, 0 and 1 values, unless they only return -1 and 0
2836 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2838 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2839 fall-through warning.
2840 (arc_compute_frame_size): Remove unused variables.
2841 (arc_print_operand): Fix fprintif format.
2842 (arc_can_follow_jump): Clean fall-through warning.
2844 2017-12-19 Marek Polacek <polacek@redhat.com>
2847 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2850 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2852 * config/arc/arc.c (overriderregs): New variable.
2853 (arc_override_options): Track fixed/call saved/call options.
2854 (arc_conditional_register_usage): Check against overrideregs
2855 variable whenever we change register properties.
2857 2017-12-19 Nathan Sidwell <nathan@acm.org>
2859 * opts.c (finish_options): Don't prefix dump_base_name if it
2860 already contains directories.
2862 2017-12-19 Martin Liska <mliska@suse.cz>
2864 PR rtl-optimization/82675
2865 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2866 more element in sbitmap.
2868 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2870 * gimplify.c (gimplify_expr): Use error_operand_p.
2872 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2875 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2878 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2880 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2881 the jump is to a label.
2883 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2885 PR tree-optimization/83444
2886 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2891 * ipa-inline.c (flatten_remove_node_hook): New function.
2892 (ipa_inline): Keep only nodes with flatten attribute at the end of
2893 the array in the order from ipa_reverse_postorder, only walk that
2894 portion of array for flattening, if there is more than one such
2895 node, temporarily register a removal hook and ignore removed nodes.
2897 PR tree-optimization/80631
2898 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2899 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2900 IFN_REDUC_MAX or IFN_REDUC_MIN.
2902 2017-12-18 Jeff Law <law@redhat.com>
2904 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2905 record symbolic equivalences from backedges in the CFG.
2908 2017-11-19 Jeff Law <law@redhat.com>
2910 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2911 of degenerates resulting from ignoring an edge.
2913 2017-12-18 Martin Sebor <msebor@redhat.com>
2916 PR tree-optimization/78450
2917 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2918 (handle_builtin_strlen): Call it.
2920 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2922 PR rtl-optimization/83424
2923 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2925 2017-12-18 Marek Polacek <polacek@redhat.com>
2928 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2929 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2932 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2934 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2935 (umaddsidi4, umaddsidi_split): Likewise.
2937 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2939 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2942 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2945 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2947 2017-12-18 Richard Biener <rguenther@suse.de>
2949 PR tree-optimization/81877
2950 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2951 (outermost_indep_loop): Adjust.
2952 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2953 (can_sm_ref_p): Adjust.
2955 2017-12-18 Richard Biener <rguenther@suse.de>
2958 * tree.c (array_at_struct_end_p): Return true if the underlying
2959 object has space for at least one element in excess of what
2960 the array domain specifies.
2962 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2964 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2965 example to avoid overfull hbox.
2966 * doc/invoke.texi (Option Summary): Add missing @gol.
2967 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2970 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2971 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2973 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2974 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2975 (AArch64 Options): Document -mverbose-cost-dump.
2976 (ARM Options): Likewise, plus -mflip-thumb.
2978 2017-12-17 Martin Sebor <msebor@redhat.com>
2981 * gimple-ssa-warn-restrict.c
2982 (builtin_memref::offset_out_of_bounds): Correct the handling of
2985 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2987 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2989 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2991 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2993 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2995 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2998 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3000 * doc/invoke.texi: Fix some typos.
3002 2017-12-16 Martin Sebor <msebor@redhat.com>
3004 PR tree-optimization/78918
3005 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3006 * builtins.c (check_sizes): Rename...
3007 (check_access): ...to this. Rename function arguments for clarity.
3008 (check_memop_sizes): Adjust names.
3009 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3010 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3011 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3012 (check_strncat_sizes, expand_builtin_strncat): Same.
3013 (expand_builtin_strncpy, expand_builtin_memset): Same.
3014 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3015 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3016 (maybe_emit_sprintf_chk_warning): Same.
3017 (expand_builtin_strcpy): Adjust.
3018 (expand_builtin_stpcpy): Same.
3019 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3020 in pointer-checking forms of memcpy, memmove, and mempcpy.
3021 (gcall_to_tree_minimal, max_object_size): Define new functions.
3022 * builtins.h (max_object_size): Declare.
3023 * calls.c (alloc_max_size): Call max_object_size instead of
3024 hardcoding ssizetype limit.
3025 (get_size_range): Handle new argument.
3026 * calls.h (get_size_range): Add a new argument.
3027 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3028 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3029 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3031 (gimple_fold_builtin_memory_chk): Same.
3032 (gimple_fold_builtin_stxcpy_chk): New function.
3033 * gimple-ssa-warn-restrict.c: New source.
3034 * gimple-ssa-warn-restrict.h: New header.
3035 * gimple.c (gimple_build_call_from_tree): Propagate location.
3036 * passes.def (pass_warn_restrict): Add new pass.
3037 * tree-pass.h (make_pass_warn_restrict): Declare.
3038 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3040 (handle_builtin_strcat): Same.
3041 (strlen_optimize_stmt): Rename...
3042 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3043 stpncpy, strncpy, and their checking forms.
3045 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3047 PR rtl-optimization/82849
3048 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3049 and get_max_loop_iterations_int.
3051 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3052 Alan Hayward <alan.hayward@arm.com>
3053 David Sherwood <david.sherwood@arm.com>
3055 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3056 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3057 size as a poly_uint64.
3058 (mode_for_vector, mode_for_int_vector): Take the number of vector
3059 elements as a poly_uint64.
3060 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3061 size as a poly_uint64.
3062 (mode_for_vector, mode_for_int_vector): Take the number of vector
3063 elements as a poly_uint64.
3065 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3066 Alan Hayward <alan.hayward@arm.com>
3067 David Sherwood <david.sherwood@arm.com>
3069 * machmode.h (MACRO_MODE): New macro.
3070 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3071 * targhooks.c (default_libcall_value, default_secondary_reload)
3072 (default_memory_move_cost, default_register_move_cost)
3073 (default_class_max_nregs): Likewise.
3075 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3076 Alan Hayward <alan.hayward@arm.com>
3077 David Sherwood <david.sherwood@arm.com>
3079 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3080 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3082 * genautomata.c (main): Likewise.
3083 * genconditions.c (write_header): Likewise.
3084 * genemit.c (main): Likewise.
3085 * genextract.c (print_header): Likewise.
3086 * genopinit.c (main): Likewise.
3087 * genoutput.c (output_prologue): Likewise.
3088 * genpeep.c (main): Likewise.
3089 * genpreds.c (write_insn_preds_c): Likewise.
3090 * genrecog.c (writer_header): Likewise.
3091 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3092 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3093 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3094 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3095 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3096 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3097 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3098 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3099 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3100 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3101 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3102 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3103 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3104 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3105 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3106 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3107 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3108 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3109 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3110 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3111 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3112 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3113 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3114 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3115 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3116 * config/darwin.c (IN_TARGET_CODE): Likewise.
3117 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3118 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3119 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3120 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3121 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3122 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3123 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3124 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3125 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3126 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3127 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3128 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3129 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3130 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3131 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3132 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3133 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3134 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3135 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3136 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3137 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3138 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3139 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3140 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3141 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3142 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3143 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3144 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3145 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3146 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3147 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3148 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3149 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3150 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3151 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3152 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3153 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3154 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3155 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3156 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3157 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3158 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3159 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3160 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3161 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3162 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3163 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3164 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3165 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3166 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3167 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3168 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3169 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3170 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3171 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3172 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3173 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3174 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3175 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3176 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3177 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3178 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3179 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3180 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3181 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3182 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3183 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3184 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3185 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3186 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3187 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3188 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3189 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3190 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3191 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3192 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3193 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3194 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3195 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3196 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3197 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3198 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3199 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3200 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3201 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3202 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3203 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3204 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3205 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3206 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3207 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3208 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3209 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3210 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3211 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3212 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3213 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3214 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3215 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3216 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3217 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3218 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3219 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3220 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3221 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3222 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3224 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3225 Alan Hayward <alan.hayward@arm.com>
3226 David Sherwood <david.sherwood@arm.com>
3228 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3231 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3232 Alan Hayward <alan.hayward@arm.com>
3233 David Sherwood <david.sherwood@arm.com>
3235 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3236 * doc/md.texi (vec_series@var{m}): Document.
3237 * tree.def (VEC_SERIES_EXPR): New tree code.
3238 * tree.h (build_vec_series): Declare.
3239 * tree.c (build_vec_series): New function.
3240 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3241 * tree-pretty-print.c (dump_generic_node): Likewise.
3242 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3243 * tree-inline.c (estimate_operator_cost): Likewise.
3244 * expr.c (expand_expr_real_2): Likewise.
3245 * optabs-tree.c (optab_for_tree_code): Likewise.
3246 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3247 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3248 * expmed.c (make_tree): Handle VEC_SERIES.
3249 * optabs.def (vec_series_optab): New optab.
3250 * optabs.h (expand_vec_series_expr): Declare.
3251 * optabs.c (expand_vec_series_expr): New function.
3252 * tree-vect-generic.c (expand_vector_operations_1): Check that
3253 the operands also have vector type.
3255 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3256 Alan Hayward <alan.hawyard@arm.com>
3257 David Sherwood <david.sherwood@arm.com>
3259 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3260 (VEC_COND_EXPR): Add missing @tindex.
3261 * doc/md.texi (vec_duplicate@var{m}): Document.
3262 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3263 * tree.c (build_vector_from_val): Add stubbed-out handling of
3264 variable-length vectors, using VEC_DUPLICATE_EXPR.
3265 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3266 * cfgexpand.c (expand_debug_expr): Likewise.
3267 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3268 * tree-inline.c (estimate_operator_cost): Likewise.
3269 * tree-pretty-print.c (dump_generic_node): Likewise.
3270 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3271 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3272 (test_vec_duplicate_folding): New function.
3273 (fold_const_c_tests): Call it.
3274 * optabs.def (vec_duplicate_optab): New optab.
3275 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3276 * optabs.h (expand_vector_broadcast): Declare.
3277 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3278 vec_duplicate_optab.
3279 * expr.c (store_constructor): Try using vec_duplicate_optab for
3281 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3283 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3286 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3287 div/mod latencies a bit.
3289 2017-12-15 Jeff Law <law@redhat.com>
3291 PR tree-optimization/36550
3292 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3293 (mark_threaded_blocks): Rewrite code to avoid block copying when
3294 optimizing for size. Don't pessimize blocks which will be
3295 copied, but all the statements will be dead.
3297 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3299 PR tree-optimization/81165
3300 * tree-ssa-threadupdate.c (uses_in_bb): New.
3301 (estimate_threading_killed_stmts): New.
3302 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3303 * tree-ssa-threadedge.c
3304 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3307 2017-12-15 Jeff Law <law@redhat.com>
3309 PR tree-optimization/83410
3310 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3311 threads when parallelizing loops.
3313 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3315 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3317 * config/alpha/alpha.c (vms_attribute_table): Swap
3318 affects_type_identity and handler fields, adjust comments.
3319 * config/mips/mips.c (mips_attribute_table): Likewise.
3320 * config/visium/visium.c (visium_attribute_table): Likewise.
3321 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3322 * config/microblaze/microblaze.c (microblaze_attribute_table):
3324 * config/spu/spu.c (spu_attribute_table): Likewise.
3325 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3326 * config/arc/arc.c (arc_attribute_table): Likewise.
3327 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3328 * config/v850/v850.c (v850_handle_interrupt_attribute,
3329 v850_handle_data_area_attribute): Formatting fixes.
3330 (v850_attribute_table): Swap affects_type_identity and handler
3331 fields, adjust comments.
3332 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3333 * config/arm/arm.c (arm_attribute_table): Likewise.
3334 * config/avr/avr.c (avr_attribute_table): Likewise.
3335 * config/s390/s390.c (s390_attribute_table): Likewise.
3336 * config/sh/sh.c (sh_attribute_table): Likewise.
3337 * config/i386/i386.c (ix86_handle_cconv_attribute,
3338 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3339 (ix86_attribute_table): Swap affects_type_identity and handler
3340 fields, adjust comments.
3341 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3342 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3343 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3344 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3345 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3346 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3347 * config/rx/rx.c (rx_attribute_table): Likewise.
3348 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3349 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3350 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3351 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3352 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3353 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3354 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3355 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3356 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3357 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3358 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3360 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3362 (register_scoped_attributes, decl_attributes): Formatting fixes.
3364 PR tree-optimization/83269
3365 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3366 subtraction in arg0's type if type is signed and arg0 is unsigned.
3370 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3371 convert? on @0 instead of convert. Check type of @1, not @0.
3372 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3373 convert? on @0 instead of convert. Check type of @1, not @0.
3374 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3375 only optimize if either both @1 and @2 types are narrower
3376 precision, or both are wider or equal precision, and in the former
3377 case only if both have undefined overflow.
3379 2017-12-15 Richard Biener <rguenther@suse.de>
3382 * internal-fn.def (IFN_NOP): Add.
3383 * internal-fn.c (expand_NOP): Do nothing.
3384 * lto-streamer-in.c (input_function): Instead of removing
3385 sanitizer calls replace them with IFN_NOP calls.
3387 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3388 Alan Hayward <alan.hayward@arm.com>
3389 David Sherwood <david.sherwood@arm.com>
3391 * dse.c (store_info, read_info_type): Replace begin and end with
3393 (print_range): New function.
3394 (set_all_positions_unneeded, any_positions_needed_p)
3395 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3397 (record_store): Likewise. Optimize the case in which all positions
3399 (get_stored_val): Replace read_begin and read_end with read_offset
3401 (replace_read): Update call accordingly.
3403 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3405 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3406 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3407 (loop_cand::loop_cand): Initialize above members.
3408 (loop_cand::supported_operations): Delete.
3409 (loop_cand::can_interchange_p): Inline above function.
3410 (loop_cand::classify_simple_reduction): Record number of constant
3411 initialized simple reductions.
3412 (should_interchange_loops): New parameters. Check stmt cost of loops
3414 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3415 Update call to should_interchange_loops.
3416 (should_interchange_loop_nest): Update call to
3417 should_interchange_loops.
3419 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3422 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3423 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3424 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3426 2017-12-15 Julia Koval <julia.koval@intel.com>
3428 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3429 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3431 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3432 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3433 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3435 2017-12-15 Julia Koval <julia.koval@intel.com>
3437 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3438 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3439 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3440 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3441 _mm_aesenc_epi128): New intrinsics.
3443 2017-12-15 Julia Koval <julia.koval@intel.com>
3445 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3446 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3448 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3449 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3450 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3452 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3454 * gimple-ssa-strength-reduction.c (analyze_increments):
3455 Distinguish replacement costs for constant strides from those for
3458 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3460 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3462 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3464 * doc/invoke.texi: Document -Wcast-function-type.
3465 * recog.h (stored_funcptr): Change signature.
3466 * tree-dump.c (dump_node): Avoid warning.
3467 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3469 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3472 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3475 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3477 PR tree-optimization/83312
3478 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3479 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3480 "val" param, and to cope with arbitrary basic blocks.
3481 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3482 handle NULL_TREE for "val", dropping "bb" param.
3483 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3484 drop "bb" param. Handle NULL_TREE for "val".
3485 (find_case_label_for_value): Make "switch_stmt" param const.
3486 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3488 (vrp_prop::check_all_array_refs): Reimplement as...
3489 (check_array_bounds_dom_walker::before_dom_children): ...this new
3490 vfunc. Replace linear search through BB block list, excluding
3491 those with non-executable in-edges via dominator walk.
3493 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3495 * config/arm/arm.opt (mverbose-cost-dump): New option.
3496 * config/arm/arm.c (arm_rtx_costs): Use it.
3498 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3501 * reload1.c (emit_input_reload_insns): Skip debug markers.
3503 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3505 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3509 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3510 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3511 * config/sh/sh.c: Skip debug insns besides notes.
3512 * config/sh/sh.md: Likewise.
3513 * config/sh/sh_treg_combine.cc: Likewise.
3514 * config/sh/sync.md: Likewise.
3516 2017-12-14 Tom de Vries <tom@codesourcery.com>
3518 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3519 Add item for weak_undefined.
3521 2017-12-14 Richard Biener <rguenther@suse.de>
3523 PR tree-optimization/67842
3524 * sese.h (bb_in_region): Remove #if 0'ed code.
3526 2017-12-14 Richard Biener <rguenther@suse.de>
3528 PR tree-optimization/83326
3529 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3530 may_be_zero parameter and handle it by not marking the first
3531 peeled copy as not exiting the loop.
3532 (try_peel_loop): Likewise.
3533 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3534 to handle the case of constant or zero iterations and perform
3535 loop header copying on-the-fly.
3537 2017-12-14 Richard Biener <rguenther@suse.de>
3539 PR tree-optimization/83418
3540 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3541 Instead of asserting we don't get unfolded comparisons deal with them.
3543 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3546 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3547 basic blocks. Assert debug bind insns don't appear outside of bbs,
3548 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3551 PR tree-optimization/83198
3552 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3553 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3554 value if arg is a REAL_CST with incompatible type.
3556 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3557 Bin Cheng <bin.cheng@arm.com>
3560 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3562 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3565 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3567 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3568 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3570 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3571 Alan Hayward <alan.hayward@arm.com>
3572 David Sherwood <david.sherwood@arm.com>
3574 * poly-int.h: New file.
3575 * poly-int-types.h: Likewise.
3576 * coretypes.h: Include them.
3577 (POLY_INT_CONVERSION): Define.
3578 * target.def (estimated_poly_value): New hook.
3579 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3580 * doc/tm.texi: Regenerate.
3581 * doc/poly-int.texi: New file.
3582 * doc/gccint.texi: Include it.
3583 * doc/rtl.texi: Describe restrictions on subreg modes.
3584 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3585 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3586 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3587 * targhooks.h (default_estimated_poly_value): Declare.
3588 * targhooks.c (default_estimated_poly_value): New function.
3589 * target.h (estimated_poly_value): Likewise.
3590 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3591 (wi::unary_traits): Delete.
3592 (wi::binary_traits::signed_shift_result_type): Define for
3593 offset_int << HOST_WIDE_INT, etc.
3594 (generic_wide_int::operator <<=): Define for all types and use
3595 wi::lshift instead of <<.
3596 (wi::hwi_with_prec): Add a default constructor.
3597 (wi::ints_for): New class.
3598 (operator <<): Define for all wide-int types.
3599 (operator /): New function.
3600 (operator %): Likewise.
3601 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3602 (ASSERT_MAYBE_NE_AT): New macros.
3604 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3605 Dominik Vogt <vogt@linux.vnet.ibm.com>
3608 * emit-rtl.c (init_emit): Remove ??? comment.
3609 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3610 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3611 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3612 alignment of 3 virtual registers to BITS_PER_WORD.
3614 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3616 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3618 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3619 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3621 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3624 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3626 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3628 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3629 with the exception of debug begin stmt markers appear before
3633 * final.c (rest_of_handle_final): Call variable_tracking_main only
3634 if !flag_var_tracking.
3636 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3637 Jakub Jelinek <jakub@redhat.com>
3641 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3642 labels debug stmts that can only appear after labels.
3644 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3646 PR rtl-optimization/82398
3647 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3648 EXPR_USEFULNESS in priority comparison.
3650 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3652 PR rtl-optimization/83393
3653 * combine.c (move_deaths): If reg_stat points to a too new insn in
3654 last_death, do not use it: find the proper insn instead.
3656 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3659 * config/rs6000/rs6000-string.c (expand_block_compare,
3660 expand_strn_compare): Fix set_mem_size() calls.
3662 2017-12-12 Jeff Law <law@redhat.com>
3664 PR tree-optimization/83298
3665 PR tree-optimization/83362
3666 PR tree-optimization/83383
3667 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3668 push_value_range a public interface. Add new argument to
3669 record_ranges_from_stmt.
3670 * gimple-ssa-evrp-analyze.c
3671 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3672 Update comments. Handle recording temporary equivalences.
3673 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3674 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3675 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3676 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3677 gimple-ssa-evrp-analyze.h.
3678 (record_temporary_equivalences_from_phis): Add new argument. When
3679 the PHI arg is an SSA_NAME, set the result's range to the range
3681 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3682 from statements too.
3683 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3684 Pass it down to children as needed.
3685 (thread_outgoing_edges): Likewise.
3686 (thread_across_edge): Likewise. Push/pop range state as needed.
3687 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3689 2017-12-12 Julia Koval <julia.koval@intel.com>
3691 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3692 (PTA_CANNONLAKE): Remove PTA_CLWB.
3694 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3697 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3698 (vcondv2div2df): Likewise.
3699 (vconduv2dfv2di): Likewise.
3701 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3703 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3704 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3705 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3706 ATTR_MATHFN_FPROUNDING_ERRNO.
3708 2017-12-12 Richard Biener <rguenther@suse.de>
3710 PR tree-optimization/83385
3711 * graphite-scop-detection.c (get_order, order): Remove.
3712 (bb_to_rpo): New global.
3714 (build_scops): Sort pbbs in RPO order.
3716 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3718 * combine.c (simplify_set): Do not transform subregs to zero_extends
3719 if the destination mode is a vector mode.
3721 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3723 PR tree-optimization/83359
3724 * tree-cfg.h (fold_loop_internal_call): Declare.
3725 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3726 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3727 (find_loop_dist_alias): New function.
3728 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3729 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3732 PR tree-optimization/80631
3733 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3734 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3735 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3736 hardcoding zero as the value if COND_EXPR is never true. For
3737 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3738 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3739 hardcoding MAX_EXPR as the reduction operation.
3740 (is_nonwrapping_integer_induction): Allow negative step.
3741 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3742 vect_create_epilog_for_reduction, if no value is suitable, don't
3743 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3745 2017-12-12 Richard Biener <rguenther@suse.de>
3747 PR tree-optimization/81889
3748 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3749 range info from the non-wrapping IV instead of just the range
3752 2017-12-12 Julia Koval <julia.koval@intel.com>
3754 * config.gcc: Add vaesintrin.h.
3755 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3757 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3758 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3760 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3761 * config/i386/immintrin.h: Include vaesintrin.h.
3762 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3763 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3764 _mm_aesdec_epi128): New intrinsics.
3766 2017-12-12 Julia Koval <julia.koval@intel.com>
3768 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3769 OPTION_MASK_ISA_VAES_UNSET): New.
3770 (ix86_handle_option): Handle -mvaes.
3771 * config/i386/cpuid.h: Define bit_VAES.
3772 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3773 * config/i386/i386-c.c (__VAES__): New.
3774 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3775 (ix86_valid_target_attribute_inner_p): Ditto.
3776 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3777 * config/i386/i386.opt: Add -mvaes.
3778 * doc/invoke.texi: Ditto.
3780 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3782 * debug.h (gcc_debug_hooks): Add inline_entry.
3783 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3784 * debug.c (do_nothing_debug_hooks): Likewise.
3785 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3786 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3787 (dwarf2_lineno_debug_hooks): Likewise.
3789 * common.opt (gstatement-frontiers): New, setting
3790 debug_nonbind_markers_p.
3791 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3792 * toplev.c (process_options): Autodetect value for debug statement
3794 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3795 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3797 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3798 markers. Integrate source bind into debug stmt expand loop.
3799 (pass_expand::execute): Check debug marker limit. Avoid deep
3800 TER and expand debug locations for debug bind insns only.
3801 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3803 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3804 * final.c (reemit_insn_block_notes): Take current block from
3805 nonbind markers. Declare note where it's first set.
3806 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3807 begin stmt markers if enabled.
3808 (notice_source_line): Handle nonbind markers. Fail if their
3809 location is unknown or that of builtins.
3810 (rest_of_handle_final): Convert begin stmt markers to notes if
3811 var-tracking didn't run.
3812 (rest_of_clean_state): Skip begin stmt markers.
3813 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3815 * function.c (allocate_struct_function): Set begin_stmt_markers.
3816 * function.h (struct function): Add debug_marker_count counter
3817 and debug_nonbind_markers flag.
3818 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3819 * gimple-low.c (lower_function_body): Adjust
3820 debug_nonbind_markers.
3821 (lower_stmt): Drop or skip gimple debug stmts.
3822 (lower_try_catch): Skip debug stmts.
3823 * gimple.c (gimple_build_debug_begin_stmt): New.
3824 (gimple_copy): Increment debug_marker_count if copying one.
3825 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3826 * gimplify.c (rexpr_location): New.
3827 (rexpr_has_location): New.
3828 (warn_switch_unreachable_r): Handle gimple debug stmts.
3829 (shortcut_cond_r): Call expr_location.
3831 (find_goto_label): New.
3832 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3834 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3836 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3837 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3838 (LANG_HOOKS_INITIALIZER): ... this.
3839 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3840 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3842 (update_ebb_live_info): Skip debug insn markers.
3843 * lra.c (debug_insn_static_data): Rename to...
3844 (debug_bind_static_data): ... this.
3845 (debug_marker_static_data): New.
3846 (lra_set_insn_recog_data): Select one of the above depending
3848 (lra_update_isn_regno_info): Don't assume debug insns have
3850 (push_insns): Skip debug insns.
3851 * lto-streamer-in.c (input_function): Drop debug stmts
3852 depending on active options. Adjust debug_nonbind_markers.
3853 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3854 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3855 begin stmt marker notes.
3856 (print_insn): Likewise.
3857 * recog.c (extract_insn): Recognize rtl for debug markers.
3858 * rtl.def (DEBUG_MARKER): New.
3859 * tree-inline.c: Include params.h.
3860 (remap_gimple_stmt): Handle nonbind markers.
3861 (maybe_move_debug_stmts_to_successors): Likewise.
3862 (copy_debug_stmt): Likewise.
3863 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3864 markers regardless of no side effects.
3865 (tsi_link_before): Don't update container's side effects when adding
3866 a begin stmt marker.
3867 (tsi_link_after): Likewise.
3868 (expr_first): Skip begin stmt markers.
3869 (expr_last): Likewise.
3870 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3871 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3872 Disregard nonbind markers.
3873 * tree.c (make_node_stat): Don't set side effects for begin stmt
3875 (build1_stat): Likewise.
3876 * tree.def (DEBUG_BEGIN_STMT): New.
3877 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3878 * var-tracking.c (delete_debug_insns): Renamed to...
3879 (delete_vta_debug_insns): ... this.
3880 (reemit_marker_as_note): New.
3881 (vt_initialize): Reemit markers.
3882 (delete_vta_debug_insns): Likewise.
3883 (vt_debug_insns_local): Reemit or delete markers.
3884 (variable_tracking_main_1): Likewise.
3885 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3886 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3887 (gimple_debug_nonbind_marker_p): New.
3888 (gimple_build_debug_bind): Adjust.
3889 (gimple_build_debug_begin_stmt): New.
3890 * doc/invoke.texi (max-debug-marker-count): New param.
3891 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3892 (debug_parameter_ref, debug_marker): New.
3893 (NOTE_INSN_BEGIN_STMT): New.
3894 (DEBUG_INSN): Describe begin stmt markers.
3896 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3897 without debug insns, we wouldn't, but clean up debug insns
3898 after a control flow insn nevertheless.
3900 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3901 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3902 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3903 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3904 (rtl_tidy_fallthru_edge): Likewise.
3905 (rtl_verify_fallthru): Likewise.
3906 (rtl_verify_bb_layout): Likewise.
3907 (skip_insns_after_block): Likewise.
3908 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3909 * dwarf2out.c: Include print-rtl.h.
3910 (dwarf2out_next_real_insn): New.
3911 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3912 Dump debug binds in asm comments.
3913 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3914 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3915 callers to use gsi_start_nondebug_bb instead.
3916 (gsi_after_labels): Skip gimple debug stmts.
3917 (gsi_start_nondebug): New.
3918 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3919 (proper_loop_form_for_interchange): Adjust.
3920 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3921 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3922 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3923 (collect_fallthrough_labels): Likewise.
3924 (should_warn_for_implicit_fallthrough): Likewise.
3925 (warn_implicit_fallthrough_r): Likewise.
3926 (expand_FALLTHROUGH_r): Likewise.
3927 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3928 (graphite_copy_stmts_from_block): Skip nonbind markers.
3929 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3930 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3931 * jump.c (clean_barriers): Skip debug insns.
3932 * omp-expand.c (expand_parallel_call): Skip debug insns.
3933 (expand_task_call): Likewise.
3934 (remove_exit_barrier): Likewise.
3935 (expand_omp_taskreg): Likewise.
3936 (expand_omp_for_init_counts): Likewise.
3937 (expand_omp_for_generic): Likewise.
3938 (expand_omp_for_static_nochunk): Likewise.
3939 (expand_omp_for_static_chunk): Likewise.
3940 (expand_omp_simd): Likewise.
3941 (expand_omp_taskloop_for_outer): Likewise.
3942 (expand_omp_taskloop_for_inner): Likewise.
3943 (expand_oacc_for): Likewise.
3944 (expand_omp_sections): Likewise.
3945 (expand_omp_single): Likewise.
3946 (expand_omp_synch): Likewise.
3947 (expand_omp_atomic_load): Likewise.
3948 (expand_omp_atomic_store): Likewise.
3949 (expand_omp_atomic_fetch_op): Likewise.
3950 (expand_omp_atomic_pipeline): Likewise.
3951 (expand_omp_atomic_mutex): Likewise.
3952 (expand_omp_target): Likewise.
3953 (grid_expand_omp_for_loop): Likewise.
3954 (grid_expand_target_grid_body): Likewise.
3955 (build_omp_regions_1): Likewise.
3956 * omp-low.c (check_combined_parallel): Skip debug stmts.
3957 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3958 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3959 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3961 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3962 (make_edges): Likewise.
3963 (cleanup_dead_labels): Likewise.
3964 (gimple_can_merge_blocks_p): Likewise.
3965 (stmt_starts_bb_p): Likewise.
3966 (gimple_block_label): Likewise.
3967 (gimple_redirect_edge_and_branch): Likewise.
3968 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3970 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3972 * tree-pretty-print (print_declaration): Omit initializer in slim
3974 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3976 (eliminate_unnecessary_stmts): Stabilize block removal order.
3977 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3978 * var-tracking.c (get_first_insn): New.
3979 (vt_emit_notes): Call it.
3980 (vt_initialize): Walk any insns before the first BB.
3981 (delete_debug_insns): Likewise.
3983 * gimple.h (enum gimple_debug_subcode): Add
3984 GIMPLE_DEBUG_BEGIN_STMT.
3985 (gimple_debug_begin_stmt_p): New.
3986 (gimple_debug_nonbind_marker_p): New.
3987 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3988 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3989 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3990 * insn-notes.def (BEGIN_STMT): New.
3991 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3992 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3993 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3994 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3995 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3996 (INSN_DEBUG_MARKER_KIND): New.
3997 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3998 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3999 (INSN_VAR_LOCATION_PTR): New.
4000 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4002 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4003 deps only if debug bind insns are enabled.
4004 (pass_expand::execute): Avoid deep TER and expand
4005 debug locations for debug bind insns only.
4006 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4007 debug stmts special handling down to debug bind stmts.
4008 * combine.c (try_combine): Narrow debug insns special handling
4009 down to debug bind insns.
4010 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4011 Narrow debug insns preexisting special handling down to debug
4013 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4014 handling down to debug bind insns.
4015 * function.c (instantiate_virtual_regs): Skip debug markers,
4016 adjust handling of debug binds.
4017 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4018 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4019 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4020 insns to debug bind insns.
4021 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4022 special handling of debug stmts to debug bind stmts.
4023 * ipa-split.c (split_function): Likewise.
4024 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4025 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4027 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4028 * regrename.c (build_def_use): Likewise.
4029 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4030 (pass_cprop_hardreg): Narrow special casing of debug insns to
4032 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4033 * reload1.c (reload): Likewise.
4034 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4035 casing of debug stmts to debug bind stmts.
4036 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4037 * ssa-iterators.h (num_imm_uses): Likewise.
4038 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4039 debug stmts to debug bind stmts.
4040 * tree-inline.c (tree_function_versioning): Narrow special casing
4041 of debug stmts to debug bind stmts.
4042 * tree-loop-distribution.c (generate_loops_for_partition):
4043 Narrow special casing of debug stmts to debug bind stmts.
4044 * tree-sra.c (analyze_access_subtree): Narrow special casing
4045 of debug stmts to debug bind stmts.
4046 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4047 stmts to debug bind stmts.
4048 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4049 casing of debug stmts to debug bind stmts.
4050 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4051 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4052 casing of debug stmts to debug bind stmts.
4053 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4055 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4056 debug stmts to debug bind stmts.
4057 (gimple_replace_ssa_lhs): Likewise.
4058 (insert_debug_temp_for_var_def): Likewise.
4059 (insert_debug_temps_for_defs): Likewise.
4060 (reset_debug_uses): Likewise.
4061 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4062 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4063 (adjust_debug_stmts): Likewise.
4064 (adjust_phi_and_debug_stmts): Likewise.
4065 (vect_do_peeling): Likewise.
4066 * tree-vect-loop.c (vect_transform_loop): Likewise.
4067 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4068 * var-tracking.c (adjust_mems): Narrow special casing of debug
4069 insns to debug bind insns.
4070 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4071 (compute_bb_dataflow, vt_find_locations): Likewise.
4072 (vt_expand_loc, emit_notes_for_changes): Likewise.
4073 (vt_init_cfa_base): Likewise.
4074 (vt_emit_notes): Likewise.
4075 (vt_initialize): Likewise.
4076 (vt_finalize): Likewise.
4078 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4079 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4080 (next_nonnote_nondebug_insn_bb): New.
4081 (prev_nonnote_nondebug_insn_bb): New.
4082 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4083 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4084 (prev_nonnote_nondebug_insn_bb): Declare.
4085 (next_nonnote_nondebug_insn_bb): Declare.
4086 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4087 * cfgrtl.c (get_last_bb_insn): Likewise.
4088 * lra.c (push_insns): Likewise.
4090 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4093 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4095 * selftest.c (selftest::test_runner::test_runner): New ctor.
4096 (selftest::test_runner::~test_runner): New dtor.
4097 * selftest.h (class selftest::test_runner): New class.
4099 2017-12-11 Carl Love <cel@us.ibm.com>
4101 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4102 vec_extract_fp32_from_shortl]): Add #defines.
4103 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4104 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4105 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4106 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4107 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4108 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4109 * doc/extend.texi: Add documentation for the added builtins.
4111 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4113 PR rtl-optimization/80693
4114 PR rtl-optimization/81019
4115 PR rtl-optimization/81020
4116 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4117 are not mentioned in i3. Place the REG_UNUSED note on i2,
4118 possibly modified to REG_DEAD, if it did not originate in i3.
4120 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4122 * recog.c (store_data_bypass_p_1): New function.
4123 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4124 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4126 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4128 PR rtl-optimization/83361
4129 * ifcvt.c (if_convert): Call fixup_partitions.
4131 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4133 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4134 early folding of splat_u{8,16,32}.
4136 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4138 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4139 output_operand_lossage first argument with capital letter.
4140 (aarch64_override_options): Don't start error and sorry first argument
4141 with capital letter.
4143 2017-12-11 Andi Kleen <ak@linux.intel.com>
4145 PR gcov-profile/83355
4146 * auto-profile.c (string_table::get_index_by_decl): Don't
4147 recurse when abstract origin points to itself.
4149 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4151 PR tree-optimization/83320
4152 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4153 (prune_datarefs_not_in_loop): Ditto.
4155 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4157 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4159 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4161 PR tree-optimization/83337
4162 * gimple-loop-interchange.cc (compute_access_stride): Handle
4163 bitfield DRs properly.
4165 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4167 PR tree-optimization/83338
4168 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4169 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4170 vector of unsigned integers to vector of signed integers.
4172 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4174 PR rtl-optimization/83317
4175 * lra-constraints.c (process_address_1): Add insn code check.
4177 2017-12-08 Michael Matz <matz@suse.de>
4179 Fix PR tree-optimization/83323
4180 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4181 head-controlled loops and loop BBs.
4182 * common.opt (funroll-and-jam): Remove, instead ...
4183 (floop-unroll-and-jam): ... reuse this option.
4184 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4185 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4186 (-floop-unroll-and-jam): ... this option.
4188 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4190 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4191 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4192 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4194 2017-12-08 Richard Biener <rguenther@suse.de>
4197 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4198 handle accesses outside of zero-sized vars.
4200 2017-12-08 Martin Jambor <mjambor@suse.cz>
4202 PR tree-optimization/83141
4203 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4204 test for MEM_REFs implicitely changing types with padding. Remove
4206 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4208 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4210 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4212 * config/avr/avr.c (avr_attribute_table): Likewise.
4213 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4214 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4215 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4216 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4217 of struct attribute_spec.
4218 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4220 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4223 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4225 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4228 * profile-count.c (profile_count::from_gcov_type): Move from
4229 profile-count.h; handle overflow.
4230 * profile-count.h (profile_count::from_gcov_type): Move offline.
4232 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4234 PR rtl-optimization/83304
4235 * combine.c (move_deaths): If we do not know where a register died,
4238 2017-12-08 Richard Biener <rguenther@suse.de>
4240 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4241 Provide -fopt-info-loop feedback when we interchange in a nest.
4243 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4245 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4246 for armv6 ARM CPU IDs.
4248 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4250 * common/config/arm/arm-common.c: Include <algorithm>.
4251 (INCLUDE_VECTOR): Define.
4252 (compare_opt_names): New function.
4253 (arm_rewrite_selected_arch): Only strip out extensions that can be
4254 expressed through -mfpu. Sort the remaining extensions
4257 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4259 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4260 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4261 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4262 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4264 2017-06-08 Tristan Gingold <gindold@adacore.com>
4267 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4268 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4270 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4272 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4273 of struct attribute_spec.
4275 2017-12-08 Julia Koval <julia.koval@intel.com>
4277 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4278 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4279 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4280 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4281 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4282 _mm_maskz_dpwssds_epi32): Ditto.
4284 2017-12-08 Richard Biener <rguenther@suse.de>
4286 PR tree-optimization/81303
4287 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4288 conditions try to create a comparison vector type matching
4289 the data vector type.
4290 (vectorizable_condition): Adjust.
4291 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4292 Leave invariant conditions alone in case we can vectorize those.
4294 2017-12-08 Julia Koval <julia.koval@intel.com>
4296 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4297 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4298 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4299 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4300 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4302 2017-12-08 Julia Koval <julia.koval@intel.com>
4304 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4305 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4306 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4307 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4308 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4309 _mm_maskz_dpbusds_epi32): New intrinsics.
4311 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4313 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4315 (ld<bh>io_signed): Likewise.
4316 (st<bhw>io): Likewise.
4317 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4318 SMALL_INT12 constant integer operand.
4320 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4322 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4324 (nios2_large_constant_allowed): ...this. Adjust uses.
4325 (nios2_plus_symbolic_constant_p): Rename to...
4326 (nios2_plus_large_constant_p): ...this. Adjust uses.
4327 (nios2_legitimate_address_p): Correct CONST_INT handling.
4328 (nios2_symbolic_memory_operand_p): Rename to...
4329 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4330 (nios2_large_constant_p): Check for large constant integers too.
4331 (nios2_split_large_constant): Handle constant integers.
4332 (nios2_split_symbolic_memory_operand): Rename to...
4333 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4334 (nios2_legitimize_constant_address): Handle constant integers.
4335 (r0rel_constant_p): Handle small constant integers.
4336 (nios2_print_operand_address): Handle r0-relative integer addresses.
4337 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4338 * config/nios2/nios2.md: Adjust for renamed functions.
4340 2017-12-07 Andrew Waterman <andrew@sifive.com>
4342 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4343 (TARGET_HAVE_SRODATA_SECTION): New define.
4344 (riscv_select_section): New function.
4346 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4347 Alexander Monakov <amonakov@ispras.ru>
4348 Jakub Jelinek <jakub@redhat.com>
4351 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4353 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4355 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4356 a bool success value. Don't call output_operand_lossage here.
4357 (aarch64_print_ldpstp_address): Return a bool success value.
4358 (aarch64_print_operand_address): Call output_addr_const if
4359 aarch64_print_address_internal fails.
4360 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4361 'y'; call output_operand_lossage instead. Call output_operand_lossage
4362 if aarch64_print_ldpstp_address fails.
4364 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4366 * tree-vector-builder.h
4367 (tree_vector_builder::binary_encoded_nelts): Declare.
4368 * tree-vector-builder.c
4369 (tree_vector_builder::binary_encoded_nelts): New function.
4370 * fold-const.c (negate_expr_p): Likewise.
4371 (operand_equal_p, fold_checksum_tree): Likewise.
4372 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4373 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4374 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4375 (uniform_vector_p): Likewise.
4376 * varasm.c (const_hash_1, compare_constant): Likewise.
4377 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4378 (valid_lattice_transition): Operate directly on the VECTOR_CST
4380 * ipa-icf.c: Include tree-vector-builder.h.
4381 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4382 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4384 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4386 * tree.c (build_vector): Delete.
4387 * tree.h (build_vector): Make static and move into the self-testing
4390 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4392 * vector-builder.h (vector_builder::derived): New const overload.
4393 (vector_builder::elt): New function.
4394 * tree-vector-builder.h (tree_vector_builder::type): New function.
4395 (tree_vector_builder::apply_step): Declare.
4396 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4398 * gimple-fold.h (tree_vector_builder): Declare.
4399 (gimple_build_vector): Take a tree_vector_builder instead of a
4400 type and vector of elements.
4401 * gimple-fold.c (gimple_build_vector): Likewise.
4402 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4404 (get_initial_defs_for_reduction): Likewise.
4405 (vectorizable_induction): Likewise.
4407 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4409 * tree-vector-builder.h
4410 (tree_vector_builder::new_binary_operation): Declare.
4411 * tree-vector-builder.c
4412 (tree_vector_builder::new_binary_operation): New function.
4413 * fold-const.c (fold_relational_const): Use it.
4414 (const_binop): Likewise. Check that both input vectors have
4415 the same number of elements, thus excluding things like WIDEN_SUM.
4416 Check whether it is possible to operate directly on the encodings
4419 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4421 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4422 new_unary_operation, operating only on the encoded elements.
4423 (const_unop): Likewise.
4424 (exact_inverse): Likewise.
4425 (distributes_over_addition_p): New function.
4426 (const_binop): Use tree_vector_builder and new_unary_operation
4427 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4428 on the encoded elements unless the encoding is strided and the
4429 operation does not distribute over addition.
4430 (fold_convert_const): Use tree_vector_builder and
4431 new_unary_operation. Operate only on the encoded elements
4432 for truncating integer conversions, or for non-stepped encodings.
4434 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4436 * config/sparc/sparc.c: Include tree-vector-builder.h.
4437 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4438 * expmed.c: Include tree-vector-builder.h.
4439 (make_tree): Use tree_vector_builder instead of build_vector.
4440 * fold-const.c: Include tree-vector-builder.h.
4441 (const_binop): Use tree_vector_builder instead of build_vector.
4442 (const_unop): Likewise.
4443 (native_interpret_vector): Likewise.
4444 (fold_vec_perm): Likewise.
4445 (fold_ternary_loc): Likewise.
4446 * gimple-fold.c: Include tree-vector-builder.h.
4447 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4449 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4450 (simplify_vector_constructor): Use tree_vector_builder instead
4452 * tree-vect-generic.c: Include tree-vector-builder.h.
4453 (add_rshift): Use tree_vector_builder instead of build_vector.
4454 (expand_vector_divmod): Likewise.
4455 (optimize_vector_constructor): Likewise.
4456 * tree-vect-loop.c: Include tree-vector-builder.h.
4457 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4458 of build_vector. Explicitly use a stepped encoding for
4460 * tree-vect-slp.c: Include tree-vector-builder.h.
4461 (vect_get_constant_vectors): Use tree_vector_builder instead
4463 (vect_transform_slp_perm_load): Likewise.
4464 (vect_schedule_slp_instance): Likewise.
4465 * tree-vect-stmts.c: Include tree-vector-builder.h.
4466 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4467 (vect_gen_perm_mask_any): Likewise.
4468 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4469 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4471 (build_vector_from_val): Likewise. Explicitly use a duplicate
4474 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4476 * doc/generic.texi (VECTOR_CST): Describe new representation of
4478 * vector-builder.h: New file.
4479 * tree-vector-builder.h: Likewise.
4480 * tree-vector-builder.c: Likewise.
4481 * Makefile.in (OBJS): Add tree-vector-builder.o.
4482 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4483 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4484 (tree_vector): Change the number of elements to
4485 vector_cst_encoded_nelts.
4486 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4487 (VECTOR_CST_ELTS): Delete.
4488 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4489 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4490 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4491 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4492 (VECTOR_CST_ENCODED_ELT): Likewise.
4493 (vector_cst_encoded_nelts): New function.
4494 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4495 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4496 (vector_cst_int_elt, vector_cst_elt): Declare.
4497 * tree.c: Include tree-vector-builder.h.
4498 (tree_code_size): Abort if passed VECTOR_CST.
4499 (tree_size): Update for new VECTOR_CST layout.
4500 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4501 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4502 (build_vector): Use tree_vector_builder.
4503 (vector_cst_int_elt, vector_cst_elt): New functions.
4504 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4505 encoded elements and then create the vector in the canonical form.
4506 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4507 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4508 (tree_c_tests): Call test_vector_cst_patterns.
4509 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4511 (hash_tree): Likewise.
4512 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4513 (streamer_write_tree_header): Likewise.
4514 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4515 (streamer_alloc_tree): Likewise. Update call to make_vector.
4516 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4518 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4520 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4521 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4522 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4524 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4525 Richard Biener <rguenther@suse.de>
4527 PR tree-optimization/81303
4528 * Makefile.in (gimple-loop-interchange.o): New object file.
4529 * common.opt (floop-interchange): Reuse the option from graphite.
4530 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4531 -floop-interchange and mention it for -O3.
4532 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4533 * gimple-loop-interchange.cc: New file.
4534 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4535 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4536 * passes.def (pass_linterchange): New pass.
4537 * timevar.def (TV_LINTERCHANGE): New time var.
4538 * tree-pass.h (make_pass_linterchange): New declaration.
4539 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4540 interchange. Record IV before/after increment in new parameters.
4541 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4542 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4544 (check_reduction_path): ...New function here.
4545 * tree-vectorizer.h (check_reduction_path): New declaration.
4547 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4550 PR rtl-optimization/80818
4551 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4552 always early clobbered.
4553 * lra-lives.c (process_bb_lives): Check input hard regs for early
4554 clobbered non-operand hard reg.
4556 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4559 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4560 types_compatible_p, just that TYPE_MODE is the same.
4562 2017-12-07 Martin Sebor <msebor@redhat.com>
4565 * attribs.c (empty_attribute_table): Initialize new member of
4566 struct attribute_spec.
4567 (decl_attributes): Add argument. Handle mutually exclusive
4568 combinations of attributes.
4569 (selftests::test_attribute_exclusions): New function.
4570 (selftests::attribute_c_tests): Ditto.
4571 * attribs.h (decl_attributes): Add default argument.
4572 * selftest.h (attribute_c_tests): Declare.
4573 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4574 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4576 * doc/extend.texi (Common Function Attributes): Update const and pure.
4577 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4578 of struct attribute_spec.
4579 * config/arc/arc.c (arc_attribute_table): Same.
4580 * config/arm/arm.c (arm_attribute_table): Same.
4581 * config/avr/avr.c ( avr_attribute_table): Same.
4582 * config/bfin/bfin.c (bfin_attribute_table): Same.
4583 * config/cr16/cr16.c (cr16_attribute_table): Same.
4584 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4585 * config/h8300/h8300.c (h8300_attribute_table): Same.
4586 * config/i386/i386.c (ix86_attribute_table): Same.
4587 * config/ia64/ia64.c (ia64_attribute_table): Same.
4588 * config/m32c/m32c.c (m32c_attribute_table): Same.
4589 * config/m32r/m32r.c (m32r_attribute_table): Same.
4590 * config/m68k/m68k.c (m68k_attribute_table): Same.
4591 * config/mcore/mcore.c (mcore_attribute_table): Same.
4592 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4593 * config/mips/mips.c (mips_attribute_table): Same.
4594 * config/msp430/msp430.c (msp430_attribute_table): Same.
4595 * config/nds32/nds32.c (nds32_attribute_table): Same.
4596 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4597 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4598 * config/rl78/rl78.c (rl78__attribute_table): Same.
4599 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4600 * onfig/rx/rx.c (rx_attribute_table): Same.
4601 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4602 * config/sh/sh.c (sh_attribute_table): Same.
4603 * config/sparc/sparc.c (sparc_attribute_table): Same.
4604 * config/spu/spu.c (spu_attribute_table): Same.
4605 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4606 * config/v850/v850.c (v850_attribute_table): Same.
4607 * config/visium/visium.c (visium_attribute_table): Same.
4609 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4612 * config/arm/arm.c (INCLUDE_STRING): Define.
4613 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4614 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4617 2017-12-07 Michael Matz <matz@suse.de>
4619 Add unroll and jam pass
4621 * gimple-loop-jam.c: New file.
4622 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4623 * common.opt (funroll-and-jam): New option.
4624 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4625 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4626 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4627 * passes.def: Add pass_loop_jam.
4628 * timevar.def (TV_LOOP_JAM): Add.
4629 * tree-pass.h (make_pass_loop_jam): Declare.
4630 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4631 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4632 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4633 to flow_loop_tree_node_add.
4634 (duplicate_subloops, copy_loops_to): Append to sibling list.
4635 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4636 * doc/invoke.texi (-funroll-and-jam): Document new option.
4637 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4639 2017-12-07 Richard Biener <rguenther@suse.de>
4641 PR tree-optimization/83296
4642 PR tree-optimization/67769
4643 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4644 flow sensitive info in an unrelated BB.
4645 (value_replacement): Use reset_flow_sensitive_info.
4646 (minmax_replacement): Reset flow sensitive info on the def
4647 we move. Do not reset flow sensitive info in the whole BB
4648 we move the stmt to.
4649 (abs_replacement): Likewise.
4651 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4654 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4655 rs6000_cpu to the given -mcpu=, or to the default processor.
4657 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4659 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4660 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4661 instead of rs6000_cpu_attr.
4662 (rs6000_adjust_cost): Ditto.
4663 (is_microcoded_insn): Ditto.
4664 (rs6000_adjust_priority): Ditto.
4665 (rs6000_issue_rate): Ditto.
4666 (rs6000_use_sched_lookahead): Ditto.
4667 (rs6000_use_sched_lookahead_guard): Ditto.
4668 (rs6000_sched_reorder): Ditto.
4669 (force_new_group): Ditto.
4670 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4671 (group_ending_nop): Ditto.
4673 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4675 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4677 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4678 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4680 (rs6000_reassociation_width): Ditto.
4681 (rs6000_emit_epilogue): Ditto.
4682 (rs6000_adjust_cost): Ditto.
4683 (is_microcoded_insn): Ditto.
4684 (is_cracked_insn): Ditto.
4685 (rs6000_adjust_priority): Ditto.
4686 (rs6000_sched_reorder): Ditto.
4687 (rs6000_sched_reorder2): Ditto.
4688 (insn_must_be_first_in_group): Ditto.
4689 (insn_must_be_last_in_group): Ditto.
4690 (rs6000_register_move_cost): Ditto.
4691 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4694 2017-12-07 Julia Koval <julia.koval@intel.com>
4696 * config.gcc: Add vaesintrin.h.
4697 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4699 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4700 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4702 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4703 * config/i386/immintrin.h: Include vaesintrin.h.
4704 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4705 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4706 _mm_aesdec_epi128): New intrinsics.
4708 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4710 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4711 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4712 c-family/c-spellcheck.cc.
4713 (best_macro_match::best_macro_match): Likewise.
4715 (struct edit_distance_traits<cpp_hashnode *>): Move to
4716 c-family/c-spellcheck.h.
4717 (class best_macro_match): Likewise.
4719 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4721 PR tree-optimization/83293
4722 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4723 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4724 might insert into empty bb.
4727 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4728 simplify for plus with :c added, and pointer_plus without that.
4729 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4730 with undefined overflow and the conversion is not widening,
4731 perform negation in utype and only convert to type afterwards.
4732 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4733 simplify for plus with :c added, and pointer_plus without that.
4734 If type is integral with undefined overflow and the conversion is
4735 not widening, perform minus in utype and only convert to type
4736 afterwards. Move the last pointer_diff_expr simplify into the
4739 2017-12-06 Martin Sebor <msebor@redhat.com>
4741 PR tree-optimization/82646
4742 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4743 strncpy, not maxlen.
4745 2017-12-06 Martin Sebor <msebor@redhat.com>
4747 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4750 PR tree-optimization/83075
4751 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4752 strncat/strncpy don't change length of source string.
4754 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4757 2017-11-29 Martin Aberg <maberg@gaisler.com>
4759 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4760 to prevent b2bst errata sequence.
4761 (sqrtdf2_fix): Likewise.
4763 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4765 PR tree-optimization/81945
4766 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4767 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4768 to dest_cfun has orig_loop_num set, either remap it to the new
4769 loop number if the loop got moved too, or clear it.
4771 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4773 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4775 (thunderx2t99_nothing): New insn reservation.
4776 (thunderx2t99_mrs): New insn reservation.
4777 (thunderx2t99_multiple): New insn reservation.
4778 (thunderx2t99_alu_basi): Add bfx to reservation.
4779 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4781 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4784 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4786 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4788 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4789 tree-ssa-pre.c::remove_dead_inserted_code.
4790 * tree-ssa-dce.h: New file.
4791 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4792 (remove_dead_inserted_code): Move and rename to function
4793 tree-ssa-dce.c::simple_dce_from_worklist.
4794 (pass_pre::execute): Update use.
4796 2017-12-05 Richard Biener <rguenther@suse.de>
4798 PR tree-optimization/83277
4799 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4800 to code-gen liveout vars.
4802 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4804 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4805 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4806 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4809 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4811 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4813 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4814 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4817 2017-12-05 Richard Biener <rguenther@suse.de>
4819 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4820 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4821 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4822 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4824 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4826 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4828 2017-12-05 Martin Liska <mliska@suse.cz>
4829 Jakub Jelinek <jakub@redhat.com>
4831 * doc/invoke.texi: Document the options.
4832 * flag-types.h (enum sanitize_code): Add
4833 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4834 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4835 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4836 * opts.c: Define new sanitizer options.
4837 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4838 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4840 2017-12-05 Julia Koval <julia.koval@intel.com>
4842 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4843 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4844 (ix86_handle_option): Handle -mavx512vnni.
4845 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4846 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4847 * config/i386/i386-c (__AVX512VNNI__): New.
4848 * config/i386/i386.c (ix86_target_string): Handle new option.
4849 (ix86_valid_target_attribute_inner_p): Handle new option.
4850 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4851 * config/i386/i386.opt (mavx512vnni): New option.
4853 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4856 * athlon.md: Disable for generic.
4857 * haswell.md: Enable for generic.
4858 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4859 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4861 (ix86_adjust_cost): Move generic to haswell path.
4863 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4865 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4866 instead of MEM_P in a couple more places. Fix formatting issues.
4868 2017-12-04 Jim Wilson <jimw@sifive.com>
4870 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4871 instead of GP_REG_LAST-1.
4872 (riscv_adjust_libcall_cfi_prologue): Likewise.
4873 (riscv_adjust_libcall_cri_epilogue): Likewise.
4874 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4877 2017-12-04 Luis Machado <luis.machado@linaro.org>
4879 * ipa-pure-const.c (check_decl): Add missing newline.
4880 (state_from_flags): Likewise.
4882 2017-12-04 Jeff Law <law@redhat.com>
4884 PR tree-optimizatin/78496
4885 * gimple-ssa-evrp-analyze.h
4886 (evrp_range_analyzer::get_vr_values): Simplify.
4887 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4888 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4889 and gimple-ssa-evrp-analyze.h.
4890 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4891 (simplify_stmt_for_jump_threading): Copy a blob of code from
4892 tree-vrp.c to use ranges to simplify statements.
4893 (dom_opt_dom_walker::before_dom_children): Call
4894 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4895 (dom_opt_dom_walker::after_dom_children): Similarly for
4896 evrp_range_analyzer::leave.
4897 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4900 * gimple-ssa-evrp-analyze.c
4901 (evrp_range_analyzer::extract_range_from_stmt): Always use
4902 vr_values::update_value_range so preexisting range info is
4903 medged with new range info, even if the new range is VR_VARYING.
4905 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4907 * combine.c: Adjust comment.
4908 (use_crosses_set_p): Delete.
4909 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4910 (try_combine): Ditto.
4912 2017-12-04 Richard Biener <rguenther@suse.de>
4914 PR tree-optimization/83255
4915 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4916 Re-add zero-iteration check.
4918 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4920 PR rtl-optimization/83245
4921 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4922 hard registers as earlyclobber, also if not in an asm.
4924 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4928 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4931 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4932 rs6000_cpu based on cpu_index, not tune_index.
4934 2017-12-04 Richard Biener <rguenther@suse.de>
4936 PR tree-optimization/83238
4937 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4938 code match comment, rejecting invalid SESE regions.
4940 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4942 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4943 require base operand is a REG_POINTER prior to reload on targets
4944 with non-equivalent space registers.
4946 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4948 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4949 (update_specialized_profile): Fix updating of counts.
4950 (perhaps_add_new_callers): Likewise.
4952 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4955 * x86-tune.def: Remove obsolette FIXMEs.
4956 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4957 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4958 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4960 (X86_TUNE_PAD_RETURNS): Disable for generic.
4962 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4964 PR tree-optimization/83170
4965 PR tree-optimization/83241
4966 * gimple-ssa-store-merging.c
4967 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4968 gimple_vuse (ins_stmt) in case it has changed.
4969 (imm_store_chain_info::output_merged_store): Likewise.
4971 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4975 * tree-cfg.c (pass_warn_function_return::execute): Handle
4976 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4977 with BUILTINS_LOCATION.
4981 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4982 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4983 instead of DECL_MODE.
4985 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4986 last argument to rtx pointer.
4987 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4988 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4989 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4990 depending on the chosen ISAs.
4991 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4992 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4993 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4995 * config/i386/sse.md (mov<mode>_internal): Likewise.
4996 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4998 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5000 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5002 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5004 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5005 parameters from prototype.
5006 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5007 parameters. Don't print a comment.
5008 (emit_fusion_gpr_load): Adjust.
5009 (emit_fusion_load_store): Adjust.
5010 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5011 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5012 comment on the second line.
5014 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5018 rs6000_cpu based on cpu_index, not tune_index.
5020 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5022 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5023 which_alternative instead of which_alternative + 1.
5024 (output_asm_insn): Print an extra tab if the template is short.
5026 2017-12-01 Jim Wilson <jimw@sifive.com>
5028 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5030 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5031 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5033 * doc/tm.texi: Regenerate.
5035 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5038 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5039 whether we can allocate pseudos before trying to fix an address.
5040 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5041 memory address is indexed or indirect.
5042 (floatuns_<mode>si2_hw2): Likewise.
5044 2017-12-01 Jason Merrill <jason@redhat.com>
5046 * Makefile.in (TAGS): Add c-family/*.cc.
5048 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5050 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5051 (call_value_insn): Likewise.
5052 (sibcall_insn): Likewise.
5053 (sibcall_value_insn): Likewise.
5054 (movsi_aarch64): Likewise.
5055 (movdi_aarch64): Likewise.
5056 (add_losym_): Likewise.
5057 (ldr_got_small_): Likewise.
5058 (ldr_got_small_sidi): Likewise.
5059 (ldr_got_small_28k_): Likewise.
5060 (ldr_got_small_28k_sidi): Likewise.
5061 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5062 Move output_addr_const to symbolic case. Add error check.
5064 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5066 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5067 * config/s390/s390.md: Disable MVC merging peephole if it would
5068 disable operand forwarding.
5069 (new peephole2): Split MVCs if it would turn them into up to 2
5072 2017-12-01 Richard Biener <rguenther@suse.de>
5074 PR tree-optimization/83232
5075 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5076 detection of same access. Instead of breaking the group here
5077 do not consider the duplicate. Add comment explaining real fix.
5079 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5081 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5083 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5085 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5087 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5089 * function.h (struct function): Remove cilk_frame_decl,
5090 is_cilk_function and calls_cilk_spawn fields.
5091 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5093 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5094 cilk_elemental field.
5095 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5096 * target.def: Adjust comment.
5097 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5098 Don't test cilk_elemental.
5100 PR tree-optimization/83233
5101 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5102 bswap_stat name for the struct.
5105 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5106 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5107 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5108 SWITCH_BREAK_LABEL_P set on the label.
5109 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5110 added for default case if it was missing and not all cases covered.
5111 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5112 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5114 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5115 to avoid -Wimplicit-fallthrough warning.
5116 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5117 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5119 PR tree-optimization/83221
5120 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5122 (init_reassoc): Formatting fix.
5125 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5126 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5130 * tree-cfg.c: Include asan.h.
5131 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5133 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5135 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5137 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5139 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5140 vpcmpeqd instruction.
5142 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5144 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5145 registers type for 128bit mode.
5147 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5149 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5151 * gengtype-state.c (read_state_pair): Likewise.
5152 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5153 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5154 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5155 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5156 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5157 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5158 * builtins.c (fold_builtin_3): Likewise.
5159 * graphite-scop-detection.c
5160 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5161 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5164 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5167 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5168 parameter. Return true if ignore_decl_rtl_set_p is true and other
5169 conditions are satisfied.
5170 * asan.h (asan_protect_global): Add new parameter.
5171 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5172 to asan_protect_global calls.
5174 2017-11-30 Jim Wilson <jimw@sifive.com>
5176 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5177 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5178 -mstrict-align, add info on default value. Delete redundant lines for
5179 -mabi. Add missing -mexplicit-relocs docs.
5181 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5183 * config/arc/arc.md (trap): New pattern.
5185 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5187 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5188 instruction to end into a delay slot.
5189 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5190 can be placed into a delay slot against reg_note.
5192 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5194 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5195 labels number of usages.
5197 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5199 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5201 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5203 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5206 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5207 multiplication by power of 2 constant into two shifts + comparison.
5209 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5212 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5214 2017-11-30 Richard Biener <rguenther@suse.de>
5216 PR tree-optimization/83202
5217 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5218 allow_peel argument and guard peeling.
5219 (canonicalize_loop_induction_variables): Likewise.
5220 (canonicalize_induction_variables): Pass false.
5221 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5222 peeling from cunrolli.
5224 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5226 * combine.c (try_combine): Print a message to dump file whenever
5227 I0, I1, or I2 cannot be combined into I3.
5229 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5231 PR rtl-optimization/83156
5232 PR rtl-optimization/82621
5233 * combine.c (try_combine): Don't split an I2 if one of the dests is
5234 set again before I3. Allow unused dests.
5236 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5238 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5240 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5242 PR rtl-optimization/80818
5243 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5244 recursively. Use insn code for clobber.
5245 (lra_set_insn_recog_data): Pass the new arg to
5246 collect_non_operand_hard_regs.
5247 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5249 (lra_update_insn_regno_info): Pass insn to
5250 add_regs_to_insn_regno_info.
5252 2017-11-29 Jim Wilson <jimw@sifive.com>
5253 Andrew Waterman <andrew@sifive.com>
5255 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5256 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5257 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5258 use SINGLE_SHIFT_COST.
5259 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5260 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5262 2017-11-29 Julia Koval <julia.koval@intel.com>
5264 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5265 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5266 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5267 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5268 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5269 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5270 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5271 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5272 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5273 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5274 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5275 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5276 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5277 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5278 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5279 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5280 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5281 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5282 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5283 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5284 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5285 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5286 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5287 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5288 __builtin_ia32_vpshldv_v2di_mask,
5289 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5290 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5291 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5293 2017-11-29 Julia Koval <julia.koval@intel.com>
5295 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5296 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5297 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5298 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5299 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5300 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5301 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5302 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5303 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5304 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5305 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5306 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5307 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5308 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5309 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5310 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5311 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5312 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5313 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5314 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5315 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5317 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5319 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5320 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5323 2017-11-29 Martin Aberg <maberg@gaisler.com>
5325 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5326 to prevent b2bst errata sequence.
5327 (sqrtdf2_fix): Likewise.
5329 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5331 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5332 (div_sqrt_insn_p): New function.
5333 (sparc_do_work_around_errata): Insert NOP instructions to
5334 prevent sequences that could trigger the TN-0013 errata for
5335 certain LEON3 processors.
5336 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5337 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5338 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5339 (in_branch_delay): Prevent div and sqrt in delay slot if
5341 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5343 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5345 * config/sparc/sparc.c (atomic_insn_p): New function.
5346 (sparc_do_work_around_errata): Insert NOP instructions to
5347 prevent sequences that could trigger the TN-0010 errata for
5349 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5350 instruction referable in atomic_insns_p.
5352 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5354 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5355 (atomic_compare_and_swap_leon3_1): Likewise.
5358 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5360 * config/sparc/sparc.c (fpop_insn_p): New function.
5361 (sparc_do_work_around_errata): Insert NOP instructions to
5362 prevent sequences that could trigger the TN-0012 errata for
5364 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5365 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5366 (in_branch_annul_delay): Prevent floating-point instructions
5367 in delay slot of annulled integer branch.
5369 2017-11-29 Richard Biener <rguenther@suse.de>
5371 PR tree-optimization/83202
5372 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5374 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5375 nodes vectorized to the same stmts multiple times.
5376 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5377 (vect_analyze_slp_instance): Adjust.
5378 (scalar_stmts_to_slp_tree_map_t): New typedef.
5379 (vect_schedule_slp_instance): Add a map recording the SLP node
5380 representing the vectorized stmts for a set of scalar stmts.
5381 Avoid code-generating redundancies.
5382 (vect_schedule_slp): Allocate map and pass it down.
5384 2017-11-29 Nathan Sidwell <nathan@acm.org>
5387 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5390 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5392 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5393 cases for printing LDP/STP memory addresses.
5394 (aarch64_print_address_internal): Renamed from
5395 aarch64_print_operand_address, added parameter, add Pmode check.
5396 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5397 (aarch64_print_operand_address): Indirect to
5398 aarch64_print_address_internal.
5399 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5400 'y' operand output specifier.
5402 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5405 * tree.c (build_simple_mem_ref_loc): Handle
5406 get_addr_base_and_unit_offset returning a MEM_REF.
5409 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5412 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5415 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5417 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5419 (layout::move_to_column): Likewise.
5421 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5424 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5425 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5427 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5428 Martin Jambor <mjambor@suse.cz>
5431 * tree.h (expr_type_first_operand_type_p): Declare
5432 * tree.c (expr_type_first_operand_type_p): New function.
5433 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5434 (ipa_value_from_jfunc): Adjust declaration.
5435 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5436 Use it as result type for arithmetics, unless it is NULL in which case
5437 be more conservative.
5438 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5439 ipa_get_jf_pass_through_result.
5440 (propagate_vals_across_pass_through): Likewise.
5441 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5442 is to propagate_vals_across_pass_through.
5443 (propagate_constants_across_call): Pass PARM_TYPE to
5444 propagate_scalar_across_jump_function.
5445 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5446 ipa_value_from_jfunc.
5447 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5448 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5449 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5450 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5451 target_type, pass it to ipa_value_from_jfunc.
5452 (update_indirect_edges_after_inlining): Pass parameter type to
5453 try_make_edge_direct_simple_call.
5455 2017-11-28 Jeff Law <law@redhat.com>
5457 * gimple-ssa-evrp-analyze.c
5458 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5459 refine ranges if scev_initialized_p returns true.
5460 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5462 2017-11-28 Julia Koval <julia.koval@intel.com>
5464 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5465 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5466 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5467 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5468 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5469 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5470 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5471 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5472 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5473 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5474 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5475 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5476 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5477 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5478 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5479 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5480 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5481 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5482 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5483 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5484 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5485 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5487 2017-11-28 Julia Koval <julia.koval@intel.com>
5489 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5490 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5491 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5492 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5493 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5494 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5495 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5496 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5497 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5498 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5499 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5500 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5501 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5502 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5503 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5504 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5505 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5506 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5507 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5508 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5509 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5510 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5511 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5512 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5513 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5514 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5515 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5516 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5517 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5518 __builtin_ia32_vpshld_v2di_mask): New builtins.
5519 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5520 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5522 2017-11-28 Richard Biener <rguenther@suse.de>
5524 PR tree-optimization/80776
5525 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5527 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5529 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5530 If the incoming edge is an effective fallthru because the other
5531 edge only reaches a __builtin_unreachable () then record ranges
5532 derived from the controlling condition in SSA info.
5533 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5534 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5536 2017-11-28 Olivier Hainque <hainque@adacore.com>
5538 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5539 on mingw build hosts.
5541 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5544 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5546 2017-11-28 Richard Biener <rguenther@suse.de>
5549 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5550 copies generated from memcpy use a character array as reference
5553 2017-11-28 Julia Koval <julia.koval@intel.com>
5554 Sebastian Peryt <sebastian.peryt@intel.com>
5556 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5557 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5558 cilk-common.o, cilk.h, cilk-common.c): Remove.
5560 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5561 * builtins.c (is_builtin_name): Remove cilkplus condition.
5562 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5563 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5564 cilk-builtins.def, cilkplus.def): Remove.
5565 * cif-code.def (CILK_SPAWN): Remove.
5566 * cilk-builtins.def: Delete.
5567 * cilk-common.c: Ditto.
5569 * cilkplus.def: Ditto.
5570 * config/darwin.h (fcilkplus): Delete.
5571 * cppbuiltin.c: Ditto.
5572 * doc/extend.texi: Remove cilkplus doc.
5573 * doc/generic.texi: Ditto.
5574 * doc/invoke.texi: Ditto.
5575 * doc/passes.texi: Ditto.
5576 * gcc.c (fcilkplus): Remove.
5577 * gengtype.c (cilk.h): Remove.
5578 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5580 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5582 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5584 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5585 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5586 cilkplus conditions.
5587 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5588 inline_read_section): Ditto.
5589 * ipa-inline-analysis.c (cilk.h): Remove.
5590 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5591 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5592 append_linker_options): Remove condition for fcilkplus.
5593 * lto/lto-lang.c (cilk.h): Remove.
5594 (lto_init): Remove condition for fcilkplus.
5595 * omp-expand.c (expand_cilk_for_call): Delete.
5596 (expand_omp_taskreg, expand_omp_for_static_chunk,
5597 expand_omp_for): Remove cilkplus
5599 (expand_cilk_for): Delete.
5600 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5601 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5602 execute_lower_omp, diagnose_sb_0): Ditto.
5603 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5604 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5605 * tree-nested.c: Ditto.
5606 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5607 (dump_generic_node): Ditto.
5608 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5609 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5611 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5613 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5615 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5616 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5617 for vector mode and !TARGET_SIMD.
5619 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5621 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5623 * tree.h (SWITCH_LABELS): Remove.
5624 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5625 assert SWITCH_BODY is non-NULL.
5626 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5628 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5630 PR tree-optimization/80788
5631 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5632 has TREE_OVERFLOW set, call drop_tree_overflow.
5634 2017-11-28 Richard Biener <rguenther@suse.de>
5636 PR tree-optimization/83158
5637 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5639 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5642 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5643 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5645 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5647 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5648 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5650 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5653 * varasm.c (compare_constant): Take the mode of the constants into
5654 account when comparing floating point constants.
5656 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5658 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5659 from explicit instantiation of debug_helper.
5660 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5662 2017-11-27 Richard Biener <rguenther@suse.de>
5664 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5667 2017-11-27 Richard Biener <rguenther@suse.de>
5669 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5670 (type_hash_canon): Also clear int_cst_hash_table entry for
5672 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5674 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5676 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5677 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5678 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5680 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5682 * hash-map.h (gt_cleare_cache): Avoid UB.
5684 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5686 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5687 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5688 (decide_unroll_runtime_iterations): Likewise.
5689 (decide_unroll_stupid): Likewise.
5691 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5694 * config/i386/i386.md: Add a loop with incssp.
5696 2017-11-27 Martin Jambor <mjambor@suse.cz>
5698 PR tree-optimization/81248
5699 * tree-sra.c (splice_param_accesses): Remove size check.
5700 (decide_one_param_reduction): Fix size check.
5701 * gimple-pretty-print.c (dump_profile): Silence warning.
5702 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5704 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5707 * dbxout.c (lastlineno): New variable.
5708 (dbx_debug_hooks): Use dbxout_switch_text_section as
5709 switch_text_section debug hook.
5710 (dbxout_function_end): Switch to current_function_section
5711 rather than function_section. If crtl->has_bb_partition,
5712 output just one N_FUN, depending on in_cold_section_p.
5713 (dbxout_source_line): Remember last lineno in lastlineno.
5714 (dbxout_switch_text_section): New function.
5715 (dbxout_function_decl): Adjust dbxout_block caller.
5716 (dbx_block_with_cold_children): New function.
5717 (dbxout_block): Return true if any LBRAC/RBRAC have been
5718 emitted. Use dbx_block_with_cold_children at depth == 0
5719 in second partition. Add PARENT_BLOCKNUM argument, pass
5720 it optionally adjusted to children. Output LBRAC/RBRAC
5721 around recursive call only if the block is in the current
5722 partition, if not and anything was output, emit empty
5724 * final.c (final_scan_insn): Compute cold_function_name
5725 before calling switch_text_section debug hook. Call
5726 that hook even if dwarf2out_do_frame if not emitting
5730 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5731 TREE_READONLY decls.
5733 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5735 PR rtl-optimization/82488
5736 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5738 2017-11-26 Julia Koval <julia.koval@intel.com>
5740 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5742 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5745 2017-11-26 Julia Koval <julia.koval@intel.com>
5747 * config/i386/driver-i386.c (host_detect_local_cpu):
5748 Detect skylake-avx512.
5750 2017-11-26 Julia Koval <julia.koval@intel.com>
5752 * config.gcc: Add -march=cannonlake.
5753 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5754 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5755 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5756 (PTA_CANNONLAKE): New.
5757 (processor_target_table): Add cannonlake.
5758 (ix86_option_override_internal): Ditto.
5759 (fold_builtin_cpu): Ditto.
5760 (get_builtin_code_for_version): Handle cannonlake.
5761 (M_INTEL_COREI7_CANNONLAKE): New.
5762 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5763 * doc/invoke.texi: Add -march=cannonlake.
5765 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5767 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5768 (try_init_one_plugin): Alternative implementation for MinGW.
5769 * Makefile.in (plugin_implib): New.
5770 (gengtype-lex.c): Fix broken AIX workaround.
5771 * configure: Regenerate.
5772 * doc/plugins.texi: Document support for MinGW.
5774 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5776 PR rtl-optimization/81553
5777 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5778 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5779 is a shift where C1 has different mode than the whole shift, use C1's
5780 mode for MULT rather than the shift's mode.
5783 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5784 builtins not enabled in the currently selected ISA.
5786 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5788 PR tree-optimization/71026
5789 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5790 (insert_reciprocals): Change to insert reciprocals before a division
5791 by a square and to insert the square of a reciprocal.
5792 (execute_cse_reciprocals_1): Change to consider division by a square.
5793 (register_division_in): Add importance parameter.
5795 2017-11-24 Richard Biener <rguenther@suse.de>
5797 PR tree-optimization/82402
5798 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5799 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5801 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5803 * match.pd (0-ptr): New transformation.
5805 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5808 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5809 when skipping checking.
5811 2017-11-24 Richard Biener <rguenther@suse.de>
5813 PR tree-optimization/83128
5814 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5815 (vn_reference_lookup_3): Likewise.
5817 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5820 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5821 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5824 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5825 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5826 check for the latter with gimple_call_builtin_p. Do not handle
5827 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5829 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5831 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5834 2017-11-23 Julia Koval <julia.koval@intel.com>
5836 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5837 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5838 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5839 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5840 _mm512_maskz_expandloadu_epi16): New intrinsics.
5841 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5842 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5843 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5844 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5845 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5846 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5847 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5848 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5849 _mm256_maskz_expandloadu_epi8): New intrinsics.
5850 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5851 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5852 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5853 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5854 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5855 * config/i386/sse.md (VI248_VLBW): New iterator.
5856 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5858 2017-11-23 Julia Koval <julia.koval@intel.com>
5860 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5861 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5862 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5863 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5864 _mm512_mask_compressstoreu_epi16): New.
5865 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5866 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5867 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5868 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5869 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5870 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5871 _mm256_mask_compressstoreu_epi8): New.
5872 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5873 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5874 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5875 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5876 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5877 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5878 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5879 __builtin_ia32_compresshi128_mask,
5880 __builtin_ia32_compressstoreuqi512_mask,
5881 __builtin_ia32_compressstoreuhi512_mask,
5882 __builtin_ia32_compressstoreuqi256_mask,
5883 __builtin_ia32_compressstoreuqi128_mask,
5884 __builtin_ia32_compressstoreuhi256_mask,
5885 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5886 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5888 (ix86_expand_special_args_builtin): Handle new types.
5889 (s4fma_expand): Handle new builtin array.
5890 * config/i386/immintrin.h: Include new headers.
5891 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5892 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5894 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5897 fixes. Declare temp and g variables at the top in order to avoid
5898 {} in most of the cases.
5900 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5902 * match.pd (ptr-0): New transformation.
5904 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5906 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5907 (struct addr_mode_cost_table): New.
5908 (struct tune_params): Add field addr_mode_costs.
5909 * config/arm/arm.c (generic_addr_mode_costs): New.
5910 (arm_slowmul_tune): Initialise addr_mode_costs field.
5911 (arm_fastmul_tune): Likewise.
5912 (arm_strongarm_tune): Likewise.
5913 (arm_xscale_tune): Likewise.
5914 (arm_9e_tune): Likewise.
5915 (arm_marvell_pj4_tune): Likewise.
5916 (arm_v6t2_tune): Likewise.
5917 (arm_cortex_tune): Likewise.
5918 (arm_cortex_a8_tune): Likewise.
5919 (arm_cortex_a7_tune): Likewise.
5920 (arm_cortex_a15_tune): Likewise.
5921 (arm_cortex_a35_tune): Likewise.
5922 (arm_cortex_a53_tune): Likewise.
5923 (arm_cortex_a57_tune): Likewise.
5924 (arm_exynosm1_tune): Likewise.
5925 (arm_xgene1_tune): Likewise.
5926 (arm_cortex_a5_tune): Likewise.
5927 (arm_cortex_a9_tune): Likewise.
5928 (arm_cortex_a12_tune): Likewise.
5929 (arm_cortex_a73_tune): Likewise.
5930 (arm_v7m_tune): Likewise.
5931 (arm_cortex_m7_tune): Likewise.
5932 (arm_v6m_tune): Likewise.
5933 (arm_fa726te_tune): Likewise.
5934 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5936 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5938 * config/arm/arm.c (arm_mem_costs): New function.
5939 (arm_rtx_costs_internal): Use arm_mem_costs.
5941 2017-11-23 Mark Wielaard <mark@klomp.org>
5943 * dwarf2out.c (init_sections_and_labels): Use generation to create
5944 unique ranges_section_label and ranges_base_label. Return generation.
5945 (output_rnglists): Add generation argument. Use generation to create
5946 unique ranges labels.
5947 (dwarf2out_finish): Get generation from init_sections_and_labels
5948 and pass generation to output_rnglists.
5950 2017-11-23 Mike Stump <mikestump@comcast.net>
5951 Eric Botcazou <ebotcazou@adacore.com>
5953 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5954 * cfgloop.h (struct loop): Add unroll field.
5955 * function.h (struct function): Add has_unroll bitfield.
5956 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5957 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5958 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5959 (pass_rtl_unroll_loops::gate): Likewise.
5960 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5961 for which loop->unroll==1.
5962 (decide_unroll_constant_iterations): Use note for consistency and
5963 take loop->unroll into account. Return early if loop->unroll is set.
5964 Fix thinko in existing test.
5965 (decide_unroll_runtime_iterations): Use note for consistency and
5966 take loop->unroll into account.
5967 (decide_unroll_stupid): Likewise.
5968 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5969 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5970 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5972 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5973 (print_loop): Print loop->unroll if set.
5974 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5975 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5976 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5978 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5979 loop->unroll is set and smaller than the trip count. Otherwise bypass
5980 entirely the heuristics if loop->unroll is set. Remove dead note.
5981 Fix off-by-one bug in other note.
5982 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5983 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5985 (tree_unroll_loops_completely): Make static.
5986 (pass_complete_unroll::execute): Use correct type for variable.
5987 (pass_complete_unrolli::execute): Fix formatting.
5988 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5990 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5992 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5993 enable when TARGET_PREFER_AVX128 is set.
5995 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5997 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5999 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6000 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6002 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6004 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6005 frequencies. Fix estimation of aggregate parameters.
6007 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6009 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6012 2017-11-23 Tom de Vries <tom@codesourcery.com>
6014 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6016 2017-11-23 Tom de Vries <tom@codesourcery.com>
6018 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6021 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6024 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6025 sibcall_value_pcrel_fdpic): Use local variable instead of
6027 (calli_tbr_rel): Add missing operand 2.
6028 (call_valuei_tbr_rel): Add missing operand 3.
6030 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6033 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6034 bitpos/bitsize covering the whole destination, use store_expr only if
6035 the complex mode is the same. Otherwise, use expand_normal and if
6036 it returns CONCAT, subreg each part separately instead of trying to
6037 subreg the whole result.
6039 2017-11-23 Richard Biener <rguenther@suse.de>
6041 PR tree-optimization/23094
6042 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6043 come here from walking over backedges in the first iteration.
6044 (vn_reference_lookup_3): Skip clobbers that store the same value.
6046 2017-11-23 Richard Biener <rguenther@suse.de>
6048 PR tree-optimization/81403
6049 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6050 a block we need a leader relative to.
6051 (phi_translate_1): For nary processing require a leader from
6052 get_representative_for given we run expression simplification
6053 using match-and-simplify. Remove previous fix.
6055 2017-11-22 Jeff Law <law@redhat.com>
6057 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6058 Use new method allocate_value_range rather than accessing the
6059 vrp_value_range_pool data member directly.
6060 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6061 to use extract_range_from_stmt method to avoid need for
6062 extract_range_from_assignment method.
6063 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6064 method rather than setting values_propgated data member directly.
6065 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6066 and values propagated data members and extract_range_from_assignment
6067 method. Reorder private data members to conform to standards.
6068 Add new methods set_lattice_propagation_complete and
6069 allocate_value_range.
6071 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6073 PR rtl-optimization/83030
6074 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6075 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6076 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6077 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6078 CROSSING_JUMP_P flag is already set before setting it.
6079 * cfgrtl.c (fixup_partition_crossing): Likewise.
6080 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6083 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6085 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6086 Handle the case where both arguments are using gen_const_vec_series.
6088 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6091 * pretty-print.c (pp_format): Move quoting implementation to
6092 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6093 to pass address of "quote" local.
6094 (pp_begin_quote): New function.
6095 (pp_end_quote): New function.
6096 * pretty-print.h (printer_fn): Convert penultimate param from bool
6098 (pp_begin_quote): New decl.
6099 (pp_end_quote): New decl.
6100 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6101 param from bool to bool *.
6102 * tree-diagnostic.h (default_tree_printer): Likewise.
6104 2017-11-22 Jeff Law <law@redhat.com>
6106 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6107 Initialize vr_values.
6108 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6109 vr_values via vr_values class instance rather than delegators.
6110 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6111 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6112 (evrp_range_analyzer::push_value_range): Likewise.
6113 (evrp_range_analyzer::pop_value_range): Likewise.
6114 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6115 most delegators. Those remaining are exposed as public interfaces.
6116 Make vr_values a pointer and private.
6117 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6119 (evrp_range_analyzer::get_vr_value): New method.
6120 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6121 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6122 (class evrp_dom_walker): Attach evrp_folder class, initialize
6123 it in the ctor. Remove temporary delegators.
6124 (evrp_dom_walker::before_dom_children): Call methods in attached
6125 evrp_range_analyzer class via class instance pointer. Use
6126 free value_range_constant_singleton to remove need for
6127 op_with_constant_singleton_value delegator method. Do not
6128 create a vrp_prop class instance for every call! Narrow
6129 scope of a couple variables.
6130 (evrp_dom_walker::cleanup): Call methods in attached
6131 evrp_range_analyzer class via class instance pointer.
6132 * vr-values.h (class vr_values): Privatize many methods and
6135 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6137 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6138 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6139 * expr.c (expand_expr_real_2): Likewise.
6140 * fold-const.c (const_unop): Likewise.
6141 * optabs-tree.c (optab_for_tree_code): Likewise.
6142 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6143 * tree-inline.c (estimate_operator_cost): Likewise.
6144 * tree-pretty-print.c (dump_generic_node): Likewise.
6145 (op_code_prio): Likewise.
6146 (op_symbol_code): Likewise.
6147 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6148 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6149 * internal-fn.c (direct_internal_fn_optab): New function.
6150 (direct_internal_fn_array, direct_internal_fn_supported_p
6151 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6152 * fold-const-call.c (fold_const_reduction): New function.
6153 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6155 * tree-vect-loop.c: Include internal-fn.h.
6156 (reduction_code_for_scalar_code): Rename to...
6157 (reduction_fn_for_scalar_code): ...this and return an internal
6159 (vect_model_reduction_cost): Take an internal_fn rather than
6161 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6163 (vectorizable_reduction): Use internal functions rather than tree
6164 codes for the reduction operation. Update calls to the functions
6166 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6167 Use calls to internal functions rather than REDUC tree codes.
6168 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6170 2017-11-22 Olivier Hainque <hainque@adacore.com>
6172 * config/vxworks.c (vxworks_override_options): Pick default
6173 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6174 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6175 DWARF_GNAT_ENCODINGS_DEFAULT.
6176 * config/vxworksae.h: Likewise.
6178 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6180 PR tree-optimization/83104
6181 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6184 2017-11-22 Marek Polacek <polacek@redhat.com>
6185 H.J. Lu <hongjiu.lu@intel.com>
6186 Jason Merrill <jason@redhat.com>
6191 * calls.c (initialize_argument_information): Call
6192 warn_parameter_passing_abi target hook.
6193 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6194 argument onto stack.
6195 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6196 * common.opt: Update -fabi-version description.
6197 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6198 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6200 (ix86_is_empty_record): New function.
6201 (ix86_warn_parameter_passing_abi): New function.
6202 (TARGET_EMPTY_RECORD_P): Redefine.
6203 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6204 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6205 * doc/tm.texi: Regenerated.
6206 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6207 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6208 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6209 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6210 instead of int_size_in_bytes.
6211 * expr.c (copy_blkmode_to_reg): Likewise.
6212 * function.c (aggregate_value_p): Return 0 for empty types.
6213 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6215 (locate_and_pad_parm): Call arg size_in_bytes instead
6217 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6218 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6219 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6221 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6222 (std_gimplify_va_arg_expr): Skip empty records. Call
6223 arg_size_in_bytes instead size_in_bytes.
6224 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6225 * tree-core.h (tree_type_common): Add empty_flag.
6226 (tree_decl_common): Update comments.
6227 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6229 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6230 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6232 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6233 * tree.c (default_is_empty_type): New function.
6234 (default_is_empty_record): New function.
6235 (arg_int_size_in_bytes): New function.
6236 (arg_size_in_bytes): New function.
6237 (get_ultimate_context): New function.
6238 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6239 TRANSLATION_UNIT_WARN_EMPTY_P.
6240 (default_is_empty_record, arg_int_size_in_bytes,
6241 arg_size_in_bytes, get_ultimate_context): Declare.
6243 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6245 * config/arm/arm.c (cmse_clear_registers): New function.
6246 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6247 code by call to cmse_clear_registers.
6248 (cmse_nonsecure_entry_clear_before_return): Likewise.
6250 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6252 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6253 (vdot_s32, vdotq_s32): New.
6254 (vdot_lane_u32, vdotq_lane_u32): New.
6255 (vdot_lane_s32, vdotq_lane_s32): New.
6258 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6261 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6262 for unsigned values with fewer HWIs than the precision.
6263 (test_overflow): New function.
6264 (wide_int_cc_tests): Call it.
6266 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6268 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6269 has the same mode class as Pmode.
6271 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6273 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6274 padding_bits_to_clear_ptr.
6275 (cmse_nonsecure_entry_clear_before_return): Likewise.
6277 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6279 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6280 auto_sbitap instead of integer bitfield to control register needing
6283 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6285 PR tree-optimization/83044
6286 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6287 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6288 that need the upper bound. Subtract offset from
6289 get_addr_base_and_unit_offset only if positive and subtract it
6290 before division by eltsize rather than after it.
6293 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6294 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6295 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6299 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6300 expand_binop with *mul_widen_optab, make sure at least one of the
6301 operands doesn't have VOIDmode.
6304 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6306 PR rtl-optimization/82044
6307 PR tree-optimization/82042
6308 * dse.c (record_store): Check for overflow.
6309 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6310 add_wild_read instead of clear_rhs_from_active_local_stores on
6311 overflow and log it into dump_file.
6313 2017-11-22 Richard Biener <rguenther@suse.de>
6315 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6316 fake edges to exit when looking for a place to insert.
6317 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6319 (insert_into_preds_of_block): Commit edge insertion immediately,
6320 assert that doesn't require new BBs.
6321 (fini_pre): Release expressions.
6322 (pass_pre::execute): Shuffle things around a bit, if the fn
6323 is too large do not compute AVAIL either as this is really the
6326 2017-11-22 Richard Biener <rguenther@suse.de>
6328 PR tree-optimization/83089
6329 * tree-if-conv.c (pass_if_conversion::execute): If anything
6330 changed reset SCEV and free the number of iteration estimates.
6332 2017-11-21 Martin Sebor <msebor@redhat.com>
6334 PR tree-optimization/82945
6335 * calls.h (warn_nonstring_bound): Remove unused function.
6337 2017-11-21 Martin Sebor <msebor@redhat.com>
6339 PR tree-optimization/82945
6340 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6341 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6342 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6344 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6345 * calls.h (get_attr_nonstring_decl): Declare new function.
6346 * doc/extend.texi (attribute nonstring): Update.
6347 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6348 get_attr_nonstring_decl and handle it.
6349 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6350 detection of nul-termination.
6351 (strlen_to_stridx): Change to a pointer.
6352 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6353 (pass_strlen::execute): Same.
6355 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6357 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6358 for the new option -mprefer-vector-width=[none|128|256|512].
6359 * config/i386/i386.c (ix86_target_string): remove old style options
6360 -mprefer-avx256 and make -mprefer-avx128 as alias.
6361 (ix86_option_override_internal): Apply defaults for the
6362 -mprefer-vector-width=[128|256] option.
6363 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6364 Implement macros to work with -mprefer-vector-width=.
6365 * config/i386/i386.opt: Implemented option
6366 -mprefer-vector-width=[none|128|256|512].
6367 * doc/invoke.texi: Documentation for
6368 -mprefer-vector-width=[none|128|256|512].
6370 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6372 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6374 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6376 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6378 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6380 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6382 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6383 * expr.c (expand_expr_real_2): Likewise.
6384 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6385 fold_binary_loc): Likewise.
6386 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6387 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6388 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6389 MINUS_EXPR transformations.
6390 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6391 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6392 * tree-inline.c (estimate_operator_cost): Likewise.
6393 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6394 op_symbol_code): Likewise.
6395 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6396 * vr-values.c (extract_range_from_binary_expr): Likewise.
6397 * varasm.c (initializer_constant_valid_p_1): Likewise.
6398 * tree.def: New tree code POINTER_DIFF_EXPR.
6400 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6402 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6403 integer suffix to movbe mnemonic.
6404 (*bswaphi2_movbe): Ditto.
6405 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6407 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6410 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6411 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6412 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6414 2017-11-21 Martin Liska <mliska@suse.cz>
6416 * tree-inline.c (expand_call_inline): Remove not needed
6419 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6420 Jakub Jelinek <jakub@redhat.com>
6423 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6424 Remove static keyword from f variable.
6426 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6428 PR tree-optimization/83086
6429 * gimple-ssa-store-merging.c
6430 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6431 rather than n.base_addr.
6433 2017-11-21 Martin Liska <mliska@suse.cz>
6435 PR rtl-optimization/82044
6436 PR tree-optimization/82042
6437 * dse.c (check_mem_read_rtx): Check for overflow.
6439 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6441 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6444 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6446 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6447 and its corresponding call arg location note.
6449 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6450 Andrew Burgess <andrew.burgess@embecosm.com>
6452 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6454 (arc_return_slot_offset): Likewise.
6455 (arc_eh_return_address_location): New declaration.
6456 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6457 (MUST_SAVE_REGISTER): Add exception handler case.
6458 (MUST_SAVE_RETURN_ADDR): Likewise.
6459 (arc_frame_pointer_required): Likewise.
6460 (arc_frame_pointer_needed): New function.
6461 (arc_compute_frame_size): Changed.
6462 (arc_expand_prologue): Likewise.
6463 (arc_expand_epilogue): Likewise.
6464 (arc_initial_elimination_offset): Likewise.
6465 (arc_return_slot_offset): Delete.
6466 (arc_eh_return_address_location): New function.
6467 (arc_builtin_setjmp_frame_value): Likewise.
6468 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6469 (EH_RETURN_STACKADJ_RTX): Define.
6470 (EH_RETURN_HANDLER_RTX): Likewise.
6471 * config/arc/arc.md (eh_return): Delete.
6473 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6475 * print-tree.h (debug_vec_tree): Remove prototype.
6476 * gdbinit.in (pvt): Remove macro.
6478 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6480 PR tree-optimization/83047
6481 * gimple-ssa-store-merging.c
6482 (imm_store_chain_info::output_merged_store): If the loads with the
6483 same vuse are in different basic blocks, for load_gsi pick a load
6484 location that is dominated by the other loads.
6487 * config/i386/i386.c (ix86_memmodel_check): Start
6488 -Winvalid-memory-model diagnostics with lowercase letter.
6491 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6492 set in_cold_section_p to the partition containing loc_list->first.
6493 When seeing loc_list->last_before_switch node, update secname and
6494 perform range_across_switch second partition handling only after that.
6497 * run-rtl-passes.c: Include debug.h.
6498 (run_rtl_passes): Call debug_hooks->assembly_start.
6499 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6503 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6504 OPTAB_DIRECT in calls to expand_simple_binop.
6506 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6509 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6511 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6514 * spellcheck.h (best_match::blithely_get_best_candidate): New
6517 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6519 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6520 with lower case letter.
6522 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6524 * config/i386/i386.md (bswaphi2): New expander.
6525 (*bswaphi2_movbe): New insn pattern.
6526 (bswaphi -> rorhi pepehole2): New peephole pattern.
6528 2017-11-20 Jeff Law <law@redhat.com>
6530 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6531 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6532 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6533 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6534 gimple-ssa-evrp-analyze.h.
6536 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6537 set BB_VISITED here.
6538 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6540 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6542 * tree-predcom.c: Add general comment on Store-Store chains.
6543 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6545 (get_chain_last_ref_at): Rename into...
6546 (get_chain_last_write_at): ...this.
6547 (get_chain_last_write_before_load): New function.
6548 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6549 CT_STORE_STORE when write reference is added.
6550 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6551 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6552 (initialize_root_vars_store_elim_1): Ditto.
6553 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6554 definition is created.
6555 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6556 chain by replacing it with dominant stored value.
6558 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6560 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6562 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6564 * vr-values.c (extract_range_from_binary_expr): Use a full range
6567 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6569 * config/arm/arm.md (R4_REGNUM): Define constant.
6570 (nonsecure_call_internal): Remove r4 clobber.
6571 (nonsecure_call_value_internal): Likewise.
6572 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6573 clobber and resequence match_operands.
6574 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6575 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6576 (nonsecure_call_value_reg_thumb2): Likewise.
6578 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6580 PR tree-optimization/78821
6581 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6582 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6583 to the address of the base rather than the base itself.
6584 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6585 (find_bswap_or_nop_finalize): New function.
6586 (find_bswap_or_nop): Use it.
6587 (bswap_replace): Return a tree rather than bool, change first
6588 argument from gimple * to gimple_stmt_iterator, allow inserting
6589 into an empty sequence, allow ins_stmt to be NULL - then emit
6590 all stmts into gsi. Fix up MEM_REF address gimplification.
6591 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6592 (struct store_immediate_info): Add N and INS_STMT non-static
6594 (store_immediate_info::store_immediate_info): Initialize them
6595 from newly added ctor args.
6596 (merged_store_group::apply_stores): Formatting fixes. Sort by
6598 (stmts_may_clobber_ref_p): For stores call also
6599 refs_anti_dependent_p.
6600 (gather_bswap_load_refs): New function.
6601 (imm_store_chain_info::try_coalesce_bswap): New method.
6602 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6603 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6604 (imm_store_chain_info::output_merged_store): Fail if number of
6605 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6606 and NOP_EXPR rhs_code.
6607 (pass_store_merging::process_store): Compute n and ins_stmt, if
6608 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6609 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6611 (pass_store_merging::execute): Calculate dominators.
6613 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6614 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6615 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6616 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6617 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6618 class pass_optimize_bswap, bswap_replace,
6619 pass_optimize_bswap::execute): Moved to ...
6620 * gimple-ssa-store-merging.c: ... this file.
6621 Include optabs-tree.h.
6622 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6623 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6624 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6625 anonymous namespace, remove static keywords.
6626 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6627 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6629 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6632 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6634 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6636 * vec.h (debug_helper): New function.
6637 (DEFINE_DEBUG_VEC): New macro.
6638 * hash-set.h (debug_helper): New function.
6639 (DEFINE_DEBUG_HASH_SET): New macro.
6640 * cfg.c (debug_slim (edge)): New function.
6641 Call DEFINE_DEBUG_VEC for edges.
6642 Call DEFINE_DEBUG_HASH_SET for edges.
6643 * cfghooks.c (debug_slim (basic_block)): New function.
6644 Call DEFINE_DEBUG_VEC for basic blocks.
6645 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6646 * print-tree.c (debug_slim): New function to handle trees.
6647 Call DEFINE_DEBUG_VEC for trees.
6648 Call DEFINE_DEBUG_HASH_SET for trees.
6649 (debug (vec<tree, va_gc>) &): Remove.
6650 (debug (<vec<tree, va_gc>) *): Remove.
6651 * print-rtl.c (debug_slim): New function to handle const_rtx.
6652 Call DEFINE_DEBUG_VEC for rtx_def.
6653 Call DEFINE_DEBUG_VEC for rtx_insn.
6654 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6655 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6656 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6657 (debug (vec<rtx_insn *> *ptr): Remove.
6658 (debug_insn_vector): Remove.
6659 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6661 2017-11-20 Tom de Vries <tom@codesourcery.com>
6663 PR rtl-optimization/82020
6664 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6665 IF_THEN_ELSE condition.
6667 2017-11-19 Jeff Law <law@redhat.com>
6669 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6670 of degenerates resulting from ignoring an edge.
6672 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6675 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6677 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6680 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6681 for uninitialied counts.
6683 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6686 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6688 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6691 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6693 2017-11-19 Tom de Vries <tom@codesourcery.com>
6695 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6698 2017-11-19 Tom de Vries <tom@codesourcery.com>
6700 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6703 2017-11-19 Tom de Vries <tom@codesourcery.com>
6705 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6708 2017-11-19 Tom de Vries <tom@codesourcery.com>
6710 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6712 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6714 2017-11-19 Tom de Vries <tom@codesourcery.com>
6716 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6718 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6719 semicolon after MCORE_EXPORT_NAME call.
6721 2017-11-19 Tom de Vries <tom@codesourcery.com>
6724 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6725 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6727 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6729 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6730 (cgraph_edge::clone): Cleanup updating of profile.
6731 * ipa-cp.c (update_profiling_info): Likewise.
6732 * ipa-inline-transform.c (inline_transform): Likewise.
6733 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6734 * ipa-split.c (execute_split_functions): Do not split when function
6736 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6737 * profile-count.c (profile_count::dump): Add global0.
6738 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6740 (profile_count::to_sreal_scale): Likewise.
6741 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6742 (profile_count::combine_with_ipa_count): New function.
6743 * profile-count.h (profile_guessed_global0adjusted): New.
6744 (profile_count::adjusted_zero): New.
6745 (profile_count::global0adjusted): New.
6746 (profile_count::combine_with_ipa_count): New.
6747 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6748 correct profile of return block of split functions.
6749 (copy_cfg_body): Remove unused profile_count.
6750 (copy_body): Likewise.
6751 (expand_call_inline): Update.
6752 (tree_function_versioning): Update.
6754 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6756 * hash-set.h (hash_set::empty): New.
6757 * tree-ssa-threadbackward.h: Delete.
6758 * tree-ssa-threadbackward.c (class thread_jumps): New.
6759 Move max_threaded_paths into class.
6760 (fsm_find_thread_path): Remove arguments that are now in class.
6761 (profitable_jump_thread_path): Rename to...
6762 (thread_jumps::profitable_jump_thread_path): ...this.
6763 (convert_and_register_jump_thread_path): Rename to...
6764 (thread_jumps::convert_and_register_current_path): ...this.
6765 (check_subpath_and_update_thread_path): Rename to...
6766 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6767 (register_jump_thread_path_if_profitable): Rename to...
6768 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6769 (handle_phi): Rename to...
6770 (thread_jumps::handle_phi): ...this.
6771 (handle_assignment): Rename to...
6772 (thread_jumps::handle_assignment): ...this.
6773 (fsm_find_control_statement_thread_paths): Rename to...
6774 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6775 (find_jump_threads_backwards): Rename to...
6776 (thread_jumps::find_jump_threads_backwards): ...this.
6777 Initialize path local data.
6778 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6779 from within thread_jumps class.
6780 (pass_early_thread_jumps::execute): Same.
6782 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6784 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6786 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6788 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6789 to canonical location.
6791 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6794 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6796 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6798 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6800 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6801 rcrt1.o%s/grcrt1.o%s for -static-pie.
6803 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6805 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6806 ix86_shift_rotate_cost): Break out from ...
6807 (ix86_rtx_costs): ... here.
6808 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6811 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6813 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6814 when entry block was promoted unlikely.
6815 (estimate_bb_frequencies): Increase frequency scale.
6816 * profile-count.h (profile_count): Export precision info.
6818 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6820 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6821 disturbin profile of entry block.
6823 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6825 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6828 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6830 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6831 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6833 2017-11-17 Jeff Law <law@redhat.com>
6835 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6836 from evrp_dom_walker class. Various methods moved into new class.
6837 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6838 (evrp_range_analyzer::enter): New method.
6839 (evrp_range_analyzer::leave): New method.
6840 (evrp_dom_walker): Remove delegators no longer needed by this class.
6841 Replace vr_values data member with evrp_range_analyzer
6843 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6844 method extracted from evrp_dom_walker::before_dom_children.
6845 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6846 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6848 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6849 Add private copy constructor and move assignment operators.
6850 Privatize methods and class data where trivially possible.
6851 (evrp_dom_walker::cleanup): New function, extracted from
6852 execute_early_vrp. Simplify access to class data.
6854 * vr-values.h (get_output_for_vrp): Prototype.
6855 * vr-values.c (get_output_for_vrp): New function extracted from
6856 vrp_visit_assignment_or_call and extract_range_from_stmt.
6857 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6859 2017-11-17 Luis Machado <luis.machado@linaro.org>
6861 * config/aarch64/aarch64.c
6862 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6863 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6864 tune_params::AUTOPREFETCHER_WEAK.
6866 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6869 * config/arm/arm.c (arm_valid_target_attribute_rec):
6870 Parse "arch=" and "+<ext>".
6871 (arm_valid_target_attribute_tree): Re-init global options.
6872 (arm_option_override): Make non-static.
6873 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6874 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6875 (__ARM_FEATURE_CRC32): Support undef.
6876 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6877 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6879 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6881 * gdbinit.in (break-on-diagnostic): New command.
6883 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6885 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6886 return to indirect jump for EH return if control-flow
6887 protection is enabled. Change explicit 'false' argument in
6888 pro_epilogue_adjust_stack with a value of
6890 * config/i386/i386.md (simple_return_indirect_internal):
6891 Remove SImode restriction to support 64-bit.
6893 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6895 * combine.c (added_notes_insn): New.
6896 (try_combine): Handle added_notes_insn like added_links_insn.
6897 Rewrite return value code.
6898 (distribute_notes): Set added_notes_insn to the earliest insn we added
6901 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6903 PR rtl-optimization/82621
6904 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6905 dest of one of those SETs is unused.
6907 2017-11-17 Richard Biener <rguenther@suse.de>
6910 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6911 * tree-pretty-print.c (dump_generic_node): Handle
6912 annot_expr_parallel_kind.
6913 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6914 * gimplify.c (gimple_boolify): Likewise.
6916 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6918 * config.gcc (extra_headers): Add cet.h for x86 targets.
6919 * config/i386/cet.h: New file.
6920 * doc/install.texi: Add --enable-cet/--disable-cet.
6922 2017-11-17 Richard Biener <rguenther@suse.de>
6924 PR tree-optimization/83017
6925 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6926 (gen_parallel_loop): Properly count iterations.
6927 (parallelize_loops): Handle loop->can_be_parallel independent
6928 of flag_loop_parallelize_all. Make static profitability test match
6930 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6931 * invoke.texi (parloops-min-per-thread): Document.
6933 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6935 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6936 upstreaming review comments.
6938 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6940 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6941 with fast unaligned access.
6942 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6944 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6946 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6947 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6949 2017-11-17 Richard Biener <rguenther@suse.de>
6951 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6952 folding of references.
6954 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6957 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6958 of replacing call to strncmp with corresponding call to strcmp when
6961 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6963 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6964 option prefer-avx256 for skylake-avx512 configuration.
6965 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6966 (get_builtin_code_for_version): Ditto.
6968 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6969 Monk Chiang <sh.chiang04@gmail.com>
6971 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6972 (FIXED_REGISTERS): Reserve more register numbers.
6973 (CALL_USED_REGISTERS): Likewise.
6974 (REG_ALLOC_ORDER): Likewise.
6975 (REG_CLASS_CONTENTS): Likewise.
6976 (REGISTER_NAMES): Likewise.
6978 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6979 Kito Cheng <kito.cheng@gmail.com>
6981 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6983 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6985 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6987 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6988 parameter list for vec_splats.
6990 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6992 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6994 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6996 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6999 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7000 with the required version of automake 1.11.6.
7002 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7004 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7006 (DU_C2_3_power9): Correct reservation combinations.
7007 (FP_DIV_power9, VEC_DIV_power9): New.
7008 (power9-alu): Split out rotate/shift...
7009 (power9-rot): ...to here, correct dispatch resource.
7010 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7012 (power9-fp): Correct latency.
7013 (power9-sdiv): Add div/sqrt resource.
7014 (power9-ddiv): Correct latency, add div/sqrt resource.
7015 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7016 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7018 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7020 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7022 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7023 switch statement mapping KF built-ins to TF built-ins if we don't
7024 have the proper ISA 3.0 assembler support.
7026 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7028 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7029 (lower_emutls_function_body): Do not compute it.
7031 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7033 * ipa-split.c (split_bb_info): Turn time to sreal.
7034 (split_point): Likewise.
7035 (dump_split_point): Likewise.
7036 (fine_split_points): Likewise.
7037 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7039 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7041 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7044 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7046 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7048 (expensive_function_p): Remove useless assert.
7049 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7051 2017-11-16 Martin Sebor <msebor@redhat.com>
7053 PR tree-optimization/82588
7054 PR tree-optimization/82583
7055 * tree-vrp.c (check_array_ref): Handle flexible array members,
7056 string literals, and inner indices.
7057 (search_for_addr_array): Add detail to diagnostics.
7059 2017-11-16 Nathan Sidwell <nathan@acm.org>
7063 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7064 (SET_DECL_ASSEMBLER_NAME): Forward to
7065 overwrite_decl_assembler_name.
7066 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7067 (overwrite_decl_assembler_name): Declare.
7068 * tree.c (overwrite_decl_assembler_name): New.
7069 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7070 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7071 (LANG_HOOKS_INITIALIZER): Add it.
7072 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7073 * langhooks.c (lhd_set_decl_assembler_name): Use
7074 SET_DECL_ASSEMBLER_NAME.
7075 (lhd_overwrite_decl_assembler_name): Default implementation.
7077 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7078 Jackson Woodruff <jackson.woodruff@arm.com>
7080 PR tree-optimization/71026
7081 * match.pd: Canonicalize constant multiplies in division.
7083 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7085 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7086 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7087 Remove OPT_fomit_frame_pointer entry.
7088 * common/config/arc/arc-common.c: Likewise.
7089 * common/config/arm/arm-common.c: Likewise.
7090 * common/config/avr/avr-common.c: Likewise.
7091 * common/config/c6x/c6x-common.c: Likewise.
7092 * common/config/cr16/cr16-common.c: Likewise.
7093 * common/config/cris/cris-common.c: Likewise.
7094 * common/config/epiphany/epiphany-common.c: Likewise.
7095 * common/config/fr30/fr30-common.c: Likewise.
7096 * common/config/frv/frv-common.c: Likewise.
7097 * common/config/ia64/ia64-common.c: Likewise.
7098 * common/config/iq2000/iq2000-common.c: Likewise.
7099 * common/config/lm32/lm32-common.c: Likewise.
7100 * common/config/m32r/m32r-common.c: Likewise.
7101 * common/config/mcore/mcore-common.c: Likewise.
7102 * common/config/microblaze/microblaze-common.c: Likewise.
7103 * common/config/mips/mips-common.c: Likewise.
7104 * common/config/mmix/mmix-common.c: Likewise.
7105 * common/config/mn10300/mn10300-common.c: Likewise.
7106 * common/config/nios2/nios2-common.c: Likewise.
7107 * common/config/pa/pa-common.c: Likewise.
7108 * common/config/pdp11/pdp11-common.c: Likewise.
7109 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7110 * common/config/riscv/riscv-common.c: Likewise.
7111 * common/config/rs6000/rs6000-common.c: Likewise.
7112 * common/config/rx/rx-common.c: Likewise.
7113 * common/config/s390/s390-common.c: Likewise.
7114 * common/config/sh/sh-common.c: Likewise.
7115 * common/config/sparc/sparc-common.c: Likewise.
7116 * common/config/tilegx/tilegx-common.c: Likewise.
7117 * common/config/tilepro/tilepro-common.c: Likewise.
7118 * common/config/v850/v850-common.c: Likewise.
7119 * common/config/visium/visium-common.c: Likewise.
7120 * common/config/xstormy16/xstormy16-common.c: Likewise.
7121 * common/config/xtensa/xtensa-common.c: Likewise.
7122 * invoke.texi (-fomit-frame-pointer): Update documentation.
7124 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7126 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7128 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7130 * cfg.c (scale_bbs_frequencies_int,
7131 cale_bbs_frequencies_gcov_type): Remove.
7132 * cfg.h (scale_bbs_frequencies_int,
7133 cale_bbs_frequencies_gcov_type): Remove.
7135 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7137 * tree-ssa-loop-manip.c
7138 (scale_dominated_blocks_in_loop): Update to profile counts.
7139 (tree_transform_and_unroll_loop): Likewise.
7141 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7143 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7144 scale_bbs_frequencies_int.
7146 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7148 * final.c (compute_alignments): Use counts rather than frequencies.
7150 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7152 * cfgloopanal.c: Include sreal.h
7153 (average_num_loop_insns): Use counts and sreal for accounting.
7155 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7157 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7160 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7161 Kito Cheng <kito.cheng@gmail.com>
7163 * config/nds32/constraints.md: Provide more constraints.
7164 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7165 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7166 support constraints usage.
7168 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7170 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7172 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7173 Kito Cheng <kito.cheng@gmail.com>
7175 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7176 * config/nds32/nds32.opt: Refine the layout.
7177 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7178 TARGET_EXT_STRING): Support new options.
7179 * config/nds32/nds32.h: Likewise.
7180 * config/nds32/nds32.md: Likewise.
7181 * config/nds32/nds32-predicates.c: Likewise.
7182 * config/nds32/constraints.md: Likewise.
7183 * common/config/nds32/nds32-common.c: Likewise.
7185 2017-11-16 Julia Koval <julia.koval@intel.com>
7188 * config/i386/gfniintrin.h: Add sse check.
7189 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7191 2017-11-16 Julia Koval <julia.koval@intel.com>
7193 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7194 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7195 (ix86_handle_option): Handle -mavx512vbmi2.
7196 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7197 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7198 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7199 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7200 (ix86_valid_target_attribute_inner_p): Ditto.
7201 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7202 * config/i386/i386.opt (mavx512vbmi2): New option.
7203 * doc/invoke.texi: Add new option.
7205 2017-11-16 Julia Koval <julia.koval@intel.com>
7207 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7208 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7209 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7210 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7211 _mm512_gf2p8mul_epi8): New intrinsics.
7212 * config/i386/i386-builtin-types.def
7213 (V64QI_FTYPE_V64QI_V64QI): New type.
7214 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7215 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7216 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7217 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7218 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7219 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7221 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7223 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7224 explicitly as a stream of bytes.
7226 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7228 * config/rs6000/altivec.h (vec_xst_be): New #define.
7229 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7230 and externalize from *altivec_vperm_<mode>_internal.
7231 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7233 (XL_BE_V8HI): Likewise.
7234 (XL_BE_V4SI): Likewise.
7235 (XL_BE_V4SI): Likewise.
7236 (XL_BE_V2DI): Likewise.
7237 (XL_BE_V4SF): Likewise.
7238 (XL_BE_V2DF): Likewise.
7239 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7240 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7241 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7242 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7243 VSX_BUILTIN_VEC_XST_BE.
7244 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7245 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7247 (altivec_init_builtins): Replace conditional calls to def_builtin
7248 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7249 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7250 with unconditional calls. Remove calls to def_builtin for
7251 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7252 __builtin_vec_xst_be.
7253 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7254 to define_expand, and add alternate RTL generation for P8.
7255 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7256 vsx_ld_elemrev_v8hi.
7257 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7258 add alternate RTL generation for P8.
7259 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7260 vsx_ld_elemrev_v16qi.
7261 (vsx_st_elemrev_v8hi): Convert define_insn
7262 to define_expand, and add alternate RTL generation for P8.
7263 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7264 vsx_st_elemrev_v8hi.
7265 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7266 add alternate RTL generation for P8.
7267 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7268 vsx_st_elemrev_v16qi.
7270 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7273 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7274 TARGET_AVX512ER check.
7275 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7276 neither -mzeroupper nor -mno-zeroupper is used and
7277 TARGET_EMIT_VZEROUPPER is set.
7278 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7279 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7281 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7284 folding of vector compares.
7285 (fold_build_vec_cmp): New helper function.
7286 (fold_compare_helper): New helper function.
7287 (builtin_function_type): Add compare builtins to the list of functions
7288 having unsigned arguments. Cosmetic updates to comment indentation.
7289 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7290 the not+eq combination.
7292 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7294 PR tree-optimization/82726
7295 PR tree-optimization/70754
7296 * tree-predcom.c (order_drefs_by_pos): New function.
7297 (combine_chains): Move code setting has_max_use_after to...
7298 (try_combine_chains): ...here. New parameter. Sort combined chains
7299 according to position information.
7300 (tree_predictive_commoning_loop): Update call to above function.
7301 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7303 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7305 PR tree-optimization/82726
7307 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7309 PR tree-optimization/70754
7310 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7311 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7312 combined stmt before it if not NULL.
7313 (combine_chains): Process refs reversely and compute dominance point
7317 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7319 PR tree-optimization/79663
7320 * tree-predcom.c (combine_chains): Process refs in reverse order
7321 only for ZERO length chains, and add explaining comment.
7323 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7325 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7326 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7327 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7329 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7331 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7333 2017-11-15 Martin Liska <mliska@suse.cz>
7335 * tree-cfg.c (pass_warn_function_return::execute):
7336 Compare warn_return_type for greater than zero.
7338 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7342 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7343 to return true on Xeon and not on Xeon Phi.
7344 (ix86_check_avx256_register): Changed to ...
7345 (ix86_check_avx_upper_register): ... this. Add extra check for
7346 VALID_AVX512F_REG_OR_XI_MODE.
7347 (ix86_avx_u128_mode_needed): Changed
7348 ix86_check_avx256_register to ix86_check_avx_upper_register.
7349 (ix86_check_avx256_stores): Changed to ...
7350 (ix86_check_avx_upper_stores): ... this. Changed
7351 ix86_check_avx256_register to ix86_check_avx_upper_register.
7352 (ix86_avx_u128_mode_after): Changed
7353 avx_reg256_found to avx_upper_reg_found. Changed
7354 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7355 (ix86_avx_u128_mode_entry): Changed
7356 ix86_check_avx256_register to ix86_check_avx_upper_register.
7357 (ix86_avx_u128_mode_exit): Ditto.
7358 * config/i386/i386.h: (host_detect_local_cpu): New define.
7360 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7362 * config/arm/xgene1.md (xgene1): Split into automatons
7363 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7364 (xgene1_f_load): Adjust reservations and/or types.
7365 (xgene1_f_store): Likewise.
7366 (xgene1_load_pair): Likewise.
7367 (xgene1_store_pair): Likewise.
7368 (xgene1_fp_load1): Likewise.
7369 (xgene1_load1): Likewise.
7370 (xgene1_store1): Likewise.
7371 (xgene1_move): Likewise.
7372 (xgene1_alu): Likewise.
7373 (xgene1_simd): Likewise.
7374 (xgene1_bfm): Likewise.
7375 (xgene1_neon_load1): Likewise.
7376 (xgene1_neon_store1): Likewise.
7377 (xgene1_neon_logic): Likewise.
7378 (xgene1_neon_st1): Likewise.
7379 (xgene1_neon_ld1r): Likewise.
7380 (xgene1_alu_cond): Added.
7381 (xgene1_shift_reg): Likwise.
7382 (xgene1_bfx): Likewise.
7383 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7385 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7388 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7390 (can_widen_mult_without_libcall): New function.
7391 (expand_mul_overflow): If only checking unsigned mul overflow,
7392 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7393 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7394 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7395 (expand_DIVMOD): Formatting fix.
7396 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7397 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7398 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7400 PR tree-optimization/82977
7401 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7402 constructed temporary to strlen_to_stridx.put.
7404 2017-11-15 Martin Liska <mliska@suse.cz>
7406 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7407 * configure: Regenerate.
7409 2017-11-15 Martin Liska <mliska@suse.cz>
7412 * config/sh/sh-mem.cc: Use proper probability for
7415 2017-11-14 Jeff Law <law@redhat.com>
7417 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7418 the red zone for stack_clash_protection_final_dynamic_probe targets
7419 when the total dynamic stack size is zero bytes.
7421 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7422 blocks is post order.
7424 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7426 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7427 * final.c (rest_of_clean_state): Set it for the
7428 -fcompare-debug dump.
7429 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7430 class when TDF_COMPARE_DEBUG is set.
7432 * dwarf2out.c (gen_producer_string): Discard
7435 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7438 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7439 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7440 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7441 (__TGMATH_REAL_2_3): Remove macros.
7442 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7443 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7445 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7447 (remquo): Define using __TGMATH_REAL_3.
7449 2017-11-14 Jeff Law <law@redhat.com>
7451 * vr-values.c: New file with contents extracted from tree-vrp.c.
7452 * Makefile.in (OBJS): Add vr-values.o
7453 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7454 (set_value_range, set_and_canonicalize_value_range): Likewise.
7455 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7456 (value_range_constant_singleton, symbolic_range_p): Likewise.
7457 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7458 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7459 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7460 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7461 (find_case_label_range, find_case_label_index): Likewise.
7462 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7463 (range_int_cst_singleton_p, value_inside_range): Likewise.
7464 (get_single_symbol): Likewise.
7465 (switch_update): Move structure definition here.
7466 (to_remove_edges, to_update_switch_stmts): Provide externs.
7467 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7468 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7469 (vrp_val_is_min, set_value_range): Likewise.
7470 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7471 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7472 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7473 (range_is_nonnull, range_int_cst_p): Likewwise.
7474 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7475 (get_single_symbol, operand_less_p): Likewise
7476 (compare_values_warnv, compare_values): Likewise.
7477 (value_inside_range, value_range_constant_singleton): Likewise.
7478 (zero_nonzero_bitgs_from_vr): Likewise.
7479 (extract_range_from_binary_expr_1): Likewise.
7480 (overflow_comparison_p): Likewise.
7481 (to_remove_edges, to_update_switch_stmts): Likewise.
7482 (find_case_label-index, find_case_label_range): Likewise.
7483 (switch_update, set_value_range_to_nonnegative): Remove.
7484 (set_value_range_to_truthvalue): Likewise.
7485 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7486 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7487 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7488 (find_case_label_ranges, test_for_singularity): Likewise.
7489 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7490 (x_vr_values): Move to its remaining use site.
7492 2017-11-10 Jeff Law <law@redhat.com>
7494 * vr-values.h (VR_INITIALIZER): Move #define here.
7495 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7496 * Makefile.in (OBJS): Add tree-evrp.o
7497 * tree-vrp.h (assert_info): Move structure definition here.
7498 (set_value_range_to_varying): Prototype.
7499 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7500 (infer_value_range, register_edge_assert_for): Likewise.
7501 (stmt_interesting_for_vrp): Likewise.
7502 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7503 (set_value_range_to_varying): No longer static.
7504 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7505 (infer_value_range, register_edge_assert_for): Likewise.
7507 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7509 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7510 generate the XXBRD instruction.
7512 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7513 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7514 as being compatible if -mabi=ieeelongdouble.
7515 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7516 to setup float128 built-ins with hardware support.
7517 (BU_FLOAT128_HW_2): Likewise.
7518 (BU_FLOAT128_HW_3): Likewise.
7519 (BU_FLOAT128_HW_VSX_1): Likewise.
7520 (BU_FLOAT128_HW_VSX_2): Likewise.
7521 (scalar_extract_expq): Change float128 built-in functions to
7522 accommodate having both KFmode and TFmode functions. Use the
7523 KFmode variant as the default.
7524 (scalar_extract_sigq): Likewise.
7525 (scalar_test_neg_qp): Likewise.
7526 (scalar_insert_exp_q): Likewise.
7527 (scalar_insert_exp_qp): Likewise.
7528 (scalar_test_data_class_qp): Likewise.
7529 (sqrtf128_round_to_odd): Delete processing the round to odd
7530 built-in functions as special built-in functions, and define them
7531 as float128 built-ins. Use the KFmode variant as the default.
7532 (truncf128_round_to_odd): Likewise.
7533 (addf128_round_to_odd): Likewise.
7534 (subf128_round_to_odd): Likewise.
7535 (mulf128_round_to_odd): Likewise.
7536 (divf128_round_to_odd): Likewise.
7537 (fmaf128_round_to_odd): Likewise.
7538 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7539 support for KFmode and TFmode xststdcqp calls.
7540 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7541 point, switch the built-in handlers for the get/set float128
7542 exponent, get float128 mantissa, float128 test built-ins, and the
7543 float128 round to odd built-in functions. Eliminate creating the
7544 float128 round to odd built-in functions as special built-ins.
7545 (rs6000_init_builtins): Eliminate special creation of the float128
7546 round to odd built-in functions.
7547 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7548 function insns to support both TFmode and KFmode variants.
7549 (xsxsigqp_<mode>): Likewise.
7550 (xsiexpqpf_<mode>): Likewise.
7551 (xsiexpqp_<mode>): Likewise.
7552 (xststdcqp_<mode>): Likewise.
7553 (xststdcnegqp_<mode>): Likewise.
7554 (xststdcqp_<mode>): Likewise.
7556 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7558 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7559 unused path_in_freq_ptr parameter.
7560 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7562 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7564 * ipa-inline.c (edge_badness): Dump sreal frequency.
7565 (compute_inlined_call_time): Match natural implementaiton ...
7566 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7567 forgotten division by CGRAPH_FREQ_BASE.
7569 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7571 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7572 Solaris 11. Update comment.
7573 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7575 * config/sol2.h (STARTFILE_SPEC): Likewise.
7576 * configure: Regenerate.
7578 2017-11-14 Carl Love <cel@us.ibm.com>
7580 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7581 le_ and be_ prefixes to swap* variables. Remove
7582 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7585 2017-11-14 Jason Merrill <jason@redhat.com>
7587 Support GTY((cache)) on hash_map.
7588 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7589 (ggc_cache_remove): Override it instead of ggc_mx.
7590 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7591 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7592 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7593 (simple_cache_map_traits): Override maybe_mx.
7594 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7595 (hash_map): Friend gt_cleare_cache.
7596 (gt_cleare_cache): New.
7597 * tree.h (tree_cache_traits): New hash_map traits class.
7598 (tree_cache_map): New typedef.
7600 2017-11-14 Richard Biener <rguenther@suse.de>
7602 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7603 paramter and handling.
7604 (cleanup_control_flow_bb): Likewise.
7605 (cleanup_control_flow_pre): New helper performing a DFS walk
7606 to call cleanup_control_flow_bb in PRE order.
7607 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7608 via cleanup_control_flow_pre.
7610 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7612 * config/aarch64/aarch64-simd.md
7613 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7614 (*aarch64_simd_bsl<mode>_alt): Likewise.
7615 (aarch64_simd_bsldi_internal): New.
7616 (aarch64_simd_bsldi_alt): Likewise.
7618 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7620 * tracer.c (better_p): Do not compare frequencies.
7621 * reg-stack.c (better_edge): Likewise.
7622 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7625 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7627 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7628 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7629 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7630 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7631 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7632 * predict.c (maybe_hot_frequency_p): Inline to ...
7633 (maybe_hot_count_p): ... here; rewrite to counts.
7634 (counts_to_freqs): Rename to ...
7635 (update_max_bb_count): ... this one.
7636 (expensive_function_p): Use counts.
7637 (estimate_bb_frequencies): Update.
7638 (rebuild_frequencies): Update.
7639 * predict.h (counts_to_freqs): Rename to ...
7640 (update_max_bb_count): ... this one.
7641 * profile.c (compute_branch_probabilities): Add debug info
7642 * tree-inline.c (expand_call_inline): Update debug info.
7643 (optimize_inline_calls): Use update_max_bb_count..
7644 (tree_function_versioning): Use update_max_bb_count..
7645 * value-prof.c (gimple_value_profile_transformations):
7646 Do not use update_max_bb_count.
7648 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7650 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7651 always use frequencies.
7653 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7655 * bb-reorder.c: Remove frequencies from comments.
7656 (better_edge_p): Use profile counts.
7657 (find_traces): Dump profile counts.
7658 (rotate_loop): Use profile counts.
7659 (find_traces_1_round): Likewise.
7660 (connect_better_edge_p): Use counts instead of probabilities for
7662 (copy_bb_p): Drop early check for non-0 frequency.
7663 (sanitize_hot_paths): Update comments.
7665 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7667 * ipa-split.c (struct split_point): Add count.
7668 (consider_split): Do not compute incoming frequency; compute incoming
7669 count and store it to split_point.
7670 (split_function): Set count of the call to split part correctly.
7672 2017-11-13 Carl Love <cel@us.ibm.com>
7674 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7676 2017-11-13 Tom Tromey <tom@tromey.com>
7678 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7680 2017-11-13 Carl Love <cel@us.ibm.com>
7682 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7683 Add support for builtins:
7684 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7685 vector {un,}signed {char,int,short},
7686 vector {un,}signed {char,int,short}) arguments.
7687 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7688 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7689 Add BU_P9V_AV_2 expansions for the builtins.
7690 * config/rs6000/altivec.h (vec_first_match_index,
7691 vec_first_mismatch_index, vec_first_match_or_eos_index,
7692 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7693 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7694 new extern declaration.
7695 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7696 * config/rs6000/vsx.md (first_match_index_<mode>,
7697 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7698 first_mismatch_or_eos_index_<mode>): Add define expand.
7699 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7700 * doc/extend.texi: Update the built-in documenation file for the new
7703 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7705 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7706 into the min/max operations for _Float<N> and _Float<N>X types.
7708 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7711 * dwarf2out.c (do_eh_frame): New static variable.
7712 (dwarf2out_begin_prologue): Set it.
7713 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7715 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7717 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7719 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7721 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7724 * debug.h (dwarf2out_do_eh_frame): Declare.
7725 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7726 (dwarf2out_do_frame): Use it.
7727 (dwarf2out_do_cfi_asm): Likewise.
7728 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7729 (dwarf2out_assembly_start): Likewise.
7730 (dwarf2out_begin_prologue): Fix comment.
7731 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7732 if the target needs either debug or unwind DWARF2 info.
7733 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7734 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7735 -ftrapping-math, -ftrapv and -fwrapv.
7737 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7739 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7740 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7741 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7742 (estimate_edge_size_and_time): Likewise.
7743 (ipa_merge_fn_summary_after_inlining): Likewise.
7744 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7745 (compute_uninlined_call_time): Use sreal_frequency.
7746 (compute_inlined_call_time): Likewise.
7747 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7748 * profile-count.c: Include sreal.h.
7749 (profile_count::to_sreal_scale): New.
7750 * profile-count.h: Forward declare sreal.
7751 (profile_count::to_sreal_scale): Declare.
7753 2017-11-13 Nathan Sidwell <nathan@acm.org>
7755 * diagnostic.c (maybe_line_and_column): New.
7756 (diagnostic_get_location_text): Use it.
7757 (diagnostic_report_current_module): Likewise.
7758 (test_diagnostic_get_location_text): Add tests.
7760 2017-11-13 Luis Machado <luis.machado@linaro.org>
7762 * doc/md.texi (Specifying processor pipeline description): Fix
7763 incorrect latency for the div instruction example.
7765 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7767 PR tree-optimization/78821
7768 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7769 that bit_not_p is the same.
7770 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7771 (split_group): Count precisely bit_not_p bits in each statement.
7772 (invert_op): New function.
7773 (imm_store_chain_info::output_merged_store): Use invert_op to
7774 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7775 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7777 2017-11-13 Martin Liska <mliska@suse.cz>
7779 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7780 (struct source_info): Likewise.
7781 (add_branch_counts): Likewise.
7782 (add_line_counts): Likewise.
7783 (function_summary): Likewise.
7784 (output_intermediate_line): Likewise.
7785 (generate_results): Likewise.
7787 2017-11-13 Martin Liska <mliska@suse.cz>
7789 * gcov.c (struct block_info): Remove typedef for block_t.
7790 (struct line_info): Likewise.
7791 (line_info::has_block): Likewise.
7792 (EXIT_BLOCK): Likewise.
7793 (unblock): Likewise.
7794 (circuit): Likewise.
7795 (get_cycles_count): Likewise.
7796 (process_file): Likewise.
7797 (read_graph_file): Likewise.
7798 (solve_flow_graph): Likewise.
7799 (find_exception_blocks): Likewise.
7800 (add_line_counts): Likewise.
7801 (accumulate_line_info): Likewise.
7802 (output_line_details): Likewise.
7804 2017-11-13 Martin Liska <mliska@suse.cz>
7806 * gcov.c (struct arc_info): Remove typedef for arc_t.
7807 (struct line_info): Likewise.
7808 (add_branch_counts): Likewise.
7809 (output_branch_count): Likewise.
7810 (function_info::~function_info): Likewise.
7811 (circuit): Likewise.
7812 (output_intermediate_line): Likewise.
7813 (read_graph_file): Likewise.
7814 (solve_flow_graph): Likewise.
7815 (find_exception_blocks): Likewise.
7816 (add_line_counts): Likewise.
7817 (accumulate_line_info): Likewise.
7818 (output_line_details): Likewise.
7819 (output_function_details): Likewise.
7821 2017-11-13 Martin Liska <mliska@suse.cz>
7823 * gcov.c (struct function_info): Remove typedef for function_t.
7824 (struct source_info): Likewise.
7825 (source_info::get_functions_at_location): Likewise.
7826 (solve_flow_graph): Likewise.
7827 (find_exception_blocks): Likewise.
7828 (add_line_counts): Likewise.
7829 (output_intermediate_file): Likewise.
7830 (process_file): Likewise.
7831 (generate_results): Likewise.
7832 (release_structures): Likewise.
7833 (read_graph_file): Likewise.
7834 (read_count_file): Likewise.
7835 (accumulate_line_counts): Likewise.
7836 (output_lines): Likewise.
7838 2017-11-13 Martin Liska <mliska@suse.cz>
7840 * gcov.c (function_info::function_info): Remove num_counts
7841 and add vector<gcov_type>.
7842 (function_info::~function_info): Use the vector.
7843 (process_file): Likewise.
7844 (read_graph_file): Likewise.
7845 (read_count_file): Likewise.
7846 (solve_flow_graph): Likewise.
7848 2017-11-13 Martin Liska <mliska@suse.cz>
7850 * gcov.c (function_info::is_artificial): New function.
7851 (process_file): Erase all artificial early.
7852 (generate_results): Skip as all artificial are already
7855 2017-11-13 Martin Liska <mliska@suse.cz>
7857 * gcov.c (read_graph_file): Store to global vector of functions.
7858 (read_count_file): Iterate the vector.
7859 (process_file): Likewise.
7860 (generate_results): Likewise.
7861 (release_structures): Likewise.
7863 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7865 PR tree-optimization/82954
7866 * gimple-ssa-store-merging.c
7867 (imm_store_chain_info::coalesce_immediate_stores): If
7868 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7870 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7872 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7873 Upddate call to ENDIAN_LANE_N.
7874 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7875 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7876 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7877 and use aarch64_endian_lane_rtx.
7878 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7880 2017-11-12 Tom de Vries <tom@codesourcery.com>
7882 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7884 2017-11-12 Tom de Vries <tom@codesourcery.com>
7886 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7888 2017-11-12 Tom de Vries <tom@codesourcery.com>
7890 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7892 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7893 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7894 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7896 2017-11-11 Martin Sebor <msebor@redhat.com>
7899 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7902 * prefic.c (translate_name): Replace strncpy with memcpy to
7903 avoid -Wstringop-truncation.
7905 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7907 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7909 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7911 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7913 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7915 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7918 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7920 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7921 enable generating XXBRH if the value is in a vector register.
7922 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7923 value is in a vector register.
7924 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7925 register to register bswap64's instead of doing the GPR sequence
7926 used on previous machines.
7927 (bswapdi2_xxbrd): New insn.
7928 (bswapdi2_reg): Disallow on ISA 3.0.
7929 (register to register bswap64 splitter): Do not split the insn on
7930 ISA 3.0 systems that use XXBRD.
7932 2017-11-10 Martin Sebor <msebor@redhat.com>
7935 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7937 (find_subframework_file): Same.
7939 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7941 * auto-profile.c (afdo_indirect_call): Drop frequency.
7942 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7943 (cgraph_node::create_edge): Drop frequency argument.
7944 (cgraph_node::create_indirect_edge): Drop frequency argument.
7945 (cgraph_edge::make_speculative): Drop frequency arguments.
7946 (cgraph_edge::resolve_speculation): Do not update frequencies
7947 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7948 (cgraph_node::dump): Check consistency in IPA mode.
7949 (cgraph_edge::maybe_hot_p): Use IPA counter.
7950 (cgraph_edge::verify_count_and_frequency): Rename to ...
7951 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7952 (cgraph_node::verify_node): Update.
7953 * cgraph.h (struct cgraph_edge): Drop frequency.
7954 (cgraph_edge::frequency): New function.
7955 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7957 (cgraph_edge::rebuild_edges): Likewise.
7958 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7959 (duplicate_thunk_for_node): Do not pass frequency.
7960 (cgraph_node::create_clone): Scale only counts.
7961 (cgraph_node::create_virtual_clone): Do not pass frequency.
7962 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7963 (cgraph_node::create_version_clone): Do not pass frequency.
7964 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7965 (cgraph_node::expand_thunk): Do not pass frequency.
7966 (cgraph_node::create_wrapper): Do not pass frequency.
7967 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7969 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7970 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7971 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7972 (gather_caller_stats): Use frequency function.
7973 (ipcp_cloning_candidate_p): Use frequency function.
7974 (ipcp_propagate_stage): Use frequency function.
7975 (get_info_about_necessary_edges): Use frequency function.
7976 (update_profiling_info): Update only IPA profile.
7977 (update_specialized_profile): Use frequency functoin.
7978 (perhaps_add_new_callers): Update only IPA profile.
7979 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7980 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7981 (dump_ipa_call_summary): Use frequency function.
7982 (estimate_edge_size_and_time): Use frequency function.
7983 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7984 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7985 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7986 (update_noncloned_counts): ... ths one; scale counts only.
7987 (clone_inlined_nodes): Do not scale frequency.
7988 (inline_call): Do not pass frequency.
7989 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7990 (compute_inlined_call_time): Use IPA profile.
7991 (want_inline_small_function_p): Use IPA profile.
7992 (want_inline_self_recursive_call_p): Use IPA profile.
7993 (edge_badness): Use IPA profile.
7994 (lookup_recursive_calls): Use IPA profile.
7995 (recursive_inlining): Do not pass frequency.
7996 (resolve_noninline_speculation): Do not update frequency.
7997 (inline_small_functions): Collect max of IPA profile.
7998 (dump_overall_stats): Dump IPA porfile.
7999 (dump_inline_stats): Dump IPA porfile.
8000 (ipa_inline): Collect IPA stats.
8001 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8002 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8003 (ipa_propagate_frequency): Use frequency function.
8004 (ipa_profile): Cleanup.
8005 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8006 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8007 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8008 (input_node): Do not stream frequency.
8009 (input_edge): Do not stream frequency.
8010 (merge_profile_summaries): Scale only IPA profiles.
8011 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8012 * predict.c (drop_profile): Do not recompute frequency.
8013 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8014 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8015 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8016 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8017 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8018 * tree-inline.c (copy_bb): Do not scale frequency.
8019 (expand_call_inline): Do not scale frequency.
8020 (tree_function_versioning): Do not scale frequency.
8021 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8023 2017-11-10 Julia Koval <julia.koval@intel.com>
8025 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8026 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8027 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8028 (_mm256_maskz_gf2p8affine_epi64_epi8)
8029 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8030 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8031 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8032 (__builtin_ia32_vgf2p8affineqb_v32qi)
8033 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8034 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8036 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8040 (arm_option_override): Refactor.
8041 (arm_option_reconfigure_globals): New.
8042 (arm_options_perform_arch_sanity_checks): New.
8043 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8045 (arm_options_perform_arch_sanity_checks): Likewise
8047 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8049 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8050 (power9-qpmul): New.
8051 * rs6000/rs6000.md ("type" attr): Add qmul.
8052 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8053 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8054 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8056 2017-11-10 Martin Sebor <msebor@redhat.com>
8059 * builtins.c (compute_objsize): Handle arrays that
8060 compute_builtin_object_size likes to fail for. Make extern.
8061 * builtins.h (compute_objsize): Declare.
8062 (check_strncpy_sizes): New function.
8063 (expand_builtin_strncpy): Call check_strncpy_sizes.
8064 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8065 -Wstringop-truncation.
8066 (gimple_fold_builtin_strncat): Same.
8067 * gimple.c (gimple_build_call_from_tree): Set call location.
8068 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8069 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8070 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8071 (handle_builtin_strlen): Use strlen_to_stridx.
8072 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8074 Use strlen_to_stridx.
8075 (pass_strlen::execute): Release strlen_to_stridx.
8076 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8077 (-Wstringop-truncation): Document new option.
8079 2017-11-10 Martin Liska <mliska@suse.cz>
8081 PR gcov-profile/82702
8082 * gcov.c (main): Handle intermediate files in a different
8084 (get_gcov_intermediate_filename): New function.
8085 (output_gcov_file): Remove support of intermediate files.
8086 (generate_results): Allocate intermediate file.
8087 (release_structures): Clean-up properly fn_end.
8088 (output_intermediate_file): Start iterating with line 1.
8090 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8092 PR tree-optimization/82929
8093 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8094 ops_swapped_p non-static data member.
8095 (store_immediate_info::store_immediate_info): Clear it.
8096 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8097 ops set ops_swapped_p.
8098 (count_multiple_uses): Handle ops_swapped_p.
8100 2017-11-10 Martin Liska <mliska@suse.cz>
8102 * coverage.c (coverage_init): Stream information about
8103 support of has_unexecuted_blocks.
8104 * doc/gcov.texi: Document that.
8105 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8106 * gcov.c (read_graph_file): Likewise.
8107 (output_line_beginning): Fix a small issue with
8110 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8112 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8113 reference of trivial component.
8115 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8118 * gimple-ssa-store-merging.c
8119 (pass_store_merging::terminate_all_aliasing_chains): For
8120 gimple_store_p stmts also call refs_output_dependent_p.
8122 PR rtl-optimization/82913
8123 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8126 2017-11-09 Jeff Law <law@redhat.com>
8128 * vr-values.h: New file with vr_values class.
8129 * tree-vrp.c: Include vr-values.h
8130 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8131 data objects into the vr_values class.
8132 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8133 (get_value_range): Make it a member function within vr_values class.
8134 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8135 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8136 (op_with_constant_singleton_value_range): Likewise.
8137 (extract_range_for_var_from_comparison_expr): Likewise.
8138 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8139 (extract_range_from_binary_expr): Likewise.
8140 (extract_range_from_unary_expr): Likewise.
8141 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8142 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8143 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8144 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8145 (compare_name_with_value, compare_names): Likewise.
8146 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8147 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8148 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8149 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8150 (extract_range_from_phi_node): Likewise.
8151 (simplify_truth_ops_using_ranges): Likewise.
8152 (simplify_div_or_mod_using_ranges): Likewise.
8153 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8154 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8155 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8156 (simplify_float_conversion_using_ranges): Likewise.
8157 (simplify_internal_call_using_ranges): Likewise.
8158 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8159 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8160 poitner via x_vr_values for calls into gimple folder.
8161 (vrp_initialize_lattice): Make this the vr_values ctor.
8162 (vrp_free_lattice): Make this the vr_values dtor.
8163 (set_vr_value): New function.
8164 (class vrp_prop): Add vr_values data member. Add various member
8165 functions as well as member functions that delegate to vr_values.
8166 (check_array_ref): Make a member function within vrp_prop class.
8167 (search_for_addr_array, vrp_initialize): Likewise.
8168 (vrp_finalize): Likewise. Revamp to avoid direct access to
8169 vr_value, values_propagated, etc.
8170 (check_array_bounds): Extract vrp_prop class instance pointer from
8171 walk info structure. Use it to call member functions.
8172 (check_all_array_refs): Make a member function within vrp_prop class.
8173 Smuggle class instance pointer via walk info structure.
8174 (x_vr_values): New local static.
8175 (vrp_valueize): Use x_vr_values to get class instance.
8176 (vr_valueize_1): Likewise.
8177 (class vrp_folder): Add vr_values data member. Add various member
8178 functions as well as member functions that delegate to vr_values.
8179 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8180 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8181 class instance from vr_values. Use it to call member functions.
8182 (vrp_dom_walker): Add vr_values data member.
8183 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8184 instance via x_vr_values.
8185 (identify_jump_threads): Accept vr_values as argument. Store
8186 it into the walker structure.
8187 (evrp_dom_walker): Add vr_values class data member. Add various
8189 (evrp_dom_walker::try_find_new_range): Use vr_values data
8190 member to access the memory allocator.
8191 (evrp_dom_walker::before_dom_children): Store vr_values class
8192 instance into the vrp_folder class.
8193 (evrp_dom_walker::push_value_range): Rework to avoid direct
8194 access to num_vr_values and vr_value.
8195 (evrp_dom_walker::pop_value_range): Likewise.
8196 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8197 Use vr_values to get to dump_all_value_ranges member function.
8198 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8199 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8200 Pass vr_values class instance down to identify_jump_threads.
8201 Remove call to vrp_free_lattice.
8202 (debug_all_value_ranges): Remove.
8204 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8205 (vrp_folder): Likewise.
8207 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8208 Get it from the existing bitmap instead.
8209 (vrp_intersect_ranges_1): Likewise.
8211 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8213 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8215 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8216 set bit_not_p to it.
8217 (imm_store_chain_info::coalesce_immediate_stores): Break group
8218 if bit_not_p is different.
8219 (count_multiple_uses, split_group,
8220 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8221 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8222 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8223 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8224 to store_immediate_info ctor.
8226 2017-11-09 Jim Wilson <jimw@sifive.com>
8228 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8229 (scan_prog_file): Likewise.
8231 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8233 * bb-reorder.c (max_entry_frequency): Remove.
8234 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8235 connect_traces, push_to_next_round_p): Remove prototypes.
8236 (find_traces_1_round): Use counts only.
8237 (push_to_next_round_p): Likewise.
8238 (find_traces): Likewise.
8239 (rotate_loop): Likewise.
8240 (find_traces_1_round): Likewise.
8241 (connect_traces): Likewise.
8242 (edge_order): Likewise.
8244 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8246 * config/arm/arm.c (output_return_instruction): Add comments to
8247 indicate requirement for cmse_nonsecure_entry return to account
8248 for the size of clearing instruction output here.
8249 (thumb_exit): Likewise.
8250 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8251 return in hardfloat mode.
8253 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8255 * config/rs6000/rs6000.c (machine_function): Add a bool,
8256 "toc_is_wrapped_separately".
8257 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8258 if it wasn't explicitly set or unset, we are optimizing for speed, and
8259 doing separate shrink-wrapping.
8260 (rs6000_get_separate_components): Enable the TOC component if
8261 saving the TOC register in the prologue.
8262 (rs6000_components_for_bb): Handle the TOC component.
8263 (rs6000_emit_prologue_components): Store the TOC register where needed.
8264 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8265 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8267 2017-11-09 Martin Jambor <mjambor@suse.cz>
8269 * ipa-param-manipulation.c: New file.
8270 * ipa-param-manipulation.h: Likewise.
8271 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8272 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8273 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8274 (ipa_parm_adjustment): Likewise.
8275 (ipa_parm_adjustment_vec): Likewise.
8276 (ipa_get_vector_of_formal_parms): Moved declaration to
8277 ipa-param-manipulation.h.
8278 (ipa_get_vector_of_formal_parm_types): Likewise.
8279 (ipa_modify_formal_parameters): Likewise.
8280 (ipa_modify_call_arguments): Likewise.
8281 (ipa_combine_adjustments): Likewise.
8282 (ipa_dump_param_adjustments): Likewise.
8283 (ipa_modify_expr): Likewise.
8284 (ipa_get_adjustment_candidate): Likewise.
8285 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8286 ipa-param-manipulation.c.
8287 (ipa_get_vector_of_formal_parm_types): Likewise.
8288 (ipa_modify_formal_parameters): Likewise.
8289 (ipa_modify_call_arguments): Likewise.
8290 (ipa_modify_expr): Likewise.
8291 (get_ssa_base_param): Likewise.
8292 (ipa_get_adjustment_candidate): Likewise.
8293 (index_in_adjustments_multiple_times_p): Likewise.
8294 (ipa_combine_adjustments): Likewise.
8295 (ipa_dump_param_adjustments): Likewise.
8296 * tree-sra.c: Also include ipa-param-manipulation.h
8297 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8300 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8301 Alan Hayward <alan.hayward@arm.com>
8302 David Sherwood <david.sherwood@arm.com>
8304 * doc/sourcebuild.texi (vect_masked_store): Document.
8306 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8307 Alan Hayward <alan.hayward@arm.com>
8308 David Sherwood <david.sherwood@arm.com>
8310 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8312 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8313 Alan Hayward <alan.hayward@arm.com>
8314 David Sherwood <david.sherwood@arm.com>
8316 * doc/sourcebuild.texi (vect_variable_length): Document.
8318 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8319 Alan Hayward <alan.hayward@arm.com>
8320 David Sherwood <david.sherwood@arm.com>
8322 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
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 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8330 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8335 previously undocumented selectors.
8336 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8338 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8340 * doc/rtl.texi (const_vector): Say that elements can be
8341 const_wide_ints too.
8342 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8343 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8344 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8345 * optabs.c (expand_vector_broadcast): Likewise.
8347 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8348 Alan Hayward <alan.hayward@arm.com>
8349 David Sherwood <david.sherwood@arm.com>
8351 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8352 scaled index even if the unscaled address was invalid.
8353 Don't increase the complexity of using a scale in that case.
8355 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8356 Alan Hayward <alan.hayward@arm.com>
8357 David Sherwood <david.sherwood@arm.com>
8359 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8360 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8361 UNITS_PER_WORD bytes.
8362 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8363 based on REGMODE_NATURAL_SIZE of the inner mode.
8364 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8365 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8366 * expmed.c (lowpart_bit_field_p): Divide the value up into
8367 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8368 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8369 whether something is likely to occupy more than one register.
8371 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8374 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8375 profile_count::adjust_for_ipa_scaling.
8376 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8377 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8379 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8381 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8383 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8384 (split_group): Add total_orig and total_new arguments, estimate the
8385 number of statements related to the store group without store merging
8386 and with store merging.
8387 (imm_store_chain_info::output_merged_store): Adjust split_group
8388 callers, punt if estimated number of statements with store merging
8389 is not smaller than estimated number of statements without it.
8391 (handled_load): Remove has_single_use checks.
8392 (pass_store_merging::process_store): Likewise.
8394 2017-11-09 Richard Biener <rguenther@suse.de>
8396 PR tree-optimization/82902
8397 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8399 2017-11-09 Martin Liska <mliska@suse.cz>
8402 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8405 2017-11-09 Martin Liska <mliska@suse.cz>
8407 PR tree-optimization/82669
8408 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8410 2017-11-09 Martin Liska <mliska@suse.cz>
8412 PR gcov-profile/48463
8413 * coverage.c (coverage_begin_function): Output also end locus
8414 of a function and information whether the function is
8416 * gcov-dump.c (tag_function): Parse and print the information.
8417 * gcov.c (INCLUDE_MAP): Add include.
8418 (INCLUDE_SET): Likewise.
8419 (struct line_info): Move earlier in the source file because
8420 of vector<line_info> in function_info structure.
8421 (line_info::line_info): Likewise.
8422 (line_info::has_block): Likewise.
8423 (struct source_info): Add new member index.
8424 (source_info::get_functions_at_location): New function.
8425 (function_info::group_line_p): New function.
8426 (output_intermediate_line): New function.
8427 (output_intermediate_file): Use the mentioned function.
8428 (struct function_start): New.
8429 (struct function_start_pair_hash): Likewise.
8430 (process_file): Add code that identifies group functions.
8431 Assign lines either to global or function scope.
8432 (generate_results): Skip artificial functions.
8433 (find_source): Assign index for each source file.
8434 (read_graph_file): Read new flag artificial and end_line.
8435 (add_line_counts): Assign it either to global of function scope.
8436 (accumulate_line_counts): Isolate core of the function to
8437 accumulate_line_info and call it for both function and global
8439 (accumulate_line_info): New function.
8440 (output_line_beginning): Fix GNU coding style.
8441 (print_source_line): New function.
8442 (output_line_details): Likewise.
8443 (output_function_details): Likewise.
8444 (output_lines): Iterate both source (global) scope and function
8446 (struct function_line_start_cmp): New class.
8447 * doc/gcov.texi: Reflect changes in documentation.
8449 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8452 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8453 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8454 and similarly for not instead of neg.
8456 2017-11-08 Andi Kleen <ak@linux.intel.com>
8458 * config/i386/i386.opt: Add -mforce-indirect-call.
8459 * config/i386/predicates.md: Check for flag_force_indirect_call.
8460 * doc/invoke.texi: Document -mforce-indirect-call
8462 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8464 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8466 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8467 config/riscv/riscv.c (predict.h): New include.
8468 (riscv_slow_unaligned_access_p): No longer static.
8469 (riscv_block_move_straight): Add require.
8470 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8472 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8475 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8476 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8477 nonimmediate_operand predicate for operand 1 instead of
8480 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8482 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8484 * config/aarch64/constraints.md (Uml): New constraint.
8485 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8488 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8490 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8491 of two vec_duplicates into a vec_concat.
8493 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8495 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8496 Simplify vec_merge of vec_duplicate and vec_concat.
8497 * config/aarch64/constraints.md (Utq): New constraint.
8498 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8501 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8503 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8504 Simplify vec_merge of vec_duplicate and const_vector.
8505 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8507 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8508 mode iterator. Update predicate on operand 1 to
8509 handle non-const_vec constants. Delete constraints.
8510 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8512 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8514 PR tree-optimization/78821
8515 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8517 (store_operand_info::store_operand_info): Initialize it to false.
8518 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8519 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8520 store in the group, and if chain_info is non-NULL, to ignore altogether
8522 (compatible_load_p): Fail if bit_not_p does not match.
8523 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8524 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8525 (pass_store_merging::process_store): Adjust
8526 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8527 call terminate_all_aliasing_chains newly when adding a store into
8528 a chain with non-NULL chain_info.
8530 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8532 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8534 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8536 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8538 (aarch64_layout_frame): Initialise emit_frame_chain.
8539 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8540 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8542 2017-11-08 Martin Liska <mliska@suse.cz>
8544 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8545 of gimple_call_internal_p.
8547 2017-11-07 Tom de Vries <tom@codesourcery.com>
8549 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8551 2017-11-07 Tom de Vries <tom@codesourcery.com>
8553 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8556 2017-11-08 Martin Liska <mliska@suse.cz>
8559 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8561 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8563 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8565 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8568 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8569 (*cmp<mode>_ccz_1): New insn with $k alternative.
8572 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8573 RTX_COMM_COMPARE as commutative as well.
8574 (ix86_binary_operator_ok): Formatting fix.
8575 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8576 *<code><mode>3<mask_name><round_saeonly_name>,
8577 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8578 *<s>mul<mode>3_highpart<mask_name>,
8579 *vec_widen_umult_even_v16si<mask_name>,
8580 *vec_widen_umult_even_v8si<mask_name>,
8581 *vec_widen_umult_even_v4si<mask_name>,
8582 *vec_widen_smult_even_v16si<mask_name>,
8583 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8584 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8585 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8586 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8587 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8588 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8589 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8590 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8591 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8592 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8593 ix86_binary_operator_ok. Formatting fixes.
8594 (*<plusminus_insn><mode>3<mask_name><round_name>,
8595 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8598 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8600 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8601 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8602 eq and ne if TARGET_ISEL.
8603 (cmp): New code_iterator.
8604 (UNS, UNSU_, UNSIK): New code_attrs.
8605 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8606 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8608 ("eq<mode>3"): ... this.
8609 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8611 ("ne<mode>3"): ... this.
8613 2017-11-07 Julia Koval <julia.koval@intel.com>
8616 * common/config/i386/i386-common.c
8617 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8618 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8619 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8620 * config/i386/i386.opt: Ditto.
8621 * config/i386/i386.c (ix86_target_string): Ditto.
8622 (ix86_option_override_internal): Ditto.
8623 (ix86_init_mpx_builtins): Move MPX to args2.
8624 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8625 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8626 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8627 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8628 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8629 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8630 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8632 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8635 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8636 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8637 (zero-extendsidi peephole2): Remove peephole.
8639 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8642 * stor-layout.c: Include attribs.h.
8643 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8644 explicit lookup of "aligned" attribute.
8646 2017-11-07 Andrew Waterman <andrew@sifive.com>
8648 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8649 (riscv_expand_block_move): Likewise.
8650 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8652 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8653 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8654 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8655 (riscv_adjust_block_mem): Likewise.
8656 (riscv_block_move_loop): Likewise.
8657 (riscv_expand_block_move): Likewise.
8658 * config/riscv/riscv.md (movmemsi): New pattern.
8660 2017-11-07 Michael Clark <michaeljclark@mac.com>
8662 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8663 (MUSL_DYNAMIC_LINKER): Likewise.
8665 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8667 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8668 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8671 2017-11-07 Richard Biener <rguenther@suse.de>
8673 * match.pd: Fix build.
8675 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8676 Jackson Woodruff <jackson.woodruff@arm.com>
8678 PR tree-optimization/71026
8679 * match.pd: Canonicalize negate in division.
8681 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8684 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8686 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8688 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8689 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8691 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8693 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8694 non-scalar integral types.
8695 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8696 (-(A-B), -(~A)): New transformations.
8698 2017-11-07 Tom de Vries <tom@codesourcery.com>
8700 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8701 semicolon after "do {} while (0)".
8702 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8703 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8704 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8705 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8706 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8708 2017-11-07 Tom de Vries <tom@codesourcery.com>
8710 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8711 after "do {} while (0)".
8712 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8713 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8714 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8715 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8716 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8718 2017-11-07 Tom de Vries <tom@codesourcery.com>
8721 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8723 (arm_rtx_costs_internal): Add missing semicolon after
8724 HANDLE_NARROW_SHIFT_ARITH call.
8726 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8728 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8729 disable isel if it was not set explicitly.
8731 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8733 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8734 (add_type_attribute) likewise.
8736 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8738 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8739 of struct ix86_frame.
8740 (ix86_initial_elimination_offset): Likewise.
8741 (ix86_expand_split_stack_prologue): Likewise.
8743 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8745 * tree-vrp.h (enum value_range_type): Update stale comment.
8747 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8748 Alan Hayward <alan.hayward@arm.com>
8749 David Sherwood <david.sherwood@arm.com>
8751 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8752 (aarch64_expand_vec_perm_const): Take the number of units too.
8753 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8754 (aarch64_expand_vec_perm_const): Likewise.
8755 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8756 (vec_perm<mode>): Update accordingly.
8758 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8759 Alan Hayward <alan.hayward@arm.com>
8760 David Sherwood <david.sherwood@arm.com>
8762 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8763 Take the number of units too.
8764 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8765 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8766 but check for a vector mode before rather than after the call.
8767 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8768 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8769 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8770 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8771 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8772 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8773 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8774 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8775 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8776 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8777 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8778 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8779 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8780 (aarch64_sqdmull2_n<mode>): Update accordingly.
8782 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8783 Alan Hayward <alan.hayward@arm.com>
8784 David Sherwood <david.sherwood@arm.com>
8786 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8787 the number of units too.
8788 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8789 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8790 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8791 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8792 (vec_store_lanesxi<mode>): Update accordingly.
8794 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8795 Alan Hayward <alan.hayward@arm.com>
8796 David Sherwood <david.sherwood@arm.com>
8798 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8799 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8800 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8801 of units rather than the mode.
8802 * config/aarch64/iterators.md (nunits): New mode attribute.
8803 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8804 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8805 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8806 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8807 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8808 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8809 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8810 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8811 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8812 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8813 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8814 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8815 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8816 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8817 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8818 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8819 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8820 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8821 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8822 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8823 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8824 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8825 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8826 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8827 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8828 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8829 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8830 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8831 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8832 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8833 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8834 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8835 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8836 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8837 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8838 (aarch64_simd_vec_setv2di): Likewise.
8840 2017-11-06 Carl Love <cel@us.ibm.com>
8842 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8844 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8845 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8847 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8849 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8850 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8851 Add power 8 macro expansions.
8852 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8853 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8854 power 8 instructions. (VSX_XXBR): Add iterator.
8856 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8858 * config/arm/arm.md (predicable_short_it): Change default to "no",
8859 improve documentation, remove uses that are identical to the default.
8860 (enabled_for_depr_it): Rename to enabled_for_short_it.
8861 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8862 Remove default uses.
8863 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8864 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8865 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8866 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8868 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8871 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8872 float128 helper macros, which are no longer used after deleting
8873 the old 'q' built-in functions, and moving the round to odd
8874 built-in functions to being special built-in functions.
8875 (BU_FLOAT128_2): Likewise.
8876 (BU_FLOAT128_1_HW): Likewise.
8877 (BU_FLOAT128_2_HW): Likewise.
8878 (BU_FLOAT128_3_HW): Likewise.
8879 (FABSQ): Delete old 'q' built-in functions.
8880 (COPYSIGNQ): Likewise.
8881 (SQRTF128_ODD): Move round to odd built-in functions to be
8882 special built-in functions, so that we can handle
8883 -mabi=ieeelongdouble.
8884 (TRUNCF128_ODD): Likewise.
8885 (ADDF128_ODD): Likewise.
8886 (SUBF128_ODD): Likewise.
8887 (MULF128_ODD): Likewise.
8888 (DIVF128_ODD): Likewise.
8889 (FMAF128_ODD): Likewise.
8890 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8891 built-in names to 'f128'.
8892 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8893 old 'q' built-in functions, as the machine independent code for
8894 'f128' built-in functions handles this.
8895 (rs6000_expand_builtin): Add expansion for float128 round to odd
8896 functions, keying off on -mabi=ieeelongdouble of whether to use
8897 the KFmode or TFmode variant.
8898 (rs6000_init_builtins): Initialize the _Float128 round to odd
8900 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8901 _Float128 'q' built-in functions are now mapped into the new
8902 'f128' built-in functions.
8904 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8906 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8907 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8909 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8912 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8913 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8914 * toplev.c: Include "ipa-fnsummary.h".
8915 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8917 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8919 PR tree-optimization/82838
8920 * gimple-ssa-store-merging.c
8921 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8922 on a separate gimple_seq which is then appended to seq.
8924 2017-11-06 Jeff Law <law@redhat.com>
8927 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8928 (get_probe_interval): New functions.
8929 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8930 (ix86_adjust_stack_and_probe): Likewise.
8931 (output_adjust_stack_and_probe): Likewise.
8932 (ix86_emit_probe_stack_range): Likewise.
8933 (ix86_expand_prologue): Likewise.
8935 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8937 PR tree-optimization/82816
8938 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8939 if the modes of the two types are the same.
8940 (convert_plusminus_to_widen): Likewise.
8942 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8944 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8946 (usadv16qi): New define_expand.
8947 (usadv8hi): New define_expand.
8949 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8952 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8954 (inline_transform): Likewise.
8955 * predict.c (counts_to_freqs): Remove useless conditional.
8956 * profile-count.h (profile_count::apply_scale): Move sanity check.
8957 * tree-inline.c (copy_bb): Always scale.
8958 (copy_cfg_body): Likewise.
8960 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8963 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8965 (*cmp_ite0): Add enabled_for_depr_it attribute.
8966 (*cmp_ite1): Likewise.
8968 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8970 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8973 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8975 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8976 return the value by pointer.
8977 (extract_range_from_multiplicative_op_1): Update accordingly.
8978 Return as soon as an operation fails.
8980 2017-11-05 Tom de Vries <tom@codesourcery.com>
8983 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8984 (DEF_SANITIZER_BUILTIN): ... here.
8985 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8986 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8988 2017-11-05 Tom de Vries <tom@codesourcery.com>
8991 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8993 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8994 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8995 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8997 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8998 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8999 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9001 2017-11-05 Tom de Vries <tom@codesourcery.com>
9004 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9007 2017-11-04 Michael Clark <michaeljclark@mac.com>
9009 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9010 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9012 (*addsi3_extended): Likewise.
9013 (*addsi3_extended2): Likewise.
9014 (<optab>si3): Likewise.
9015 (<optab>di3): Likewise.
9016 (<optab><mode>3): Likewise.
9017 (<*optabe>si3_internal): Likewise.
9018 (zero_extendqi<SUPERQI:mode>2): Likewise.
9019 (*add<mode>hi3): Likewise.
9020 (*xor<mode>hi3): Likewise.
9021 (<optab>di3): Likewise.
9022 (*<optab>si3_extend): Likewise.
9023 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9024 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9025 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9027 2017-11-04 Andrew Waterman <andrew@sifive.com>
9029 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9030 TARGET_STRICT_ALIGN based upon -mtune argument.
9032 2017-11-04 Andrew Waterman <andrew@sifive.com>
9034 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9036 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9038 * config/i386/i386.c (choose_basereg): Use optional scratch
9039 register and add assertion.
9040 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9041 needed, and don't allocate stack.
9042 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9043 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9044 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9046 2017-11-03 Jeff Law <law@redhat.com>
9048 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9049 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9050 to probe at the start of a noreturn function.
9052 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9054 PR tree-optimization/78821
9055 * gimple-ssa-store-merging.c: Update the file comment.
9056 (MAX_STORE_ALIAS_CHECKS): Define.
9057 (struct store_operand_info): New type.
9058 (store_operand_info::store_operand_info): New constructor.
9059 (struct store_immediate_info): Add rhs_code and ops data members.
9060 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9061 and op1r arguments to the ctor, initialize corresponding data members.
9062 (struct merged_store_group): Add load_align_base and load_align
9064 (merged_store_group::merged_store_group): Initialize them.
9065 (merged_store_group::do_merge): Update them.
9066 (merged_store_group::apply_stores): Pick the constant for
9067 encode_tree_to_bitpos from one of the two operands, or skip
9068 encode_tree_to_bitpos if neither operand is a constant.
9069 (class pass_store_merging): Add process_store method decl. Remove
9070 bool argument from terminate_all_aliasing_chains method decl.
9071 (pass_store_merging::terminate_all_aliasing_chains): Remove
9072 var_offset_p argument and corresponding handling.
9073 (stmts_may_clobber_ref_p): New function.
9074 (compatible_load_p): New function.
9075 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9076 if there is overlap and rhs_code is not INTEGER_CST. For
9077 non-overlapping stores terminate group if rhs is not mergeable.
9078 (get_alias_type_for_stmts): Change first argument from
9079 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9080 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9081 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9083 (get_location_for_stmts): Change first argument from
9084 auto_vec<gimple *> & to vec<gimple *> &.
9085 (struct split_store): Remove orig_stmts data member, add orig_stores.
9086 (split_store::split_store): Create orig_stores rather than orig_stmts.
9087 (find_constituent_stmts): Renamed to ...
9088 (find_constituent_stores): ... this. Change second argument from
9089 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9090 to info structures rather than the statements.
9091 (split_group): Rename ALLOW_UNALIGNED argument to
9092 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9093 it. Adjust find_constituent_stores caller.
9094 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9095 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9096 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9097 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9098 (mem_valid_for_store_merging): New function.
9099 (handled_load): New function.
9100 (pass_store_merging::process_store): New method.
9101 (pass_store_merging::execute): Use process_store method. Adjust
9102 terminate_all_aliasing_chains caller.
9104 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9106 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9107 Return true for more constants, symbols and label references.
9108 (aarch64_valid_floating_const): Remove unused function.
9110 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9113 * config/aarch64/aarch64.c (aarch64_layout_frame):
9114 Undo forcing of LR at bottom of frame.
9116 2017-11-03 Jeff Law <law@redhat.com>
9119 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9120 for int_registers_saved.
9122 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9123 extracted from tree-ssa-dom.c.
9124 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9125 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9126 (record_equivalences_from_incoming_edge): Add additional argument
9127 to single_pred_edge_ignoring_loop_edges call.
9128 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9129 (uncprop_dom_walker::before_dom_children): Add additional argument
9130 to single_pred_edge_ignoring_loop_edges call.
9131 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9132 single_pred_edge_ignoring_loop_edges rather than open coding.
9133 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9135 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9137 * match.pd (-(-A)): Rewrite.
9139 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9141 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9142 (rs6000_emit_int_cmove): New declaration.
9143 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9144 (rs6000_emit_sISEL): Delete.
9145 (rs6000_emit_int_cmove): Make non-static.
9146 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9147 instead of rs6000_emit_sISEL.
9149 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9151 * asan.c (create_cond_insert_point): Maintain profile.
9152 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9154 * basic-block.h (struct basic_block_def): Remove frequency.
9155 (EDGE_FREQUENCY): Use to_frequency
9156 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9158 (find_traces): Update to use to_frequency.
9159 (find_traces_1_round): Likewise; use only IPA counts.
9160 (bb_to_key): Likewise.
9161 (connect_traces): Use IPA counts only.
9162 (copy_bb_p): Update to use to_frequency.
9163 (fix_up_crossing_landing_pad): Likewise.
9164 (sanitize_hot_paths): Likewise.
9165 * bt-load.c (basic_block_freq): Likewise.
9166 * cfg.c (init_flow): Set count_max to uninitialized.
9167 (check_bb_profile): Remove frequencies; check counts.
9168 (dump_bb_info): Do not dump frequencies.
9169 (update_bb_profile_for_threading): Update counts only.
9170 (scale_bbs_frequencies_int): Likewise.
9171 (MAX_SAFE_MULTIPLIER): Remove.
9172 (scale_bbs_frequencies_gcov_type): Update counts only.
9173 (scale_bbs_frequencies_profile_count): Update counts only.
9174 (scale_bbs_frequencies): Update counts only.
9175 * cfg.h (struct control_flow_graph): Add count-max.
9176 (update_bb_profile_for_threading): Update prototype.
9177 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9178 (find_many_sub_basic_blocks): Likewise.
9179 * cfgcleanup.c (try_forward_edges): Likewise.
9180 (try_crossjump_to_edge): Likewise.
9181 * cfgexpand.c (expand_gimple_cond): Likewise.
9182 (expand_gimple_tailcall): Likewise.
9183 (construct_init_block): Likewise.
9184 (construct_exit_block): Likewise.
9185 * cfghooks.c (verify_flow_info): Check consistency of counts.
9186 (dump_bb_for_graph): Do not dump frequencies.
9187 (split_block_1): Do not update frequencies.
9188 (split_edge): Do not update frequencies.
9189 (make_forwarder_block): Do not update frequencies.
9190 (duplicate_block): Do not update frequencies.
9191 (account_profile_record): Do not update frequencies.
9192 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9193 for global heuristics.
9194 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9195 (expected_loop_iterations_unbounded): Use counts only.
9196 * cfgloopmanip.c (scale_loop_profile): Simplify.
9197 (create_empty_loop_on_edge): Simplify
9199 (duplicate_loop_to_header_edge): Simplify
9200 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9201 (update_br_prob_note): Take care of removing note when profile
9203 (relink_block_chain): Do not dump frequency.
9204 (rtl_account_profile_record): Use to_frequency.
9205 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9206 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9207 (cgraph_update_edges_for_call_stmt_node): Likewise.
9208 (cgraph_edge::verify_count_and_frequency): Update.
9209 (cgraph_node::verify_node): Temporarily disable frequency verification.
9210 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9211 to_cgraph_frequency.
9212 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9213 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9215 (cgraph_node::expand_thunk): Update profile.
9216 * except.c (dw2_build_landing_pads): Do not update frequency.
9217 * final.c (compute_alignments): Use to_frequency.
9218 (dump_basic_block_info): Do not dump frequency.
9219 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9220 (dump_gimple_bb_header): Do not dump frequency.
9221 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9223 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9224 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9225 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9226 (init_before_recovery): Do not update frequency.
9227 (sched_create_recovery_edges): Do not update frequency.
9228 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9229 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9230 (ipa_cp_c_finalize): Set max_count to uninitialized.
9231 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9232 (param_change_prob): Use counts.
9233 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9235 * ipa-split.c (consider_split): Use to_frequency.
9236 (split_function): Use to_frequency.
9237 * ira-build.c (loop_compare_func): Likewise.
9238 (mark_loops_for_removal): Likewise.
9239 (mark_all_loops_for_removal): Likewise.
9240 * loop-doloop.c (doloop_modify): Do not update frequency.
9241 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9243 * lto-streamer-in.c (input_function): Update count_max.
9244 * omp-expand.c (expand_omp_taskreg): Update count_max.
9245 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9246 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9247 (maybe_hot_count_p): Use ipa counts only.
9248 (maybe_hot_bb_p): Simplify.
9249 (maybe_hot_edge_p): Simplify.
9250 (probably_never_executed): Do not take frequency argument.
9251 (probably_never_executed_bb_p): Do not pass frequency.
9252 (probably_never_executed_edge_p): Likewise.
9253 (combine_predictions_for_bb): Check that profile is nonzero.
9254 (propagate_freq): Do not set frequency.
9255 (drop_profile): Simplify.
9256 (counts_to_freqs): Simplify.
9257 (expensive_function_p): Use to_frequency.
9258 (propagate_unlikely_bbs_forward): Simplify.
9259 (determine_unlikely_bbs): Simplify.
9260 (estimate_bb_frequencies): Add hack to silence graphite issues.
9261 (compute_function_frequency): Use ipa counts.
9262 (pass_profile::execute): Update.
9263 (rebuild_frequencies): Use counts only.
9264 (force_edge_cold): Use counts only.
9265 * profile-count.c (profile_count::dump): Dump new count types.
9266 (profile_count::differs_from_p): Check compatiblity.
9267 (profile_count::to_frequency): New function.
9268 (profile_count::to_cgraph_frequency): New function.
9269 * profile-count.h (struct function): Declare.
9270 (enum profile_quality): Add profile_guessed_local and
9271 profile_guessed_global0.
9272 (class profile_proability): Decrease number of bits to 29;
9273 update from_reg_br_prob_note and to_reg_br_prob_note.
9274 (class profile_count: Update comment; decrease number of bits
9275 to 61. Check compatibility.
9276 (profile_count::compatible_p): New private member function.
9277 (profile_count::ipa_p): New member function.
9278 (profile_count::operator<): Handle global zero correctly.
9279 (profile_count::operator>): Handle global zero correctly.
9280 (profile_count::operator<=): Handle global zero correctly.
9281 (profile_count::operator>=): Handle global zero correctly.
9282 (profile_count::nonzero_p): New member function.
9283 (profile_count::force_nonzero): New member function.
9284 (profile_count::max): New member function.
9285 (profile_count::apply_scale): Handle IPA scalling.
9286 (profile_count::guessed_local): New member function.
9287 (profile_count::global0): New member function.
9288 (profile_count::ipa): New member function.
9289 (profile_count::to_frequency): Declare.
9290 (profile_count::to_cgraph_frequency): Declare.
9291 * profile.c (OVERLAP_BASE): Delete.
9292 (compute_frequency_overlap): Delete.
9293 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9294 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9295 * sched-ebb.c (rank): Use counts only.
9296 * shrink-wrap.c (handle_simple_exit): Use counts only.
9297 (try_shrink_wrapping): Use counts only.
9298 (place_prologue_for_one_component): Use counts only.
9299 * tracer.c (find_best_predecessor): Use to_frequency.
9300 (find_trace): Use to_frequency.
9301 (tail_duplicate): Use to_frequency.
9302 * trans-mem.c (expand_transaction): Do not update frequency.
9303 * tree-call-cdce.c: Do not update frequency.
9304 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9305 (gimple_merge_blocks): Likewise.
9306 (gimple_split_edge): Likewise.
9307 (gimple_duplicate_sese_region): Likewise.
9308 (gimple_duplicate_sese_tail): Likewise.
9309 (move_sese_region_to_fn): Likewise.
9310 (gimple_account_profile_record): Likewise.
9311 (insert_cond_bb): Likewise.
9312 * tree-complex.c (expand_complex_div_wide): Likewise.
9313 * tree-eh.c (lower_resx): Update profile.
9314 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9316 (initialize_cfun): Do not initialize frequencies
9317 (freqs_to_counts): Delete.
9318 (copy_cfg_body): Ignore count parameter.
9319 (copy_body): Update.
9320 (expand_call_inline): Update count_max.
9321 (optimize_inline_calls): Update count_max.
9322 (tree_function_versioning): Update count_max.
9323 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9324 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9326 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9327 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9328 (try_peel_loop): Likewise.
9329 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9331 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9332 (tree_transform_and_unroll_loop): Do not use frequencies
9333 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9334 Use reliable prediction only.
9335 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9336 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9337 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9338 probability scaling.
9339 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9340 not update frequency
9341 (any_remaining_duplicated_blocks): Likewise.
9342 (update_profile): Likewise.
9343 (estimated_freqs_path): Delete.
9344 (freqs_to_counts_path): Delete.
9345 (clear_counts_path): Delete.
9346 (ssa_fix_duplicate_block_edges): Likewise.
9347 (duplicate_thread_path): Likewise.
9348 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9349 * tree-tailcall.c (decrease_profile): Do not update frequency.
9350 (eliminate_tail_call): Likewise.
9351 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9352 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9353 (optimize_mask_stores): Likewise.
9354 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9355 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9356 (ubsan_expand_ptr_ifn): Update profile.
9357 * value-prof.c (gimple_ic): Simplify.
9358 * value-prof.h (gimple_ic): Update prototype.
9359 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9360 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9362 (want_inline_self_recursive_call_p): Likewise.
9363 (resolve_noninline_speculation): Only cummulate defined counts.
9364 (inline_small_functions): Use nonzero_p.
9365 (ipa_inline): Do not access freed node.
9367 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9369 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9370 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9372 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9374 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9375 non-legitimate address.
9377 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9379 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9380 (*lt0_<mode>di, *lt0_<mode>si): New.
9382 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9384 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9385 TARGET_PAIRED_FLOAT.
9387 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9388 Jim Wilson <jim.wilson@linaro.org>
9390 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9391 * config/aarch64/aarch64-tune.md: Regenerated.
9392 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9393 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9395 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9397 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9398 (arc_expand_prologue): Restore blink for millicode.
9399 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9401 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9404 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9405 gen_vec_duplicate after forcing the scalar into a register.
9407 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9409 * combine (try_combine): Print the insns input to try_combine to the
9412 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9415 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9416 Remove second argument from aarch64_process_target_attr call.
9417 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9419 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9421 (aarch64_handle_attr_arch): Remove second argument.
9422 (aarch64_handle_attr_cpu): Ditto.
9423 (aarch64_handle_attr_tune): Ditto.
9424 (aarch64_handle_attr_isa_flags): Ditto.
9425 (aarch64_process_one_target_attr): Ditto.
9426 (aarch64_process_target_attr): Ditto.
9427 (aarch64_option_valid_attribute_p): Remove second argument.
9428 on aarch64_process_target_attr call.
9430 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9432 * diagnostic.c: Include "selftest-diagnostic.h".
9433 (selftest::assert_location_text): New function.
9434 (selftest::test_diagnostic_get_location_text): New function.
9435 (selftest::diagnostic_c_tests): Call it.
9437 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9439 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9440 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9441 (class selftest::test_diagnostic_context): Move to...
9442 * selftest-diagnostic.c: New file.
9443 * selftest-diagnostic.h: New file.
9445 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9447 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9448 offset range for FT32B.
9449 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9450 * config/ft32/ft32.md: Add TARGET_NOPM.
9451 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9452 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9454 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9456 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9458 2017-11-02 Jeff Law <law@redhat.com>
9460 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9461 virtual keyword on FINAL OVERRIDE members.
9463 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9464 virtuals together. Add virtual destructor.
9465 (substitute_and_fold_engine): Similarly.
9467 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9469 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9471 2017-11-02 Richard Biener <rguenther@suse.de>
9473 PR tree-optimization/82795
9474 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9476 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9478 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9480 (gcc_GAS_CHECK_FEATURE): Remove.
9481 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9482 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9483 * configure: Regenerate.
9485 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9487 * config/arc/arc.c (hwloop_optimize): Account for empty
9490 2017-11-02 Richard Biener <rguenther@suse.de>
9493 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9494 Truncate ARRAY_REF index and element size.
9496 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9498 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9500 2017-11-01 Jeff Law <law@redhat.com>
9502 * tree-ssa-ccp.c (ccp_folder): New class derived from
9503 substitute_and_fold_engine.
9504 (ccp_folder::get_value): New member function.
9505 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9506 (ccp_fold_stmt): Remove prototype.
9507 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9508 * tree-ssa-copy.c (copy_folder): New class derived from
9509 substitute_and_fold_engine.
9510 (copy_folder::get_value): Renamed from get_value.
9511 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9512 * tree-vrp.c (vrp_folder): New class derived from
9513 substitute_and_fold_engine.
9514 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9515 (vrp_folder::get_value): New member function.
9516 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9517 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9518 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9519 provide a class interface to folder/substitute routines.
9520 (ssa_prop_fold_stmt_fn): Remove typedef.
9521 (ssa_prop_get_value_fn): Likewise.
9522 (subsitute_and_fold): Remove prototype.
9523 (replace_uses_in): Likewise.
9524 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9525 Renamed from replace_uses_in. Call the virtual member function
9526 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9527 (substitute_and_fold_dom_walker): Remove initialization of
9528 data member entries for calbacks. Add substitute_and_fold_engine
9529 member and initialize it.
9530 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9531 member functions for get_value, replace_phi_args_in c
9532 replace_uses_in, and fold_stmt calls.
9533 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9534 substitute_and_fold. Remove assert. Update ctor call.
9536 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9537 (ssa_prop_visit_phi_fn): Likewise.
9538 (class ssa_propagation_engine): New class to provide an interface
9540 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9542 (ssa_prop_visit_phi): Likewise.
9543 (ssa_propagation_engine::simulate_stmt): Moved into class.
9544 Call visit_phi/visit_stmt from the class rather than via
9545 file scoped static variables.
9546 (ssa_propagation_engine::simulate_block): Moved into class.
9547 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9548 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9549 set file scoped statics for the visit_stmt/visit_phi callbacks.
9550 * tree-complex.c (complex_propagate): New class derived from
9551 ssa_propagation_engine.
9552 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9553 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9554 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9556 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9557 ssa_propagation_engine.
9558 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9559 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9560 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9561 * tree-ssa-copy.c (copy_prop): New class derived from
9562 ssa_propagation_engine.
9563 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9564 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9565 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9566 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9567 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9568 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9569 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9571 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9573 PR rtl-optimization/82778
9574 PR rtl-optimization/82597
9575 * compare-elim.c (struct comparison): Add in_a_setter field.
9576 (find_comparison_dom_walker::before_dom_children): Remove killed
9577 bitmap and df_simulate_find_defs call, instead walk the defs.
9578 Compute last_setter and initialize in_a_setter. Merge definitions
9579 with first initialization for a few variables.
9580 (try_validate_parallel): Use insn_invalid_p instead of
9581 recog_memoized. Return insn rather than just the pattern.
9582 (try_merge_compare): Fix up comment. Don't uselessly test if
9583 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9585 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9588 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9589 Alan Hayward <alan.hayward@arm.com>
9590 David Sherwood <david.sherwood@arm.com>
9592 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9593 aarch64_hard_regno_nregs to get the number of registers
9596 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9597 Alan Hayward <alan.hayward@arm.com>
9598 David Sherwood <david.sherwood@arm.com>
9600 * config/aarch64/constraints.md (Upl): Rename to...
9602 * config/aarch64/aarch64.md
9603 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9606 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9607 Alan Hayward <alan.hayward@arm.com>
9608 David Sherwood <david.sherwood@arm.com>
9610 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9611 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9614 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9615 Alan Hayward <alan.hayward@arm.com>
9616 David Sherwood <david.sherwood@arm.com>
9618 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9619 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9620 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9622 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9623 of a single element.
9624 * config/aarch64/iterators.md: Add a comment above the permute
9625 unspecs to say that they are generated directly by
9626 aarch64_expand_vec_perm_const.
9627 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9629 2017-11-01 Nathan Sidwell <nathan@acm.org>
9631 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9633 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9635 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9636 and medany code models, and describe what they do.
9638 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9640 Revert accidental duplicate:
9642 * combine.c (can_change_dest_mode): Reject changes in
9643 REGMODE_NATURAL_SIZE.
9645 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9647 PR rtl-optimization/64682
9648 PR rtl-optimization/69567
9649 PR rtl-optimization/69737
9650 PR rtl-optimization/82683
9651 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9652 register mentioned in the note, drop the note, unless it came from I3,
9653 in which case it should go to I3 again.
9655 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9657 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9658 and return false if not.
9659 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9663 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9664 range is known to be empty.
9666 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9667 Alan Hayward <alan.hayward@arm.com>
9668 David Sherwood <david.sherwood@arm.com>
9670 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9671 and CONST_VECTOR_NUNITS instead of computing the number of units from
9672 the byte sizes of the vector and element.
9673 (simplify_binary_operation_1): Likewise.
9674 (simplify_const_binary_operation): Likewise.
9675 (simplify_ternary_operation): Likewise.
9677 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9678 Alan Hayward <alan.hayward@arm.com>
9679 David Sherwood <david.sherwood@arm.com>
9681 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9682 (int_mem_offset): ...this new function.
9683 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9684 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9685 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9688 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9689 Alan Hayward <alan.hayward@arm.com>
9690 David Sherwood <david.sherwood@arm.com>
9692 * lower-subreg.c (interesting_mode_p): New function.
9693 (compute_costs, find_decomposable_subregs, decompose_register)
9694 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9695 (resolve_clobber, dump_choices): Use it.
9697 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9698 Alan Hayward <alan.hayward@arm.com>
9699 David Sherwood <david.sherwood@arm.com>
9701 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9703 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9704 Alan Hayward <alan.hayward@arm.com>
9705 David Sherwood <david.sherwood@arm.com>
9707 * alias.c (find_base_value, find_base_term): Only process integer
9708 truncations. Check the precision rather than the size.
9710 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9711 Alan Hayward <alan.hayward@arm.com>
9712 David Sherwood <david.sherwood@arm.com>
9714 * machmode.h (is_narrower_int_mode): New function
9715 * optabs.c (expand_float, expand_fix): Use it.
9716 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9718 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9719 Alan Hayward <alan.hayward@arm.com>
9720 David Sherwood <david.sherwood@arm.com>
9722 * rtl.h (narrower_subreg_mode): New function.
9723 * ira-color.c (update_costs_from_allocno): Use it.
9725 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9726 Alan Hayward <alan.hayward@arm.com>
9727 David Sherwood <david.sherwood@arm.com>
9729 * optabs-query.h (convert_optab_p): New function, split out from...
9730 (convert_optab_handler): ...here.
9731 (widening_optab_handler): Delete.
9732 (find_widening_optab_handler): Remove permit_non_widening parameter.
9733 (find_widening_optab_handler_and_mode): Likewise. Provide an
9734 override that operates on mode class wrappers.
9735 * optabs-query.c (widening_optab_handler): Delete.
9736 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9737 parameter. Assert that the two modes are the same class and that
9738 the "from" mode is narrower than the "to" mode. Use
9739 convert_optab_handler instead of widening_optab_handler.
9740 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9741 instead of widening_optab_handler.
9742 * expr.c (expand_expr_real_2): Update calls to
9743 find_widening_optab_handler.
9744 * optabs.c (expand_widen_pattern_expr): Likewise.
9745 (expand_binop_directly): Take the insn_code as a parameter.
9746 (expand_binop): Only call find_widening_optab_handler for
9747 conversion optabs; use optab_handler otherwise. Update calls
9748 to find_widening_optab_handler and expand_binop_directly.
9749 Use convert_optab_handler instead of widening_optab_handler.
9750 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9751 find_widening_optab_handler and use scalar_mode rather than
9753 (convert_plusminus_to_widen): Likewise.
9755 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9756 Alan Hayward <alan.hayward@arm.com>
9757 David Sherwood <david.sherwood@arm.com>
9759 * machmode.h (fixed_size_mode): New class.
9760 * rtl.h (get_pool_mode): Return fixed_size_mode.
9761 * gengtype.c (main): Add fixed_size_mode.
9762 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9763 (get_raw_arg_mode): Likewise.
9764 * doc/tm.texi: Regenerate.
9765 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9766 * targhooks.c (default_get_reg_raw_mode): Likewise.
9767 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9768 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9769 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9770 (msp430_get_raw_result_mode): Likewise.
9771 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9772 * dbxout.c (dbxout_parms): Require fixed-size modes.
9773 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9774 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9775 * omp-low.c (lower_oacc_reductions): Likewise.
9776 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9777 (simplify_subreg): Update accordingly.
9778 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9779 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9780 that aren't fixed-size.
9781 (get_pool_mode): Return a fixed_size_mode.
9782 (output_constant_pool_2): Take a fixed_size_mode.
9784 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9785 Alan Hayward <alan.hayward@arm.com>
9786 David Sherwood <david.sherwood@arm.com>
9788 * doc/rtl.texi (vec_series): Document.
9789 (const): Say that the operand can be a vec_series.
9790 * rtl.def (VEC_SERIES): New rtx code.
9791 * rtl.h (const_vec_series_p_1): Declare.
9792 (const_vec_series_p): New function.
9793 * emit-rtl.h (gen_const_vec_series): Declare.
9794 (gen_vec_series): Likewise.
9795 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9796 (gen_vec_series): Likewise.
9797 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9798 * simplify-rtx.c (simplify_unary_operation): Handle negations
9800 (simplify_binary_operation_series): New function.
9801 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9802 (test_vector_ops_series): New function.
9803 (test_vector_ops): Call it.
9804 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9805 gen_const_vec_series.
9806 (altivec_lvsr): Likewise.
9807 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9809 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9810 Alan Hayward <alan.hayward@arm.com>
9811 David Sherwood <david.sherwood@arm.com>
9813 * doc/rtl.texi (const): Update description of address constants.
9814 Say that vector constants are allowed too.
9815 * common.md (E, F): Use CONSTANT_P instead of checking for
9817 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9818 checking for CONST_VECTOR.
9819 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9821 * expr.c (expand_expr_real_2): Check for vector modes instead
9822 of checking for CONST_VECTOR.
9823 * rtl.h (const_vec_p): New function.
9824 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9825 (unwrap_const_vec_duplicate): Handle them here too.
9827 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9828 David Malcolm <dmalcolm@redhat.com>
9829 Alan Hayward <alan.hayward@arm.com>
9830 David Sherwood <david.sherwood@arm.com>
9832 * rtl.h (vec_duplicate_p): New function.
9833 * selftest-rtl.c (assert_rtx_eq_at): New function.
9834 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9835 (assert_rtx_eq_at): Declare.
9836 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9837 * selftest-run-tests.c (selftest::run_tests): Call it.
9838 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9839 (simplify_unary_operation_1): Recursively handle vector duplicates.
9840 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9842 (simplify_subreg): Handle subregs of vector duplicates.
9843 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9844 (selftest::simplify_rtx_c_tests): New functions.
9846 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9847 Alan Hayward <alan.hayward@arm.com>
9848 David Sherwood <david.sherwood@arm.com>
9850 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9851 (gen_vec_duplicate): Likewise.
9852 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9854 (gen_const_vector): ...here.
9855 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9856 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9857 whose elements are all equal.
9858 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9859 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9860 (simplify_relational_operation): Likewise.
9861 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9863 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9864 (aarch64_expand_vector_init): Likewise.
9865 * config/arm/arm.c (neon_vdup_constant): Likewise.
9866 (neon_expand_vector_init): Likewise.
9867 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9868 (arm_block_set_unaligned_vect): Likewise.
9869 (arm_block_set_aligned_vect): Likewise.
9870 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9871 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9872 (expand_vec_perm_even_odd_pack): Likewise.
9873 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9874 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9875 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9876 gen_const_vec_duplicate.
9877 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9878 * config/mips/mips.c (mips_gen_const_int_vector): Use
9879 gen_const_vec_duplicate.
9880 (mips_expand_vector_init): Use CONST0_RTX.
9881 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9882 (define_split): Use gen_const_vec_duplicate.
9883 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9884 (define_split): Use gen_const_vec_duplicate.
9885 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9886 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9887 * config/spu/spu.c (spu_const): Likewise.
9889 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9890 Alan Hayward <alan.hayward@arm.com>
9891 David Sherwood <david.sherwood@arm.com>
9893 * combine.c (can_change_dest_mode): Reject changes in
9894 REGMODE_NATURAL_SIZE.
9896 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9898 * configure.ac (--enable-libssp): New.
9899 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9900 trying to determine target-specific default. Adjust indentation.
9901 * configure: Regenerated.
9902 * doc/install.texi (Configuration): Expand --disable-libssp
9905 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9907 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9910 2017-10-31 Martin Jambor <mjambor@suse.cz>
9913 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9915 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9917 * auto-profile.c (autofdo_source_profile::read): Use
9918 UNKNOWN_LOCATION rather than 0.
9919 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9920 (warning_at): ...this overload.
9921 (warning_at_rich_loc_n): Rename to...
9922 (warning_n): ...this overload.
9923 (error_at_rich_loc): Rename to...
9924 (error_at): ...this overload.
9925 (pedwarn_at_rich_loc): Rename to...
9926 (pedwarn): ...this overload.
9927 (permerror_at_rich_loc): Rename to...
9928 (permerror): ...this overload.
9929 (inform_at_rich_loc): Rename to...
9930 (inform): ...this overload.
9931 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9932 (diagnostic_n_impl_richloc): Rename to...
9933 (diagnostic_n_impl): ...this rich_location *-based decl.
9934 (inform_at_rich_loc): Rename to...
9935 (inform): ...this, and add an assertion.
9936 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9937 (warning_at_rich_loc): Rename to...
9938 (warning_at): ...this, and add an assertion.
9939 (warning_at_rich_loc_n): Rename to...
9940 (warning_n): ...this, and add an assertion.
9941 (warning_n): Update location_t-based implementation for removal of
9942 location_t-based diagnostic_n_impl.
9943 (pedwarn_at_rich_loc): Rename to...
9944 (pedwarn): ...this, and add an assertion.
9945 (permerror_at_rich_loc): Rename to...
9946 (permerror): ...this, and add an assertion.
9947 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9948 (error_at_rich_loc): Rename to...
9949 (error_at): ...this, and add an assertion.
9950 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9951 (driver::do_spec_on_infiles): Likewise.
9952 * substring-locations.c (format_warning_va): Update for renaming
9953 of inform_at_rich_loc.
9955 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9957 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9958 _Float<N>X built-in functions so that the variant without the
9959 "__builtin_" prefix is only enabled for the GNU C and Objective C
9960 languages when they are in non-strict ANSI/ISO mode.
9961 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9962 * target.def (floatn_builtin_p): Add a target hook to control
9963 whether _Float<N> and _Float<N>X built-in functions without the
9964 "__builtin_" prefix are enabled, and return true for C and
9965 Objective C in the default hook. Include langhooks.h in
9967 * targhooks.h (default_floatn_builtin_p): Likewise.
9968 * targhooks.c (default_floatn_builtin_p): Likewise.
9969 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9970 floatn_builtin_p target hook.
9971 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9973 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9974 Eric Botcazou <ebotcazou@adacore.com>
9976 PR rtl-optimization/81803
9977 * lra-constraints.c (curr_insn_transform): Also reload the whole
9978 register for a strict subreg no wider than a word if this is for
9979 a WORD_REGISTER_OPERATIONS target.
9981 2017-10-31 Jason Merrill <jason@redhat.com>
9983 * gdbinit.in: Skip over inlines from timevar.h.
9985 2017-10-31 Martin Liska <mliska@suse.cz>
9987 * doc/gcov.texi: Document new option.
9988 * gcov.c (print_usage): Likewise print it.
9989 (process_args): Support the argument.
9990 (format_count): New function.
9991 (format_gcov): Use the function.
9993 2017-10-31 Martin Liska <mliska@suse.cz>
9995 * gcov.c (struct name_map): do not use typedef.
9996 Define operator== and operator<.
9997 (name_search): Remove.
9998 (name_sort): Remove.
9999 (main): Do not allocate names.
10000 (process_file): Add vertical space.
10001 (generate_results): Use std::find.
10002 (release_structures): Do not release memory.
10003 (find_source): Use std::find.
10005 2017-10-31 Martin Liska <mliska@suse.cz>
10007 * gcov.c (struct line_info): Remove it's typedef.
10008 (line_info::line_info): Add proper ctor.
10009 (line_info::has_block): Do not use a typedef.
10010 (struct source_info): Do not use typedef.
10011 (circuit): Likewise.
10012 (get_cycles_count): Likewise.
10013 (output_intermediate_file): Iterate via vector iterator.
10014 (add_line_counts): Use std::vector methods.
10015 (accumulate_line_counts): Likewise.
10016 (output_lines): Likewise.
10018 2017-10-31 Martin Liska <mliska@suse.cz>
10020 * gcov.c (struct source_info): Remove typedef.
10021 (source_info::source_info): Add proper ctor.
10022 (accumulate_line_counts): Use struct, not it's typedef.
10023 (output_gcov_file): Likewise.
10024 (output_lines): Likewise.
10025 (main): Do not allocate an array.
10026 (output_intermediate_file): Use size of vector container.
10027 (process_file): Resize the vector.
10028 (generate_results): Do not preallocate, use newly added vector
10030 (release_structures): Do not release sources.
10031 (find_source): Use vector methods.
10032 (add_line_counts): Do not use typedef.
10034 2017-10-31 Martin Liska <mliska@suse.cz>
10036 * doc/gcov.texi: Document that.
10037 * gcov.c (add_line_counts): Mark lines with a non-executed
10039 (output_line_beginning): Handle such lines.
10040 (output_lines): Pass new argument.
10041 (output_intermediate_file): Print it in intermediate format.
10043 2017-10-31 Martin Liska <mliska@suse.cz>
10045 * color-macros.h: New file.
10046 * diagnostic-color.c: Factor out color related to macros to
10048 * doc/gcov.texi: Document -k option.
10049 * gcov.c (INCLUDE_STRING): Include string.h.
10050 (print_usage): Add -k option.
10051 (process_args): Parse it.
10052 (pad_count_string): New function.
10053 (output_line_beginning): Likewise.
10054 (DEFAULT_LINE_START): New macro.
10055 (output_lines): Support color output.
10057 2017-10-31 Martin Liska <mliska@suse.cz>
10059 PR gcov-profile/82633
10060 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10061 their interaction with GCOV infrastructure.
10062 * configure.ac: Add -fkeep-{inline,static}-functions to
10064 * configure: Regenerate.
10066 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10069 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10071 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10074 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10075 into a register if it does not fit into an immediate offset field.
10077 2017-10-31 Olivier Hainque <hainque@adacore.com>
10079 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10081 2017-10-31 Julia Koval <julia.koval@intel.com>
10083 * config.gcc: Add gfniintrin.h.
10084 * config/i386/gfniintrin.h: New.
10085 * config/i386/i386-builtin-types.def
10086 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10087 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10088 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10089 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10090 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10091 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10092 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10093 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10094 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10095 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10096 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10097 * config/i386/immintrin.h: Include gfniintrin.h.
10098 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10100 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10102 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10104 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10106 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10107 (arm_ashldi3_1bit): Remove pattern.
10108 (ashrdi3): Remove shift by 1 expansion.
10109 (arm_ashrdi3_1bit): Remove pattern.
10110 (lshrdi3): Remove shift by 1 expansion.
10111 (arm_lshrdi3_1bit): Remove pattern.
10112 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10113 cost of ashldi3 by 1.
10114 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10115 (<shift>di3_neon): Likewise.
10117 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10119 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10120 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10121 and (*aarch64_simd_mov<VQ:mode>).
10122 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10123 pattern alternative.
10124 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10125 attributes to match pattern alternative.
10127 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10129 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10130 * config/rs6000/emmintrin.h: New file.
10131 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10133 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10135 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10136 * (movdi_vfp_cortexa8): Remove pattern.
10138 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10140 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10142 (Specific, alpha*-dec-osf5.1): Remove.
10143 (Specific, mips-sgi-irix5): Remove.
10144 (Specific, mips-sgi-irix6): Remove.
10146 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10148 PR middle-end/22141
10149 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10150 arguments to clear_bit_region_be.
10152 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10154 * gimplify.c: Include memmodel.h.
10156 2017-10-30 Martin Jambor <mjambor@suse.cz>
10158 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10159 condition whether loop should be executed at all.
10161 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10164 gimple folding of vec_madd() intrinsics.
10165 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10166 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10167 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10169 2017-10-30 Richard Biener <rguenther@suse.de>
10171 PR tree-optimization/82762
10173 2017-10-23 Richard Biener <rguenther@suse.de>
10175 PR tree-optimization/82129
10177 2017-08-01 Richard Biener <rguenther@suse.de>
10179 PR tree-optimization/81181
10180 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10181 (compute_antic): ... end of iteration here.
10183 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10185 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10186 -std=iso9899:2017 and -std=gnu17.
10187 * doc/standards.texi (C Language): Document C17 support.
10188 * doc/cpp.texi (Overview): Mention -std=c17.
10189 (Standard Predefined Macros): Document C11 and C17 values of
10190 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10191 * doc/extend.texi (Inline): Do not list individual options for
10192 standards newer than C99.
10193 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10195 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10198 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10200 * asan.c (asan_finish_file): Align asan globals array by shadow
10203 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10205 PR middle-end/22141
10206 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10207 (struct store_immediate_info): Add bitregion_start and bitregion_end
10209 (store_immediate_info::store_immediate_info): Add brs and bre
10210 arguments and initialize bitregion_{start,end} from those.
10211 (struct merged_store_group): Add bitregion_start, bitregion_end,
10212 align_base and mask fields. Drop unnecessary struct keyword from
10213 struct store_immediate_info. Add do_merge method.
10214 (clear_bit_region_be): Use memset instead of loop storing zeros.
10215 (merged_store_group::do_merge): New method.
10216 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10217 stores as long as the surrounding bitregions have no gaps.
10218 (merged_store_group::merge_overlapping): Use do_merge.
10219 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10220 is byte aligned, rather than requiring that start and width are
10221 byte aligned. Drop unnecessary struct keyword from
10222 struct store_immediate_info. Allocate and populate also mask array.
10223 Make start of the arrays relative to bitregion_start rather than
10224 start and size them according to bitregion_{end,start} difference.
10225 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10226 struct store_immediate_info.
10227 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10228 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10229 struct keyword from struct store_immediate_info.
10230 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10231 between stores as long as the surrounding bitregions have no gaps.
10233 (struct split_store): Add orig non-static data member.
10234 (split_store::split_store): Initialize orig to false.
10235 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10236 if there is exactly a single original stmt. Change stmts argument
10237 to pointer from reference, if NULL, don't push anything to it. Add
10238 first argument, use it to optimize skipping over orig stmts that
10239 are known to be before bitpos already. Simplify.
10240 (split_group): Return unsigned int count how many stores are or
10241 would be needed rather than a bool. Add allow_unaligned argument.
10242 Change split_stores argument from reference to pointer, if NULL,
10243 only do a dry run computing how many stores would be produced.
10244 Rewritten algorithm to use both alignment and misalign if
10245 !allow_unaligned and handle bitfield stores with gaps.
10246 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10247 from bitregion_start instead of start. Compute allow_unaligned
10248 here, if true, do 2 split_group dry runs to compute which one
10249 produces fewer stores and prefer aligned if equal. Punt if
10250 new count is bigger or equal than original before emitting any
10251 statements, rather than during that. Remove no longer needed
10252 new_ssa_names tracking. Replace num_stmts with
10253 split_stores.length (). Use 32-bit stack allocated entries
10254 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10255 if possible. Handle bitfields with gaps.
10256 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10257 Compute bitregion_{start,end} for the stores and construct
10258 store_immediate_info with that. Formatting fixes.
10260 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10263 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10264 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10266 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10268 * gimplify.c: Include tm_p.h.
10270 * common.opt (gcoff): Re-add as ignored option.
10271 (gcoff1, gcoff2, gcoff3): Likewise.
10273 * Makefile.in (OBJS): Delete sdbout.o.
10274 (GTFILES): Delete $(srcdir)/sdbout.c.
10275 * debug.h: Delete sdb_debug_hooks.
10276 * final.c: Delete sdbout.h include.
10277 (final_scan_insn): Delete SDB_DEBUG check.
10278 (rest_of_clean_state): Likewise.
10279 * output.h: Delete sdb_begin_function_line.
10280 * sdbout.c: Delete.
10281 * sdbout.h: Delete.
10282 * toplev.c: Delete sdbout.h include.
10283 (process_options): Delete SDB_DEBUG check.
10284 * tree-core.h (tree_type_common): Delete pointer field of
10286 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10287 TYPE_SYMTAB_POINTER.
10288 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10289 (TYPE_SYMTAB_IS_POINTER): Delete.
10290 (TYPE_SYMTAB_IS_DIE): Renumber.
10291 * xcoffout.c: Refer to former sdbout.c file.
10292 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10294 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10295 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10296 (Debugging Options): Delete -gcoff.
10297 (-gstabs): Delete SDB reference.
10299 (-gcoff@var{level}): Delete.
10300 * doc/passes.texi (Debugging information output): Delete SDB and
10301 sdbout.c references.
10302 * doc/tm.texi: Regenerate.
10303 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10304 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10306 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10307 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10308 (SDB_DEBUGGING_INFO): Delete.
10309 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10310 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10311 * target.def (output_source_filename): Delete COFF reference.
10313 * common.opt (gcoff): Delete.
10314 (gxcoff+): Update Negative chain.
10315 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10317 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10318 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10319 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10320 and SDB references.
10321 (expand_function_start): Change sdb reference to past tense.
10322 (expand_function_end): Change sdb reference to past tense.
10323 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10324 * opts.c (debug_type_names): Delete coff entry.
10325 (common_handle_option): Delete OPT_gcoff case.
10326 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10328 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10329 * config/cris/cris.h: Delete SDB reference in comment.
10330 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10331 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10332 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10333 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10335 (ix86_expand_prologue): Likewise.
10336 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10337 * config/ia64/ia64.h: Likewise.
10338 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10339 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10341 * config/mmix/mmix.h: Likewise.
10342 * config/nds32/nds32.c: Likewise.
10343 * config/stormy/storym16.h: Likewise.
10344 * config/visium/visium.h: Likewise.
10345 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10347 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10349 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10350 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10351 FRAME_POINTER_REGNUM point at high end of local var area.
10353 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10355 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10356 Move comment around. Do not reset best_edge for a copiable
10357 destination if the copy would cause a partition change.
10358 (better_edge_p): Remove redundant check.
10360 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10362 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10364 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10366 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10367 for math functions that have _Float<N> and _Float<N>X variants.
10368 (mathfn_built_in_2): Add support for math functions that have
10369 _Float<N> and _Float<N>X variants.
10370 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10371 (expand_builtin_mathfn_ternary): Add support for fma with
10372 _Float<N> and _Float<N>X variants.
10373 (expand_builtin): Likewise.
10374 (fold_builtin_3): Likewise.
10375 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10376 create math function _Float<N> and _Float<N>X variants as external
10378 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10379 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10380 the __builtin_ prefix and if not strict ansi, without the prefix.
10381 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10382 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10383 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10384 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10385 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10386 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10387 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10388 function signatures for fma _Float<N> and _Float<N>X variants.
10389 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10390 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10391 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10392 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10393 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10394 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10395 * gencfn-macros.c (print_case_cfn): Add support for math functions
10396 that have _Float<N> and _Float<N>X variants.
10397 (print_define_operator_list): Likewise.
10398 (fltfn_suffixes): Likewise.
10400 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10401 for math functions that have _Float<N> and _Float<N>X variants.
10402 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10403 and _Float<N>X variants.
10404 (COPYSIGN): Likewise.
10407 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10408 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10410 (integer_valued_read_call_p): Likewise.
10411 * fold-const-call.c (fold_const_call_ss): Likewise.
10412 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10413 _Float<N> and _Float<N>X variants.
10414 (fold_const_call_ssss): Add support for fma _Float<N> and
10415 _Float<N>X variants.
10416 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10417 support for copysign and fma _Float<N> and _Float<N>X variants.
10418 (backprop::process_builtin_call_use): Likewise.
10419 * tree-call-cdce.c (can_test_argument_range); Add support for
10420 sqrt _Float<N> and _Float<N>X variants.
10421 (edom_only_function): Likewise.
10422 (get_no_error_domain): Likewise.
10423 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10424 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10425 copysign _Float<N> and _Float<N>X variants.
10426 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10427 handled by machine independent code.
10428 (FMAF128): Likewise.
10429 * doc/cpp.texi (Common Predefined Macros): Document defining
10430 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10431 fma _Float<N> and _Float<N>X variants.
10433 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10436 * config/i386/i386-modes.def (CCFPU): Remove definition.
10437 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10438 (ix86_cc_modes_compatible): Ditto.
10439 (ix86_expand_carry_flag_compare): Ditto.
10440 (ix86_expand_int_movcc): Ditto.
10441 (ix86_expand_int_addcc): Ditto.
10442 (ix86_reverse_condition): Ditto.
10443 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10444 Return true/false for unordered/ordered fp comparisons.
10445 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10446 (ix86_prepare_fp_compare_args): Update for rename.
10447 (ix86_expand_fp_compare): Update for rename. Generate unordered
10448 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10449 (ix86_expand_sse_compare_and_jump): Ditto.
10450 * config/i386/predicates.md (fcmov_comparison_operator):
10451 Remove CCFPU mode handling.
10452 (ix86_comparison_operator): Ditto.
10453 (ix86_carry_flag_operator): Ditto.
10454 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10455 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10456 (*cmpu<mode>_cc_i387): Ditto.
10457 (FPCMP): Remove mode iterator.
10458 (unord): Remove mode attribute.
10459 (unord_subst): New define_subst transformation
10460 (unord): New define_subst attribute.
10461 (unordered): Ditto.
10462 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10463 (*cmpi<unord>xf_i387): Ditto.
10464 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10465 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10466 using unord_subst transformation.
10467 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10468 (round_saeonly): Also handle CCFP mode.
10469 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10470 Remove UNSPEC_SAHF unspec handling.
10472 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10474 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10476 2017-10-27 Jeff Law <law@redhat.com>
10478 * gimple-ssa-sprintf.c: Include domwalk.h.
10479 (class sprintf_dom_walker): New class, derived from dom_walker.
10480 (sprintf_dom_walker::before_dom_children): New function.
10481 (struct call_info): Moved into sprintf_dom_walker class
10482 (compute_formath_length, handle_gimple_call): Likewise.
10483 (sprintf_length::execute): Call the dominator walker rather
10484 than walking the statements.
10486 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10487 gimple statement locations.
10488 (check_array_bounds): Corresponding changes. Get the statement's
10489 location directly from wi->stmt.
10491 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10494 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10496 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10498 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10499 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10501 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10504 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10505 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10506 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10507 maybe_get_pool_constant.
10508 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10511 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10513 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10514 2.26 caveat. Update gas and gld versions.
10515 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10518 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10520 * cgraph.h (set_malloc_flag): Declare.
10521 * cgraph.c (set_malloc_flag_1): New function.
10522 (set_malloc_flag): Likewise.
10523 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10524 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10526 (read_ipa_call_summary): Add support for reading is_return_callee.
10527 (write_ipa_call_summary): Stream is_return_callee.
10528 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10529 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10530 ipa-prop.h, ipa-fnsummary.h.
10531 (pure_const_names): Change to static.
10532 (malloc_state_e): Define.
10533 (malloc_state_names): Define.
10534 (funct_state_d): Add field malloc_state.
10535 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10536 (check_retval_uses): New function.
10537 (malloc_candidate_p): Likewise.
10538 (analyze_function): Add support for malloc attribute.
10539 (pure_const_write_summary): Stream malloc_state.
10540 (pure_const_read_summary): Add support for reading malloc_state.
10541 (dump_malloc_lattice): New function.
10542 (propagate_malloc): New function.
10543 (warn_function_malloc): New function.
10544 (ipa_pure_const::execute): Call propagate_malloc and
10545 ipa_free_fn_summary.
10546 (pass_local_pure_const::execute): Add support for malloc attribute.
10547 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10548 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10550 2017-10-27 Martin Liska <mliska@suse.cz>
10552 PR gcov-profile/82457
10553 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10554 for fork-like functions to be properly instrumented.
10556 2017-10-27 Richard Biener <rguenther@suse.de>
10558 PR middle-end/81659
10559 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10560 info when we redirected EH.
10562 2017-10-26 Michael Collison <michael.collison@arm.com>
10564 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10566 (<optab>_trunchf<GPI:mode>2: New pattern.
10567 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10568 * config/aarch64/iterators.md (wv): New mode attribute.
10569 (vf, VF): New mode attributes.
10570 (vgp, VGP): New mode attributes.
10571 (s): Update attribute with SImode and DImode prefixes.
10573 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10575 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10576 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10577 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10578 (nios2_option_overide): Initialize it. Don't allow R0-relative
10579 addressing with PIC.
10580 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10581 (nios2_symbolic_constant_p): Likewise.
10582 (nios2_legitimate_address_p): Likewise.
10583 (nios2_r0rel_section_name_p): New.
10584 (nios2_symbol_ref_in_r0rel_data_p): New.
10585 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10586 (r0rel_constant_p): New.
10587 (nios2_print_operand_address): Handle r0rel_constant_p.
10588 (nios2_cdx_narrow_form_p): Likewise.
10589 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10590 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10591 (Nios II Options): Document -mr0rel-sec.
10593 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10595 * config/nios2/nios2.c: Include xregex.h.
10596 (nios2_gprel_sec_regex): New.
10597 (nios2_option_overide): Initialize it. Don't allow GP-relative
10598 addressing with PIC.
10599 (nios2_small_section_name_p): Check for regex match.
10600 * config/nios2/nios2.opt (mgprel-sec=): New option.
10601 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10602 (Nios II Options): Document -mgprel-sec.
10604 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10606 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10608 2017-10-26 Tom de Vries <tom@codesourcery.com>
10610 PR tree-optimization/82707
10611 * gimple.c (gimple_copy): Fix unsharing of
10612 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10614 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10616 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10617 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10618 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10619 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10620 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10621 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10622 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10623 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10624 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10625 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10626 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10627 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10628 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10629 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10630 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10631 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10632 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10634 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10636 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10638 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10639 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10640 warning to rs6000.c. Remove the Undocumented flag, since it has
10642 (-mabi=ibmlongdouble): Likewise.
10643 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10644 already set, set the default format for long double.
10645 (rs6000_debug_reg_global): Print whether long double is IBM or
10647 (rs6000_option_override_internal): Rework setting long double
10648 format. Only warn if the user is changing the long double default
10649 and they did not use -Wno-psabi.
10650 * doc/invoke.texi (PowerPC options): Update the documentation for
10651 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10653 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10654 Alan Hayward <alan.hayward@arm.com>
10655 David Sherwood <david.sherwood@arm.com>
10657 * rtl.h (wider_subreg_mode): New function.
10658 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10659 rather than an unsigned int *.
10660 * ira-color.c (regno_max_ref_width): Replace with...
10661 (regno_max_ref_mode): ...this new variable.
10662 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10663 Use wider_subreg_mode.
10664 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10665 rather than an unsigned int *.
10666 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10667 (process_alt_operands): Likewise.
10668 (invariant_p): Likewise.
10669 * lra-spills.c (assign_mem_slot): Likewise.
10670 (add_pseudo_to_slot): Likewise.
10671 * lra.c (collect_non_operand_hard_regs): Likewise.
10672 (add_regs_to_insn_regno_info): Likewise.
10673 * reload1.c (regno_max_ref_width): Replace with...
10674 (regno_max_ref_mode): ...this new variable.
10675 (reload): Update accordingly. Update call to
10676 ira_sort_regnos_for_alter_reg.
10677 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10678 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10679 (scan_paradoxical_subregs): Likewise.
10681 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10683 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10684 (aarch64_frame): Add emit_frame_chain boolean.
10685 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10686 Move eh_return case to aarch64_layout_frame.
10687 (aarch64_layout_frame): Initialize emit_frame_chain.
10688 (aarch64_expand_prologue): Use emit_frame_chain.
10690 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10692 * config/aarch64/aarch64.c (aarch64_layout_frame):
10693 Ensure LR is always stored at the bottom of the callee-saves.
10694 Remove rarely used frame layout which saves callee-saves at top of
10695 frame, so the store of LR can be used as a valid probe in all cases.
10697 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10699 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10700 Improve unaligned TImode/TFmode base/offset split.
10702 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10703 Alan Hayward <alan.hayward@arm.com>
10704 David Sherwood <david.sherwood@arm.com>
10706 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10707 * combine.c (find_single_use_1): Likewise.
10708 (expand_field_assignment): Likewise.
10709 (move_deaths): Likewise.
10710 * lra-constraints.c (simplify_operand_subreg): Likewise.
10711 (curr_insn_transform): Likewise.
10712 * lra.c (collect_non_operand_hard_regs): Likewise.
10713 (add_regs_to_insn_regno_info): Likewise.
10714 * rtlanal.c (reg_referenced_p): Likewise.
10715 (covers_regno_no_parallel_p): Likewise.
10717 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10719 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10720 Don't print any bits outside the precision of the value.
10721 * wide-int.cc (test_printing): Add some new tests.
10723 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10725 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10726 supports -xbrace_comment option.
10727 * configure: Regenerate.
10728 * config.in: Regenerate.
10729 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10730 (ASM_CPU_SPEC): Use it.
10732 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10734 * target.def (static_rtx_alignment): New hook.
10735 * targhooks.h (default_static_rtx_alignment): Declare.
10736 * targhooks.c (default_static_rtx_alignment): New function.
10737 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10738 * doc/tm.texi: Regenerate.
10739 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10740 instead of targetm.constant_alignment. Remove call to
10741 set_mem_attributes.
10742 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10743 (cris_preferred_mininum_alignment): New function, split out from...
10744 (cris_constant_alignment): ...here.
10745 (cris_static_rtx_alignment): New function.
10746 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10748 (ix86_constant_alignment): ...here.
10749 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10750 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10751 (mmix_static_rtx_alignment): New function.
10752 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10753 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10755 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10758 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10759 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10761 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10763 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10764 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10765 mark as a sign-extending load.
10766 (local_pic_load_u): Define.
10768 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10770 PR middle-end/82062
10771 * fold-const.c (operand_equal_for_comparison_p): Also return true
10772 if ARG0 is a simple variant of ARG1 with narrower precision.
10773 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10775 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10777 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10779 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10780 scatter_static, scatter_per_elt.
10781 * x86-tune-costs.h: Add new cost entries.
10783 2017-10-25 Richard Biener <rguenther@suse.de>
10785 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10786 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10787 class pass_fre): Move to ...
10788 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10789 class pass_fre): ... here and adjust for statistics.
10791 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10794 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10795 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10796 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10797 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10800 2017-10-25 Richard Biener <rguenther@suse.de>
10802 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10803 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10804 eliminate_push_avail, eliminate_insert): Move inside...
10805 (class eliminate_dom_walker): ... this class in preparation
10807 (fini_eliminate): Remove by merging with ...
10808 (eliminate): ... this function. Adjust for class changes.
10809 (pass_pre::execute): Remove fini_eliminate call.
10810 (pass_fre::execute): Likewise.
10812 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10815 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10816 (VPERMI2, VPERMI2I): New mode iterators.
10817 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10818 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10820 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10821 mode iterator. Remove 3 old define_insn patterns.
10822 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10823 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10824 VPERMI2 mode iterator, remove the other two expanders.
10825 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10826 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10827 instructions, remove the other two patterns.
10828 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10829 mode iterator, remove the other two patterns.
10830 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10831 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10832 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10833 and adjust argument order accordingly.
10834 (ix86_expand_vec_perm): Adjust caller.
10835 (expand_vec_perm_1): Likewise.
10836 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10837 (expand_vec_perm_vpermt2_vpshub2): ... this.
10838 (ix86_expand_vec_perm_const_1): Adjust caller.
10839 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10842 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10843 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10844 (vec_shl_<mode>): Remove unused expander.
10845 (avx512bw_<shift_insn><mode>3): New define_insn.
10846 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10847 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10849 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10852 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10855 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10857 PR rtl-optimization/82396
10858 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10859 (autopref_multipass_init): Simplify initialization.
10860 (autopref_rank_data): Simplify sort order.
10861 * gcc/sched-int.h (autopref_multipass_data_): Remove
10862 multi_mem_insn_p, min_offset and max_offset.
10864 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10866 PR middle-end/60580
10867 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10868 Check special value of flag_omit_frame_pointer.
10869 (aarch64_can_eliminate): Likewise.
10870 (aarch64_override_options_after_change_1): Simplify handling of
10871 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10873 2017-10-24 Richard Biener <rguenther@suse.de>
10875 PR tree-optimization/82697
10876 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10877 zero for conditional load and unconditional store.
10879 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10881 * doc/install.texi: Document bootstrap-cet.
10883 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10886 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10887 ENDBR instruction at function entrance if function is only
10890 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10893 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10894 patterns to better describe from which operation the CF is computed.
10895 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10896 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10897 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10898 is 0, use _0 suffixed expanders instead of emitting a comparison
10901 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10903 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10904 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10906 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10908 PR middle-end/82569
10909 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10910 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10911 * loop-iv.c (iv_get_reaching_def): Likewise.
10912 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10913 variable is promoted and the partition contains undefined values.
10915 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10917 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10919 (nios2_address_cost): Define.
10920 (nios2_legitimize_address): Recognize (exp + constant) directly.
10921 (TARGET_ADDRESS_COST): Define.
10923 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10925 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10926 (nios2_symbolic_memory_operand_p): Declare.
10927 (nios2_split_large_constant): Declare.
10928 (nios2_split_symbolic_memory_operand): Declare.
10929 * config/nios2/nios2.c: Adjust includes.
10930 (nios2_symbolic_constant_allowed): New.
10931 (nios2_symbolic_constant_p): New.
10932 (nios2_plus_symbolic_constant_p): New.
10933 (nios2_valid_addr_expr_p): Recognize addresses involving
10934 symbolic constants.
10935 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10936 (nios2_symbolic_memory_operand_p): New.
10937 (nios2_large_constant_p): New.
10938 (nios2_split_large_constant): New.
10939 (nios2_split_plus_large_constant): New.
10940 (nios2_split_symbolic_memory_operand): New.
10941 (nios2_legitimize_address): Code refactoring. Handle addresses
10942 involving symbolic constants.
10943 (nios2_emit_move_sequence): Likewise.
10944 (nios2_print_operand): Improve error output.
10945 (nios2_print_operand_address): Handle LO_SUM.
10946 (nios2_cdx_narrow_form_p): Likewise.
10947 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10948 operands involving symbolic constants.
10949 (movhi_internal, movsi_internal): Likewise.
10950 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10951 (extendhisi2, extendqi<mode>2): Likewise.
10953 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10955 * tree-pass.h (PROP_rtl_split_insns): Define.
10956 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10958 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10960 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10962 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10965 * target.def (const_not_ok_for_debug_p): Default to
10966 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10967 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10968 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10969 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10970 which targetm.const_not_ok_for_debug_p returned true.
10971 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10973 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10975 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10976 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10977 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10979 (ix86_const_not_ok_for_debug_p): New function.
10980 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10981 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10983 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10986 * system.h: Conditionally include "unique-ptr.h" if
10987 INCLUDE_UNIQUE_PTR is defined.
10988 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10989 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10991 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10993 * config/rl78/rl78.md: New define_expand "subdi3".
10995 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10998 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10999 DF_REF_INSN if DF_REF_INSN_INFO is false.
11001 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11003 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11004 xmm_move instead of sse_move.
11005 (sse_store_index): New function.
11006 (ix86_register_move_cost): Be more sensible about mismatch stall;
11007 model AVX moves correctly; make difference between sse->integer and
11009 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11010 moves; make difference between SSE and AVX.
11011 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11012 and zmm_move. Increase size of sse load and store tables;
11013 add unaligned load and store tables; add ssemmx_to_integer.
11014 * x86-tune-costs.h: Update all entries according to real
11015 move latencies from Agner Fog's manual and chip documentation.
11017 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11020 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11021 * config/i386/constraints.md (Wf): New constraint.
11022 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11023 (cmp<dwi>_doubleword): Removed.
11024 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11025 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11026 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11027 expand with cmp<dwi>_doubleword. For LTU and GEU use
11028 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11030 * common.opt (gcolumn-info): Enable by default.
11031 * doc/invoke.texi (gcolumn-info): Document new default.
11033 2017-10-23 Richard Biener <rguenther@suse.de>
11035 PR tree-optimization/82672
11036 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11037 Fold the stmt if we propagated into it.
11039 2017-10-23 Richard Biener <rguenther@suse.de>
11041 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11042 (bitmap_remove_expr_from_set): ... this. All callers call this
11043 for non-constant values.
11044 (bitmap_set_subtract): Rename to...
11045 (bitmap_set_subtract_expressions): ... this. Adjust and
11047 (bitmap_set_contains_value): Remove superfluous check.
11048 (bitmap_set_replace_value): Inline into single caller ...
11049 (bitmap_value_replace_in_set): ... here and simplify.
11050 (dependent_clean): Merge into ...
11051 (clean): ... this using an overload. Adjust.
11052 (prune_clobbered_mems): Adjust.
11053 (compute_antic_aux): Likewise.
11054 (compute_partial_antic_aux): Likewise.
11056 2017-10-23 Richard Biener <rguenther@suse.de>
11058 PR tree-optimization/82129
11060 2017-08-01 Richard Biener <rguenther@suse.de>
11062 PR tree-optimization/81181
11063 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11064 (compute_antic): ... end of iteration here.
11066 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11068 * target.def (starting_frame_offset): New hook.
11069 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11070 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11071 * doc/tm.texi.in: Regenerate.
11072 * hooks.h (hook_hwi_void_0): Declare.
11073 * hooks.c (hook_hwi_void_0): New function.
11074 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11075 STARTING_FRAME_OFFSET.
11076 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11077 * reload1.c (reload): Likewise.
11078 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11079 instead of STARTING_FRAME_OFFSET.
11080 * function.c (try_fit_stack_local): Likewise.
11081 (assign_stack_local_1): Likewise
11082 (instantiate_virtual_regs): Likewise.
11083 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11084 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11085 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11086 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11087 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11088 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11089 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11090 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11091 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11092 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11093 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11094 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11095 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11096 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11097 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11098 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11099 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11100 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11101 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11102 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11103 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11104 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11105 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11106 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11107 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11108 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11109 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11110 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11111 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11112 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11113 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11114 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11115 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11116 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11117 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11118 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11119 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11120 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11121 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11122 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11123 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11124 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11125 return a HOST_WIDE_INT.
11126 (avr_builtin_setjmp_frame_value): Use it instead of
11127 STARTING_FRAME_OFFSET.
11128 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11129 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11130 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11132 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11133 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11134 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11135 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11136 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11137 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11138 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11139 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11140 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11141 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11142 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11143 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11145 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11146 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11147 * config/mips/mips.c (mips_compute_frame_info): Refer to
11148 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11149 (mips_starting_frame_offset): New function.
11150 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11151 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11152 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11153 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11154 and return a HOST_WIDE_INT.
11155 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11156 (mmix_initial_elimination_offset): Refer to
11157 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11158 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11159 * config/pa/pa.c (pa_starting_frame_offset): New function.
11160 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11161 (pa_expand_prologue): Likewise.
11162 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11163 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11164 !FRAME_GROWS_DOWNWARD handling to...
11165 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11166 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11167 !FRAME_GROWS_DOWNWARD handling to...
11168 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11169 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11170 !FRAME_GROWS_DOWNWARD handling to...
11171 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11172 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11174 (rs6000_starting_frame_offset): New function.
11175 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11176 !FRAME_GROWS_DOWNWARD handling to...
11177 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11178 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11179 !FRAME_GROWS_DOWNWARD handling to...
11180 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11181 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11182 !FRAME_GROWS_DOWNWARD handling to...
11183 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11184 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11185 (rs6000_starting_frame_offset): New function.
11186 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11187 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11188 * config/vax/vax.c (vax_starting_frame_offset): New function.
11189 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11190 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11191 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11192 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11193 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11194 * system.h (STARTING_FRAME_OFFSET): Poison.
11196 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11198 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11199 SCALAR_TYPE_MODE instead of TYPE_MODE.
11201 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11202 Alan Hayward <alan.hayward@arm.com>
11203 David Sherwood <david.sherwood@arm.com>
11205 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11207 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11208 Alan Hayward <alan.hayward@arm.com>
11209 David Sherwood <david.sherwood@arm.com>
11211 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11213 2017-10-23 Richard Biener <rguenther@suse.de>
11215 PR tree-optimization/82129
11216 * tree-ssa-pre.c (bitmap_set_and): Remove.
11217 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11218 canonicalizing expressions in the set to those with lowest
11219 ID rather than taking that from the first edge.
11221 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11223 * combine.c (rtx_equal_for_field_assignment_p): Use
11224 byte_lowpart_offset.
11226 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11227 Alan Hayward <alan.hayward@arm.com>
11228 David Sherwood <david.sherwood@arm.com>
11230 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11231 to a SUBREG_PROMOTED_VAR.
11233 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11234 Alan Hayward <alan.hayward@arm.com>
11235 David Sherwood <david.sherwood@arm.com>
11237 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11238 (expand_debug_source_expr): Likewise.
11239 * combine.c (combine_simplify_rtx): Likewise.
11240 * cse.c (fold_rtx): Likewise.
11241 * optabs.c (expand_float): Likewise.
11242 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11243 (simplify_binary_operation_1): Likewise.
11245 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11246 Alan Hayward <alan.hayward@arm.com>
11247 David Sherwood <david.sherwood@arm.com>
11249 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11250 (record_promoted_value): Likewise.
11251 * expr.c (expand_expr_real_2): Likewise.
11252 * ree.c (update_reg_equal_equiv_notes): Likewise.
11253 (combine_set_extension): Likewise.
11254 * rtlanal.c (low_bitmask_len): Likewise.
11255 * simplify-rtx.c (neg_const_int): Likewise.
11256 (simplify_binary_operation_1): Likewise.
11258 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11259 Alan Hayward <alan.hayward@arm.com>
11260 David Sherwood <david.sherwood@arm.com>
11262 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11263 * regcprop.c (maybe_mode_change): Likewise.
11264 * reload1.c (alter_reg): Likewise.
11266 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11268 * inchash.h (inchash::hash::add_wide_int): New function.
11269 * lto-streamer-out.c (hash_tree): Use it.
11271 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11273 * inchash.h (inchash::hash::add_wide_int): Rename to...
11274 (inchash::hash::add_hwi): ...this.
11275 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11276 (polymorphic_call_target_hasher::hash): Likewise.
11277 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11278 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11279 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11280 * lto-streamer-out.c (hash_tree): Likewise.
11281 * optc-save-gen.awk: Likewise.
11282 * tree.c (add_expr): Likewise.
11284 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11287 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11288 for ordered inequality comparisons even with TARGET_IEEE_FP.
11290 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11293 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11294 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11295 Expand with cmp<dwi>_doubleword.
11297 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11299 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11300 List CET intrinsics.
11301 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11302 specific to -fcf-protection option.
11304 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11306 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11307 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11308 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11309 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11310 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11311 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11312 (extra_objs): Add cet.o for Linux/x86 targets.
11313 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11314 * config/i386/cet.c: New file.
11315 * config/i386/cetintrin.h: Likewise.
11316 * config/i386/t-cet: Likewise.
11317 * config/i386/cpuid.h (bit_SHSTK): New.
11318 (bit_IBT): Likewise.
11319 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11320 pass IBT and SHSTK bits.
11321 * config/i386/i386-builtin-types.def
11322 (VOID_FTYPE_UNSIGNED_PVOID): New.
11323 (VOID_FTYPE_UINT64_PVOID): Likewise.
11324 * config/i386/i386-builtin.def: Add CET intrinsics.
11325 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11326 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11327 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11328 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11330 * config/i386/i386.c (rest_of_insert_endbranch): New.
11331 (pass_data_insert_endbranch): Likewise.
11332 (pass_insert_endbranch): Likewise.
11333 (make_pass_insert_endbranch): Likewise.
11334 (ix86_notrack_prefixed_insn_p): Likewise.
11335 (ix86_target_string): Add -mibt, -mshstk flags.
11336 (ix86_option_override_internal): Add flag_cf_protection
11338 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11339 (ix86_print_operand): Add 'notrack' prefix output.
11340 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11341 (ix86_expand_builtin): Expand CET intrinsics.
11342 (x86_output_mi_thunk): Add 'endbranch' instruction.
11343 * config/i386/i386.h (TARGET_IBT): New.
11344 (TARGET_IBT_P): Likewise.
11345 (TARGET_SHSTK): Likewise.
11346 (TARGET_SHSTK_P): Likewise.
11347 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11348 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11349 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11350 (builtin_setjmp_setup): New pattern.
11351 (builtin_longjmp): Likewise.
11352 (rdssp<mode>): Likewise.
11353 (incssp<mode>): Likewise.
11354 (saveprevssp): Likewise.
11355 (rstorssp): Likewise.
11356 (wrss<mode>): Likewise.
11357 (wruss<mode>): Likewise.
11358 (setssbsy): Likewise.
11359 (clrssbsy): Likewise.
11360 (nop_endbr): Likewise.
11361 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11363 * config/i386/immintrin.h: Include <cetintrin.h>.
11364 * config/i386/linux-common.h
11365 (file_end_indicate_exec_stack_and_cet): New prototype.
11366 (TARGET_ASM_FILE_END): New.
11368 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11370 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11371 latencies instead of having separate table; make difference between
11372 integer and float costs.
11373 * i386.h (processor_costs): Remove scalar_stmt_cost,
11374 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11375 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11377 * x86-tune-costs.h: Remove entries which has been removed in
11378 procesor_costs from all tables; make cond_taken_branch_cost
11379 and cond_not_taken_branch_cost COST_N_INSNS based.
11381 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11383 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11385 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11387 * config/i386/i386.md (isa): Remove fma_avx512f.
11388 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11389 <avx512>_fmadd_<mode>_mask3<round_name>,
11390 <avx512>_fmsub_<mode>_mask<round_name>,
11391 <avx512>_fmsub_<mode>_mask3<round_name>,
11392 <avx512>_fnmadd_<mode>_mask<round_name>,
11393 <avx512>_fnmadd_<mode>_mask3<round_name>,
11394 <avx512>_fnmsub_<mode>_mask<round_name>,
11395 <avx512>_fnmsub_<mode>_mask3<round_name>,
11396 <avx512>_fmaddsub_<mode>_mask<round_name>,
11397 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11398 <avx512>_fmsubadd_<mode>_mask<round_name>,
11399 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11400 (*vec_widen_umult_even_v16si<mask_name>,
11401 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11402 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11404 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11406 * extend.texi: Add 'nocf_check' documentation.
11407 * gimple.texi: Add second parameter to
11408 gimple_build_call_from_tree.
11409 * invoke.texi: Add -fcf-protection documentation.
11410 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11412 2017-10-20 Richard Biener <rguenther@suse.de>
11414 PR tree-optimization/82473
11415 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11416 the largest input type.
11418 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11420 * c-attribs.c (handle_nocf_check_attribute): New function.
11421 (c_common_attribute_table): Add 'nocf_check' handling.
11422 * gimple-parser.c: Add second argument NULL to
11423 gimple_build_call_from_tree.
11424 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11425 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11427 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11428 * common.opt: Add fcf-protection flag.
11429 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11430 * flag-types.h: Add enum cf_protection_level.
11431 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11432 Add 'nocf_check' attribute propagation to gimple call.
11433 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11434 (gimple_build_call_from_tree): Update prototype.
11435 (gimple_call_nocf_check_p): New function.
11436 (gimple_call_set_nocf_check): Likewise.
11437 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11438 * ipa-icf.c: Add nocf_check attribute in statement hash.
11439 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11440 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11441 * toplev.c (process_options): Add flag_cf_protection handling.
11443 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11445 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11447 2017-10-20 Richard Biener <rguenther@suse.de>
11449 PR tree-optimization/82603
11450 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11451 remove false predicated stores.
11453 2017-10-20 Richard Biener <rguenther@suse.de>
11455 * graphite-isl-ast-to-gimple.c
11456 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11457 Remove return value and simplify, dump copied stmt after lhs
11459 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11460 Reduce dump verbosity.
11461 (gsi_insert_earliest): Likewise.
11462 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11463 * graphite.c (print_global_statistics): Adjust dumping.
11464 (print_graphite_scop_statistics): Likewise.
11465 (print_graphite_statistics): Do not dump loops here.
11466 (graphite_transform_loops): But here.
11468 2017-10-20 Nicolas Roche <roche@adacore.com>
11470 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11471 * configure: Regenerate.
11473 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11476 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11477 functions when optimizing replace GIMPLE_RETURN stmts with
11478 calls to __builtin_unreachable ().
11481 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11482 for -fsanitize=thread link of executables.
11483 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11484 link of executables.
11487 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11488 New mode iterators.
11489 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11490 define_insns for logical vector shifts to use VI248_AVX512BW
11491 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11492 condition, useless isa and prefix attributes. Change the first
11493 2 of these define_insns to ...
11494 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11495 define_insn for avx512vl.
11496 (<shift_insn><mode>3): ... and this, new define_insn without
11497 masking for non-avx512vl.
11500 * config/i386/sse.md (*andnot<mode>3,
11501 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11502 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11503 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11504 not applied use empty suffix even for TARGET_AVX512VL.
11505 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11506 is applied, supply evex,evex or evex,evex,evex instead of just
11509 2017-10-20 Julia Koval <julia.koval@intel.com>
11511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11512 (OPTION_MASK_ISA_GFNI_UNSET): New.
11513 (ix86_handle_option): Handle OPT_mgfni.
11514 * config/i386/cpuid.h (bit_GFNI): New.
11515 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11516 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11517 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11518 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11519 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11520 * config/i386/i386.opt: Add mgfni.
11522 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11524 * config/msp430/msp430.c (msp430_option_override): Disable
11525 -fdelete-null-pointer-checks.
11526 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11528 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11530 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11531 of x87 and SSE instructions.
11533 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11535 * asan.c (create_cond_insert_point): Do not update edge count.
11536 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11537 (afdo_propagate_circuit): Likewise.
11538 (afdo_calculate_branch_prob): Likewise.
11539 (afdo_annotate_cfg): Likewise.
11540 * basic-block.h (struct edge_def): Remove count.
11541 (edge_def::count): New accessor.
11542 * bb-reorder.c (rotate_loop): Update.
11543 (find_traces_1_round): Update.
11544 (connect_traces): Update.
11545 (sanitize_hot_paths): Update.
11546 * cfg.c (unchecked_make_edge): Update.
11547 (make_single_succ_edge): Update.
11548 (check_bb_profile): Update.
11549 (dump_edge_info): Update.
11550 (update_bb_profile_for_threading): Update.
11551 (scale_bbs_frequencies_int): Update.
11552 (scale_bbs_frequencies_gcov_type): Update.
11553 (scale_bbs_frequencies_profile_count): Update.
11554 (scale_bbs_frequencies): Update.
11555 * cfganal.c (connect_infinite_loops_to_exit): Update.
11556 * cfgbuild.c (compute_outgoing_frequencies): Update.
11557 (find_many_sub_basic_blocks): Update.
11558 * cfgcleanup.c (try_forward_edges): Update.
11559 (try_crossjump_to_edge): Update
11560 * cfgexpand.c (expand_gimple_cond): Update
11561 (expand_gimple_tailcall): Update
11562 (construct_exit_block): Update
11563 * cfghooks.c (verify_flow_info): Update
11564 (redirect_edge_succ_nodup): Update
11565 (split_edge): Update
11566 (make_forwarder_block): Update
11567 (duplicate_block): Update
11568 (account_profile_record): Update
11569 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11570 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11571 * cfgloopmanip.c (scale_loop_profile): Update.
11573 (lv_adjust_loop_entry_edge): Update.
11574 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11575 (force_nonfallthru_and_redirect): Update.
11576 (purge_dead_edges): Update.
11577 (rtl_flow_call_edges_add): Update.
11578 * cgraphunit.c (init_lowered_empty_function): Update.
11579 (cgraph_node::expand_thunk): Update.
11580 * gimple-pretty-print.c (dump_probability): Update.
11581 (dump_edge_probability): Update.
11582 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11583 * haifa-sched.c (sched_create_recovery_edges): Update.
11584 * hsa-gen.c (convert_switch_statements): Update.
11585 * ifcvt.c (dead_or_predicable): Update.
11586 * ipa-inline-transform.c (inline_transform): Update.
11587 * ipa-split.c (split_function): Update.
11588 * ipa-utils.c (ipa_merge_profiles): Update.
11589 * loop-doloop.c (add_test): Update.
11590 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11591 * lto-streamer-in.c (input_cfg): Update.
11592 (input_function): Update.
11593 * lto-streamer-out.c (output_cfg): Update.
11594 * modulo-sched.c (sms_schedule): Update.
11595 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11596 * predict.c (maybe_hot_edge_p): Update.
11597 (unlikely_executed_edge_p): Update.
11598 (probably_never_executed_edge_p): Update.
11599 (dump_prediction): Update.
11600 (drop_profile): Update.
11601 (propagate_unlikely_bbs_forward): Update.
11602 (determine_unlikely_bbs): Update.
11603 (force_edge_cold): Update.
11604 * profile.c (compute_branch_probabilities): Update.
11605 * reg-stack.c (better_edge): Update.
11606 * shrink-wrap.c (handle_simple_exit): Update.
11607 * tracer.c (better_p): Update.
11608 * trans-mem.c (expand_transaction): Update.
11609 (split_bb_make_tm_edge): Update.
11610 * tree-call-cdce.c: Update.
11611 * tree-cfg.c (gimple_find_sub_bbs): Update.
11612 (gimple_split_edge): Update.
11613 (gimple_duplicate_sese_region): Update.
11614 (gimple_duplicate_sese_tail): Update.
11615 (gimple_flow_call_edges_add): Update.
11616 (insert_cond_bb): Update.
11617 (execute_fixup_cfg): Update.
11618 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11619 * tree-complex.c (expand_complex_div_wide): Update.
11620 * tree-eh.c (lower_resx): Update.
11621 (unsplit_eh): Update.
11622 (cleanup_empty_eh_move_lp): Update.
11623 * tree-inline.c (copy_edges_for_bb): Update.
11624 (freqs_to_counts): Update.
11625 (copy_cfg_body): Update.
11626 * tree-ssa-dce.c (remove_dead_stmt): Update.
11627 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11628 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11629 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11630 (unloop_loops): Update.
11631 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11632 * tree-ssa-loop-split.c (connect_loops): Update.
11633 (split_loop): Update.
11634 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11635 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11636 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11637 * tree-ssa-reassoc.c (branch_fixup): Update.
11638 * tree-ssa-tail-merge.c (replace_block_by): Update.
11639 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11640 (compute_path_counts): Update.
11641 (update_profile): Update.
11642 (recompute_probabilities): Update.
11643 (update_joiner_offpath_counts): Update.
11644 (estimated_freqs_path): Update.
11645 (freqs_to_counts_path): Update.
11646 (clear_counts_path): Update.
11647 (ssa_fix_duplicate_block_edges): Update.
11648 (duplicate_thread_path): Update.
11649 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11650 (case_bit_test_cmp): Update.
11651 (collect_switch_conv_info): Update.
11652 (gen_inbound_check): Update.
11653 (do_jump_if_equal): Update.
11654 (emit_cmp_and_jump_insns): Update.
11655 * tree-tailcall.c (decrease_profile): Update.
11656 (eliminate_tail_call): Update.
11657 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11658 (vect_do_peeling): Update.
11659 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11660 * ubsan.c (ubsan_expand_null_ifn): Update.
11661 (ubsan_expand_ptr_ifn): Update.
11662 * value-prof.c (gimple_divmod_fixed_value): Update.
11663 (gimple_mod_pow2): Update.
11664 (gimple_mod_subtract): Update.
11665 (gimple_ic): Update.
11666 (gimple_stringop_fixed_value): Update.
11668 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11671 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11673 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11675 PR rtl-optimization/82395
11676 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11677 based on non_spilled_static_chain_regno_p.
11679 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11681 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11682 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11683 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11685 2017-10-19 Martin Sebor <msebor@redhat.com>
11687 PR tree-optimization/82596
11688 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11690 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11692 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11693 (get_mem_refs_of_builtin_call): Likewise.
11694 * builtins.c (expand_builtin_apply): Adjust call to
11695 allocate_dynamic_stack_space.
11696 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11697 the third argument to allocate_dynamic_stack_space, otherwise -1.
11698 (expand_builtin): Deal with all alloca variants.
11699 (is_inexpensive_builtin): Likewise.
11700 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11701 * calls.c (special_function_p): Deal with all alloca variants.
11702 (initialize_argument_information): Adjust call to
11703 allocate_dynamic_stack_space.
11704 (expand_call): Likewise.
11705 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11706 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11707 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11708 use it for the stack usage computation.
11709 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11710 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11711 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11712 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11713 (in_loop_p): Remove first argument and useless check.
11714 (pass_walloca::execute): Remove useless test and adjust call to above.
11715 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11716 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11717 (gimplify_call_expr): Deal with all alloca variants.
11718 * hsa-gen.c (gen_hsa_alloca): Likewise.
11719 (gen_hsa_insns_for_call): Likewise.
11720 * ipa-pure-const.c (special_builtin_state): Likewise.
11721 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11722 * tree-object-size.c (alloc_object_size): Likewise.
11723 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11724 (call_may_clobber_ref_p_1): Likewise.
11725 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11726 (ccp_fold_stmt): Likewise.
11727 (optimize_stack_restore): Likewise.
11728 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11729 (mark_all_reaching_defs_necessary_1): Likewise.
11730 (propagate_necessity): Likewise.
11731 (eliminate_unnecessary_stmts): Likewise.
11732 * tree.c (build_common_builtin_nodes): Build
11733 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11734 (build_alloca_call_expr): New function.
11735 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11736 (CASE_BUILT_IN_ALLOCA): Likewise.
11737 (build_alloca_call_expr): Declare.
11738 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11740 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11743 * dwarf2out.c (new_die_raw): New static inline function.
11744 (new_die): Use it to create the DIE.
11745 (add_AT_external_die_ref): Likewise.
11746 (clone_die): Likewise.
11747 (clone_as_declaration): Likewise.
11748 (dwarf2out_vms_debug_main_pointer): Likewise.
11749 (base_type_die): Likewise. Remove early return for corner cases.
11750 Do not call add_pubtype on the DIE here.
11751 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11752 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11753 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11754 native order exists for base types, attach the DIE manually and call
11755 add_pubtype on it. Do not equate a reverse order DIE to the type.
11757 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11759 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11760 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11762 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11764 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11766 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11767 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11768 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11769 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11770 * opts.c (sanitizer_opts): Add builtin.
11771 * ubsan.c (instrument_builtin): New function.
11772 (pass_ubsan::execute): Call it.
11773 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11774 * doc/invoke.texi: Document -fsanitize=builtin.
11776 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11777 builtins, store max (log2 (align), 0) into uchar field instead of
11778 align into uptr field.
11779 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11780 store uchar 0 field instead of uptr 0 field.
11781 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11782 instead of passing one address of struct with 2 locations pass
11783 two addresses of structs with 1 location each.
11784 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11785 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11786 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11787 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11788 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11789 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11790 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11791 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11793 2017-10-19 Martin Liska <mliska@suse.cz>
11796 * file-find.c (remove_prefix): Remove.
11797 * file-find.h (remove_prefix): Likewise.
11798 * gcc-ar.c: Remove smartness of lookup.
11800 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11802 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11803 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11804 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11806 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11809 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11810 (setcc + and to xor + setcc): New peephole2.
11812 2017-10-19 Tom de Vries <tom@codesourcery.com>
11814 * doc/sourcebuild.texi (Test Directives, Variants of
11815 dg-require-support): Add dg-require-stack-size.
11817 2017-10-19 Martin Liska <mliska@suse.cz>
11820 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11821 that have a large alignment.
11822 (gimplify_target_expr): Likewise.
11824 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11826 PR rtl-optimization/82602
11827 * ira.c (rtx_moveable_p): Return false for volatile asm.
11829 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11832 * config/i386/i386-modes.def (CCGZ): New CC mode.
11833 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11834 * config/i386/predicates.md (ix86_comparison_operator):
11836 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11837 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11838 with double-word subtraction.
11839 (put_condition_code): Handle CCGZmode.
11841 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11843 * wide-int.cc (debug (const wide_int &)): New.
11844 (debug (const wide_int *)): New.
11845 (debug (const widest_int &)): New.
11846 (debug (const widest_int *)): New.
11848 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11850 PR middle-end/82556
11851 * lra-constraints.c (curr_insn_transform): Use non-input operand
11852 instead of output one for matched reload.
11854 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11856 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11857 (tree-ssa-loop-ivopts.h): New header file.
11858 (struct builtin_info): New fields.
11859 (classify_builtin_1): Compute and record base and offset parts for
11860 memset builtin partition by calling strip_offset.
11861 (offset_cmp, fuse_memset_builtins): New functions.
11862 (finalize_partitions): Fuse adjacent memset partitions by calling
11864 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11865 Expose the interface.
11866 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11868 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11870 PR tree-optimization/82574
11871 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11872 that data reference must be executed exactly once per iteration
11873 against the outermost loop in nest.
11874 (classify_partition): Update call to above function.
11876 2017-10-18 Richard Biener <rguenther@suse.de>
11878 PR tree-optimization/82591
11879 * graphite.c (graphite_transform_loops): Move code gen message
11881 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11882 Here. Handle scop_to_isl_ast failing.
11883 (scop_to_isl_ast): Limit the number of ISL operations.
11885 2017-10-18 Richard Biener <rguenther@suse.de>
11887 * graphite-isl-ast-to-gimple.c
11888 (translate_isl_ast_to_gimple::set_rename): Simplify.
11889 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11890 (graphite_copy_stmts_from_block): ... here.
11891 (copy_bb_and_scalar_dependences): Simplify.
11892 (add_parameters_to_ivs_params): Canonicalize.
11893 (generate_entry_out_of_ssa_copies): Simplify.
11894 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11895 by passing in ISL dimension.
11896 (parameter_index_in_region_1): Rename to ...
11897 (parameter_index_in_region): ... this.
11898 (extract_affine): Adjust assert, pass down parameter index.
11899 (add_param_constraints): Use range-info when available.
11900 (build_scop_context): Adjust.
11901 * sese.c (new_sese_info): Adjust.
11902 (free_sese_info): Likewise.
11903 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11904 Remove unused typedefs.
11905 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11907 2017-10-18 Martin Liska <mliska@suse.cz>
11909 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11911 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11913 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11914 (s390_sched_init): Do not reset s390_sched_state if we entered the
11915 current basic block via a fallthru edge and all others are unlikely.
11917 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11919 * config/s390/s390.c (NUM_SIDES): New variable.
11920 (LONGRUNNING_THRESHOLD): New variable.
11921 (LATENCY_FACTOR): New variable.
11922 (s390_sched_score): Decrease score for long-running instructions on
11924 (s390_sched_variable_issue): Perform bookkeeping for long-running
11927 2017-10-18 Richard Biener <rguenther@suse.de>
11929 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11930 Simplify with removal of the parameter rename map.
11931 (set_rename): Likewise.
11932 (should_copy_to_new_region): Likewise.
11933 (graphite_copy_stmts_from_block): Likewise.
11934 (copy_bb_and_scalar_dependences): Remove initialization of
11935 unused copied_bb_map.
11936 (copy_def): Remove.
11937 (copy_internal_parameters): Likewise.
11938 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11939 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11940 Use INTEGRAL_TYPE_P.
11941 (parameter_index_in_region_1): Rename to ...
11942 (assign_parameter_index_in_region): ... this. Assert we have
11943 a parameter we handle.
11944 (scan_tree_for_params): Adjust.
11945 * sese.h (parameter_rename_map_t): Remove.
11946 (struct sese_info_t): Remove unused parameter_rename_map and
11947 copied_bb_map members.
11948 * sese.c (new_sese_info): Adjust.
11949 (free_sese_info): Likewise.
11951 2017-10-18 Martin Liska <mliska@suse.cz>
11954 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11955 on an abnormal edge.
11957 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11959 * doc/invoke.texi (ffunction-sections and fdata-sections):
11962 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11964 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11965 the use statement can throw internally.
11967 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11969 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11970 any RTX present on the RHS of a SET.
11971 * compare-elim.c (try_eliminate_compare): Restore comment.
11973 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11975 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11976 may be also called on tcc_type nodes.
11977 * langhooks.c (lhd_tree_size): Likewise.
11979 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11981 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11982 format_warning_at_substring.
11983 (maybe_warn): Convert source_range * param to a location_t. Pass
11984 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11985 (format_directive): Remove code to extract source_ranges and
11986 source_range * in favor of just a location_t.
11987 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11989 * substring-locations.c (format_warning_va): Convert
11990 source_range * param to a location_t.
11991 (format_warning_at_substring): Likewise.
11992 * substring-locations.h (format_warning_va): Likewise.
11993 (format_warning_at_substring): Likewise.
11995 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11997 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11999 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12000 and scatter/gather ops.
12002 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12003 vec_gather_load and vec_scatter_store.
12004 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12005 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12006 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12007 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12008 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12009 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12011 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12013 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12014 Detect FTST instruction and handle its register pops. Only pop
12015 second operand if can_pop_second_op is true.
12016 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12017 set can_pop_second_op to false in the compare_for_stack_reg call.
12019 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12020 output_fp_compare for stack register operands.
12021 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12022 instructions here. Do not emit stack register pops here. Assert
12023 that FCOMPP pops next to top stack register. Rewrite function.
12025 2017-10-17 Nathan Sidwell <nathan@acm.org>
12027 PR middle-end/82577
12028 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12029 use DECL_ASSEMBLER_NAME_RAW.
12031 PR middle-end/82546
12032 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12035 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12036 Wilco Dijkstra <wilco.dijkstra@arm.com>
12038 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12039 converstion to Pmode from the buf_addr.
12041 2017-10-17 Richard Biener <rguenther@suse.de>
12043 * graphite-dependences.c (scop_get_reads_and_writes): Change
12044 output parameters to references.
12046 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12048 PR 71026/tree-optimization
12049 * fold-const.c (distribute_real_division): Removed.
12050 (fold_binary_loc): Remove calls to distribute_real_divison.
12052 2017-10-17 Richard Biener <rguenther@suse.de>
12054 * graphite-scop-detection.c
12055 (scop_detection::stmt_has_simple_data_refs_p): Always use
12056 the full nest as region.
12057 (try_generate_gimple_bb): Likewise.
12058 * sese.c (scalar_evolution_in_region): Simplify now that
12059 SCEV can handle instantiation in regions.
12060 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12061 in the non-loop part of a function if requested.
12063 2017-10-17 Richard Biener <rguenther@suse.de>
12065 PR tree-optimization/82563
12066 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12068 (graphite_regenerate_ast_isl): Call it.
12069 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12071 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12073 PR tree-optimization/82549
12074 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12075 Formatting fixes. Instead of calling make_bit_field_ref with negative
12078 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12080 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12081 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12082 _mm_maskz_reduce_ss): New.
12083 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12084 __builtin_ia32_reducess_mask): Ditto..
12085 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12086 * config/i386/sse.md (reduces<mode>): Renamed to ...
12087 (reduces<mode><mask_scalar_name>): ... this.
12088 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12090 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12091 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12093 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12095 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12096 * selftest-run-tests.c (selftest::run_tests): Call
12097 selftest::unique_ptr_tests_cc_tests.
12098 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12099 * unique-ptr-tests.cc: New file.
12101 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12104 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12106 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12107 (make_hard_regno_born, make_hard_regno_dead): Update
12108 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12110 2017-10-16 Jeff Law <law@redhat.com>
12112 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12114 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12116 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12118 2017-10-16 Olivier Hainque <hainque@adacore.com>
12120 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12121 with_cpu if we were configured for an e500v2 target cpu name.
12123 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12125 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12126 * doc/invoke.texi: Document +nodsp as a valid extension for
12129 2017-10-16 Martin Liska <mliska@suse.cz>
12131 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12132 (test_set_range): Likewise.
12133 (test_range_functions): Rename to ...
12134 (test_bit_in_range): ... this.
12135 (sbitmap_c_tests): Add new test.
12137 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12139 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12141 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12142 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12144 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12146 * config/aarch64/aarch64-builtins.c
12147 (aarch64_types_quadopu_lane_qualifiers): New.
12148 (TYPES_QUADOPU_LANE): New.
12149 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12150 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12151 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12152 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12153 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12154 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12155 (Vdottype, DOTPROD): New.
12156 (sur): Add SDOT and UDOT.
12158 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12160 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12161 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12162 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12163 Add TARGET_DOTPROD.
12164 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12165 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12166 Enable TARGET_DOTPROD.
12167 (cortex-a75.cortex-a55): Likewise.
12168 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12170 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12172 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12173 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12175 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12177 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12178 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12179 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12180 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12181 * config/arm/types.md (neon_dot, neon_dot_q): New.
12182 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12184 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12186 * config/arm/arm.h (TARGET_DOTPROD): New.
12187 * config/arm/arm.c (arm_arch_dotprod): New.
12188 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12189 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12190 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12191 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12192 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12193 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12194 * doc/invoke.texi (armv8.2-a): Document dotprod
12196 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12198 * i386.c (ix86_vec_cost): New function.
12199 (ix86_rtx_costs): Handle vector operations better.
12200 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12201 * x86-tune-costs.h: Add new costs to all tables.
12203 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12205 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12207 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12208 divsd, sqrtss and sqrtsd
12209 * x86-tune-costs.h: Add new entries to all costs.
12210 (znver1_cost): Fix to match real instruction latencies.
12212 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12213 Michael Collison <michael.collison@arm.com>
12215 * compare-elim.c: Include emit-rtl.h.
12216 (can_merge_compare_into_arith): New function.
12217 (try_validate_parallel): Likewise.
12218 (try_merge_compare): Likewise.
12219 (try_eliminate_compare): Call the above when no previous clobber
12221 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12224 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12226 PR middle-end/62263
12227 PR middle-end/82498
12228 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12229 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12231 PR middle-end/62263
12232 PR middle-end/82498
12233 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12234 to be any operand_equal_p operands. For & (B - 1) require
12235 B to be power of 2. Recognize
12236 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12238 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12241 * optabs.c (expand_memory_blockage): Fix call of
12242 targetm.have_memory_blockage.
12244 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12247 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12248 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12249 objects to extra_objs instead of overwriting it.
12251 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12253 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12254 Use any_fp_register_operand as operand[3] predicate. Simplify
12255 equality test for operands[2] and operands[4] memory location.
12256 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12257 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12258 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12259 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12260 any_fp_register_operand as operand[1] predicate. Simplify
12261 equality test for operands[0] and operands[3] memory location.
12262 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12263 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12264 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12266 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12268 * target-insns.def: Add memory_blockage.
12269 * optabs.c (expand_memory_blockage): New function.
12270 (expand_asm_memory_barrier): Rename ...
12271 (expand_asm_memory_blockage): ... to this.
12272 (expand_mem_thread_fence): Call expand_memory_blockage
12273 instead of expand_asm_memory_barrier.
12274 (expand_mem_singnal_fence): Ditto.
12275 (expand_atomic_load): Ditto.
12276 (expand_atomic_store): Ditto.
12277 * doc/md.texi (Standard Pattern Names For Generation):
12278 Document memory_blockage instruction pattern.
12280 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12282 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12283 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12284 * config/rl78/rl78.md: New define_expand "adddi3".
12286 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12288 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12291 2017-10-13 Jeff Law <law@redhat.com>
12293 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12295 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12298 * internal-fn.c (expand_mul_overflow): If both operands have
12299 the same highpart of -1 or 0 and the topmost bit of lowpart
12300 is different, overflow is if res <= 0 rather than res < 0.
12302 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12304 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12305 TARGET_P9_VECTOR code for unaligned_load case.
12307 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12309 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12311 2017-10-13 Nathan Sidwell <nathan@acm.org>
12313 * tree-core.h (tree_contains_struct): Make bool.
12314 * tree.c (tree_contains_struct): Likewise.
12315 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12316 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12317 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12318 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12319 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12321 2017-10-13 Richard Biener <rguenther@suse.de>
12323 * graphite-isl-ast-to-gimple.c
12324 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12325 parameters and dominance check.
12326 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12327 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12328 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12329 Do not update SSA form here or do intermediate IL verification.
12330 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12331 (graphite_initialize): Remove check on the number of loops in
12332 the function and inline into graphite_transform_loops.
12333 (graphite_finalize): Inline into graphite_transform_loops.
12334 (graphite_transform_loops): Perform SSA update and IL verification
12336 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12338 2017-10-13 Richard Biener <rguenther@suse.de>
12340 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12341 graphite_expression_type_precision): Avoid global constructor
12343 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12344 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12345 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12346 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12347 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12348 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12350 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12353 * config/i386/i386.h (ix86_red_zone_size): New.
12354 * config/i386/i386.md (push peephole2s): Replace
12355 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12357 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12358 Alan Hayward <alan.hayward@arm.com>
12359 David Sherwood <david.sherwood@arm.com>
12361 * combine.c (can_change_dest_mode): Reject changes in
12362 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12369 (expand_debug_source_expr): Likewise.
12370 * combine.c (combine_simplify_rtx): Likewise.
12371 * cse.c (fold_rtx): Likewise.
12372 * fwprop.c (canonicalize_address): Likewise.
12373 * targhooks.c (default_shift_truncation_mask): 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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12380 (widened_mode): Likewise.
12381 (expand_unop): Likewise.
12382 * ree.c (transform_ifelse): Likewise.
12383 (merge_def_and_ext): Likewise.
12384 (combine_reaching_defs): Likewise.
12385 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12387 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12388 Alan Hayward <alan.hayward@arm.com>
12389 David Sherwood <david.sherwood@arm.com>
12391 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12392 * combine.c (gen_lowpart_for_combine): Likewise.
12393 * dwarf2out.c (rtl_for_decl_location): Likewise.
12394 * final.c (alter_subreg): Likewise.
12395 * rtlhooks.c (gen_lowpart_general): Likewise.
12396 (gen_lowpart_if_possible): Likewise.
12398 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12399 Alan Hayward <alan.hayward@arm.com>
12400 David Sherwood <david.sherwood@arm.com>
12402 * calls.c (expand_call): Use subreg_lowpart_offset.
12403 * cse.c (cse_insn): Likewise.
12404 * regcprop.c (copy_value): Likewise.
12405 (copyprop_hardreg_forward_1): Likewise.
12407 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12410 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12411 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12412 =Q constraints to +Q and into insn condition add check
12413 that operands[0] and operands[1] are equal.
12414 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12415 =Q constraints to +Q and into insn condition add check
12416 that operands[0] is equal to either operands[1] or operands[2].
12419 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12420 instead of handling MINUS_EXPR twice (once for each argument),
12421 canonicalize operand order and handle just once, use rtype where
12422 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12425 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12426 any values of __C while still being pattern recognizable as a simple
12427 rotate instruction.
12429 2017-10-13 Richard Biener <rguenther@suse.de>
12431 PR tree-optimization/82451
12433 2017-10-02 Richard Biener <rguenther@suse.de>
12435 PR tree-optimization/82355
12436 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12437 a mapping for the enclosing loop but avoid generating one for
12438 the loop tree root.
12439 (copy_bb_and_scalar_dependences): Remove premature codegen
12440 error on PHIs in blocks duplicated into multiple places.
12441 * graphite-scop-detection.c
12442 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12443 in the region use it as loop and nest to analyze the DR in.
12444 (try_generate_gimple_bb): Likewise.
12445 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12446 (add_loop_constraints): For blocks in a loop not in the region
12447 create a dimension with a single iteration.
12448 * sese.h (gbb_loop_at_index): Remove assert.
12450 * cfgloop.c (loop_preheader_edge): For the loop tree root
12451 return the single successor of the entry block.
12452 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12453 Reset the SCEV hashtable and niters.
12454 * graphite-scop-detection.c
12455 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12456 assert that we only have POLYNOMIAL_CHREC that vary in loops
12457 contained in the region.
12458 (scop_detection::graphite_can_represent_expr): Adjust.
12459 (scop_detection::stmt_has_simple_data_refs_p): For loops
12460 not in the region set loop to NULL. The nest is now the
12461 entry edge to the region.
12462 (try_generate_gimple_bb): Likewise.
12463 * sese.c (scalar_evolution_in_region): Adjust for
12464 instantiate_scev change.
12465 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12466 Make nest parameter the edge into the region.
12467 (create_data_ref): Likewise.
12468 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12469 entry edge into a region and adjust instantiate_scev calls.
12470 (create_data_ref): Likewise.
12471 (graphite_find_data_references_in_stmt): Likewise.
12472 (find_data_references_in_stmt): Pass the loop preheader edge
12473 from the nest argument.
12474 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12475 parameter the edge into the region.
12476 (instantiate_parameters): Use the loop preheader edge as entry.
12477 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12479 (get_instantiated_value_entry): Make instantiate_below parameter
12480 the edge into the region.
12481 (instantiate_scev_name): Likewise. Adjust dominance checks,
12482 when we cannot use loop-based instantiation instantiate by
12483 walking use-def chains.
12484 (instantiate_scev_poly): Adjust.
12485 (instantiate_scev_binary): Likewise.
12486 (instantiate_scev_convert): Likewise.
12487 (instantiate_scev_not): Likewise.
12488 (instantiate_array_ref): Remove.
12489 (instantiate_scev_3): Likewise.
12490 (instantiate_scev_2): Likewise.
12491 (instantiate_scev_1): Likewise.
12492 (instantiate_scev_r): Do not blindly handle N-operand trees.
12493 Do not instantiate array-refs. Handle all constants and invariants.
12494 (instantiate_scev): Make instantiate_below parameter
12495 the edge into the region.
12496 (resolve_mixers): Use the loop preheader edge for the region
12497 parameter to instantiate_scev_r.
12498 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12500 2017-10-13 Richard Biener <rguenther@suse.de>
12502 PR tree-optimization/82525
12503 * graphite-isl-ast-to-gimple.c
12504 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12506 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12507 Fail code generation when we cannot represent the isl integer.
12508 (binary_op_to_tree): Elide modulo operations that are no-ops
12509 in the type we code generate. Remove now superfluous code
12512 2017-10-13 Richard Biener <rguenther@suse.de>
12514 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12515 (scop_detection::harmful_loop_in_region): Remove premature
12516 IV type restriction.
12517 (scop_detection::graphite_can_represent_scev): We can handle
12518 pointer IVs just fine.
12520 2017-10-13 Alan Modra <amodra@gmail.com>
12522 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12523 "Clobbers and Scratch Registers". Add paragraph on
12524 alternative to clobbers for scratch registers and OpenBLAS
12527 2017-10-13 Alan Modra <amodra@gmail.com>
12529 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12530 example of a memory input for a string of known length. Move
12531 commentary out of table. Add a number of new examples
12532 covering array memory inputs.
12534 2017-10-12 Martin Liska <mliska@suse.cz>
12536 PR tree-optimization/82493
12537 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12538 (test_range_functions): New function.
12539 (sbitmap_c_tests): Likewise.
12540 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12541 * selftest.h (sbitmap_c_tests): New function.
12543 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12545 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12547 * config/rs6000/amo.h: Fix spacing issue.
12549 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12552 * config/i386/i386.md (*ashl<mode>3_mask_1,
12553 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12554 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12557 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12559 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12560 (profile_probability): Set max_probability
12561 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12563 * profile-count.c (profile_probability::differs_from_p): Do not
12564 rely on max_probaiblity == 10000
12566 2017-10-12 Jeff Law <law@redhat.com>
12568 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12571 2017-10-12 Martin Sebor <msebor@redhat.com>
12575 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12576 (handle_alias_pairs): Call it.
12577 * common.opt (-Wattribute-alias): New option.
12578 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12579 * doc/invoke.texi (-Wattribute-alias): Document.
12581 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12584 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12586 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12588 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12589 (make_hard_regno_born, make_hard_regno_dead): Update
12590 bb_killed_pseudos and bb_gen_pseudos.
12592 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12594 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12596 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12598 * config/alpha/alpha.c (alpha_split_conditional_move):
12599 Use std::swap instead of manually swapping.
12600 (alpha_stdarg_optimize_hook): Ditto.
12601 (alpha_canonicalize_comparison): Ditto.
12603 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12605 * tree-loop-distribution.c (struct builtin_info): New struct.
12606 (struct partition): Refactor fields into struct builtin_info.
12607 (partition_free): Free struct builtin_info.
12608 (build_size_arg_loc, build_addr_arg_loc): Delete.
12609 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12610 information from struct builtin_info.
12611 (find_single_drs): New function refactored from classify_partition.
12612 Also moved builtin validity checks to this function.
12613 (compute_access_range, alloc_builtin): New functions.
12614 (classify_builtin_st, classify_builtin_ldst): New functions.
12615 (classify_partition): Refactor code into functions find_single_drs,
12616 classify_builtin_st and classify_builtin_ldst.
12617 (distribute_loop): Don't do runtime alias check when distributing
12619 (find_seed_stmts_for_distribution): New function.
12620 (pass_loop_distribution::execute): Refactor code finding seed
12621 stmts into above function. Support distribution for the innermost
12622 two-level loop nest. Adjust dump information.
12624 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12626 * tree-loop-distribution.c: Adjust the general comment.
12627 (NUM_PARTITION_THRESHOLD): New macro.
12628 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12629 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12630 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12631 in call to build_partition_graph.
12632 (finalize_partitions): New parameter. Make loop distribution more
12633 conservative by fusing more partitions.
12634 (distribute_loop): Don't do runtime alias check in case of loop nest
12636 (find_seed_stmts_for_distribution): New function.
12637 (prepare_perfect_loop_nest): New function.
12638 (pass_loop_distribution::execute): Refactor code finding seed stmts
12639 and loop nest into above functions. Support loop nest distribution.
12640 Adjust dump information accordingly.
12642 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12644 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12645 and set PTYPE_SEQUENTIAL for merged partition.
12647 2017-10-12 Richard Biener <rguenther@suse.de>
12649 PR tree-optimization/69728
12651 2017-09-19 Richard Biener <rguenther@suse.de>
12653 PR tree-optimization/69728
12654 * graphite-sese-to-poly.c (schedule_error): New global.
12655 (add_loop_schedule): Handle empty domain by failing the
12657 (build_original_schedule): Handle schedule_error.
12659 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12660 domain by returning an unchanged schedule.
12662 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12664 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12665 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12667 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12669 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12672 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12675 * expr.c (store_field): Don't optimize away bitsize == 0 store
12676 from CALL_EXPR with addressable return type.
12678 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12680 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12681 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12682 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12683 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12684 TARGET_ISEL instead of TARGET_ISEL<sel>.
12686 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12688 * config/rs6000/rs6000.c
12689 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12691 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12693 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12695 (reg_or_cint_operand): Fix comment.
12696 (reg_or_zero_operand): New predicate.
12697 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12698 * config/rs6000/rs6000.c (output_isel): Delete.
12699 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12700 instead of reg_or_cint_operand. Output instruction directly (not via
12702 (isel_unsigned_<mode>): Ditto.
12703 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12704 gpc_reg_operand. Add an instruction alternative for this. Output
12705 instruction directly.
12706 (*isel_reversed_unsigned_<mode>): Ditto.
12708 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12710 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12711 (TARGET_CANONICALIZE_COMPARISON): Define.
12713 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12716 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12717 Check whether the dest is REG before adding REG_EQUIV note.
12719 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12722 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12724 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12725 (make_hard_regno_born, make_hard_regno_dead): Update
12726 bb_killed_pseudos and bb_gen_pseudos.
12728 2017-10-11 Nathan Sidwell <nathan@acm.org>
12730 * incpath.h (enum incpath_kind): Name enum, prefix values.
12731 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12732 * incpath.c (heads, tails): Use INC_MAX.
12733 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12734 (merge_include_chains, split_quote_chain,
12735 register_include_chains): Update incpath_kind names.
12736 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12737 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12739 (add_framework_path, darwin_register_objc_includes): Likewise.
12740 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12742 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12744 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12745 Do not use float_operator operator predicate.
12746 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12747 * config/i386/predicates.md (float_operator): Remove predicate.
12749 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12751 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12752 (*jccxf_i387): Ditto.
12753 (*jcc<mode>_i387): Ditto.
12754 (*jccu<mode>_i387): Ditto.
12755 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12756 (*jcc_*_i387 splitters): Remove.
12757 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12758 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12759 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12762 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12764 * profile-count.h (slow_safe_scale_64bit): New function.
12765 (safe_scale_64bit): New inline.
12766 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12767 * profile-count.c: Include wide-int.h
12768 (slow_safe_scale_64bit): New.
12770 2017-10-11 Nathan Sidwell <nathan@acm.org>
12772 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12773 HAS_DECL_ASSEMBLER_NAME_P.
12774 * gimple-expr.c (gimple_decl_printable_name: Check
12775 HAS_DECL_ASSEMBLER_NAME_P too.
12776 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12777 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12778 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12779 * passes.c (rest_of_decl_compilation): Check
12780 HAS_DECL_ASSEMBLER_NAME_P too.
12781 * recog.c (verify_changes): Likewise.
12782 * tree-pretty-print.c (dump_decl_name): Likewise.
12783 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12785 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12786 (SET_DECL_ASSEMBLER_NAME): Use it.
12787 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12788 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12789 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12791 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12793 * config.gcc (i386, x86_64): Add extra objects.
12794 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12795 (ix86_min_insn_size): Declare.
12796 (ix86_issue_rate): Declare.
12797 (ix86_adjust_cost): Declare.
12798 (ia32_multipass_dfa_lookahead): Declare.
12799 (ix86_macro_fusion_p): Declare.
12800 (ix86_macro_fusion_pair_p): Declare.
12801 (ix86_bd_has_dispatch): Declare.
12802 (ix86_bd_do_dispatch): Declare.
12803 (ix86_core2i7_init_hooks): Declare.
12804 (ix86_atom_sched_reorder): Declare.
12805 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12806 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12807 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12808 (rip_relative_addr_p): Rename to ...
12809 (ix86_rip_relative_addr_p): ... this one; export.
12810 (memory_address_length): Update.
12811 (ix86_issue_rate): Move to x86-tune-sched.c.
12812 (ix86_flags_dependent): Move to x86-tune-sched.c.
12813 (ix86_agi_dependent): Move to x86-tune-sched.c.
12814 (exact_dependency_1): Move to x86-tune-sched.c.
12815 (exact_store_load_dependency): Move to x86-tune-sched.c.
12816 (ix86_adjust_cost): Move to x86-tune-sched.c.
12817 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12818 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12819 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12820 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12821 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12822 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12823 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12824 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12825 (min_insn_size): Rename to ...
12826 (ix86_min_insn_size): ... this one; export.
12827 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12828 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12829 (core2i7_first_cycle_multipass_backtrack): Move to
12830 x86-tune-sched-core.c.
12831 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12832 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12833 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12834 (ix86_avoid_jump_mispredicts): Update.
12835 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12836 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12837 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12838 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12839 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12840 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12841 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12842 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12843 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12844 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12845 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12846 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12847 (BIG): Move to ix86-tune-sched-bd.c.
12848 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12849 (enum insn_path): Move to ix86-tune-sched-bd.c.
12850 (get_mem_group): Move to ix86-tune-sched-bd.c.
12851 (is_cmp): Move to ix86-tune-sched-bd.c.
12852 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12853 (is_branch): Move to ix86-tune-sched-bd.c.
12854 (is_prefetch): Move to ix86-tune-sched-bd.c.
12855 (init_window): Move to ix86-tune-sched-bd.c.
12856 (allocate_window): Move to ix86-tune-sched-bd.c.
12857 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12858 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12859 (process_end_window): Move to ix86-tune-sched-bd.c.
12860 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12861 (find_constant): Move to ix86-tune-sched-bd.c.
12862 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12863 (has_immediate): Move to ix86-tune-sched-bd.c.
12864 (get_insn_path): Move to ix86-tune-sched-bd.c.
12865 (get_insn_group): Move to ix86-tune-sched-bd.c.
12866 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12867 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12868 (add_insn_window): Move to ix86-tune-sched-bd.c.
12869 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12870 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12871 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12872 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12873 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12874 (do_dispatch): Move to ix86-tune-sched-bd.c.
12875 (has_dispatch): Move to ix86-tune-sched-bd.c.
12876 * config/i386/t-i386: Add new object files.
12877 * config/i386/x86-tune-costs.h: New file.
12878 * config/i386/x86-tune-sched-atom.c: New file.
12879 * config/i386/x86-tune-sched-bd.c: New file.
12880 * config/i386/x86-tune-sched-core.c: New file.
12881 * config/i386/x86-tune-sched.c: New file.
12883 2017-10-11 Liu Hao <lh_mouse@126.com>
12885 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12886 the generic version below instead.
12887 (should_colorize): Recognize Windows consoles as terminals
12889 * pretty-print.c [__MINGW32__] (write_all): New function.
12890 [__MINGW32__] (find_esc_head): Likewise.
12891 [__MINGW32__] (find_esc_terminator): Likewise.
12892 [__MINGW32__] (eat_esc_sequence): Likewise.
12893 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12895 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12898 2017-10-11 Richard Biener <rguenther@suse.de>
12900 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12901 Properly call analyze_scalar_evolution with the loop of the stmt.
12903 2017-10-11 Richard Biener <rguenther@suse.de>
12905 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12906 * tree-core.h (tree_base): Add chrec_var union member.
12907 * tree.h (CHREC_VAR): Remove.
12908 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12909 * tree-chrec.h (build_polynomial_chrec): Adjust.
12910 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12911 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12913 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12915 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12916 * match.pd: ... here.
12917 ((T) X == (T) Y): Relax condition.
12919 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12921 PR tree-optimization/82472
12922 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12924 (break_alias_scc_partitions): Update postorder number.
12926 2017-10-11 Martin Liska <mliska@suse.cz>
12929 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12931 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12933 2017-10-11 Martin Liska <mliska@suse.cz>
12935 * print-rtl.c (print_insn): Move declaration of idbuf
12936 to same scope as name.
12938 2017-10-11 Martin Liska <mliska@suse.cz>
12943 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12944 volatile arguments.
12946 2017-10-11 Martin Liska <mliska@suse.cz>
12949 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12950 volatile arguments.
12952 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12954 * config.gcc (default_gnu_indirect_function): Default to yes for
12955 arm*-*-linux* with glibc.
12957 2017-10-11 Richard Biener <rguenther@suse.de>
12959 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12960 default-defs and types we do not want to analyze.
12961 (interpret_loop_phi): Replace unreachable code with an assert.
12962 (compute_scalar_evolution_in_loop): Remove and inline ...
12963 (analyze_scalar_evolution_1): ... here, replacing condition with
12964 what makes the intent clearer. Remove handling of cases
12965 get_scalar_evolution now handles.
12967 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12969 PR rtl-optimization/81434
12970 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12971 comment for main loop. In sched_group_found if, also add checks for
12972 pass and min_cost_group.
12974 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12976 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12977 (rs6000_insn_cost): New function.
12978 * config/rs6000/rs6000.md (cost): New attribute.
12980 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12981 H.J. Lu <hongjiu.lu@intel.com>
12985 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12986 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12987 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12988 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12989 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12990 OPTION_MASK_ISA_MMX and some other bit set require both
12991 mmx and the ISAs without the mmx bit.
12992 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12993 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12994 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12995 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12996 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12997 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12998 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12999 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13000 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13001 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13002 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13003 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13004 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13005 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13006 Add OPTION_MASK_ISA_MMX.
13008 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13010 * config.gcc (armv7*-*-freebsd*): New target.
13011 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13013 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13015 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13016 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13019 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13021 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13022 a template parameter.
13023 (WIDE_INT_REF_FOR): Update accordingly.
13024 * tree.h (wi::int_traits <const_tree>): Delete.
13025 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13026 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13027 (wi::tree_to_wide_ref): New typedef.
13028 (wi::to_wide): New function.
13029 * calls.c (get_size_range): Use wi::to_wide when operating on
13030 trees as wide_ints.
13031 * cgraph.c (cgraph_node::create_thunk): Likewise.
13032 * config/i386/i386.c (ix86_data_alignment): Likewise.
13033 (ix86_local_alignment): Likewise.
13034 * dbxout.c (stabstr_O): Likewise.
13035 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13036 * expr.c (const_vector_from_tree): Likewise.
13037 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13038 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13039 (fold_negate_expr_1, int_const_binop_1, const_binop)
13040 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13041 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13042 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13043 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13044 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13045 (fold_not_const, round_up_loc): Likewise.
13046 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13047 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13048 (alloca_call_type): Likewise.
13049 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13050 * godump.c (go_output_typedef): Likewise.
13051 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13052 * internal-fn.c (get_min_precision): Likewise.
13053 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13054 * ipa-polymorphic-call.c
13055 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13056 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13057 (ipa_modify_call_arguments): Likewise.
13058 * match.pd: Likewise.
13059 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13060 * print-tree.c (print_node_brief, print_node): Likewise.
13061 * stmt.c (expand_case): Likewise.
13062 * stor-layout.c (layout_type): Likewise.
13063 * tree-affine.c (tree_to_aff_combination): Likewise.
13064 * tree-cfg.c (group_case_labels_stmt): Likewise.
13065 * tree-data-ref.c (dr_analyze_indices): Likewise.
13066 (prune_runtime_alias_test_list): Likewise.
13067 * tree-dump.c (dequeue_and_dump): Likewise.
13068 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13069 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13070 * tree-pretty-print.c (dump_generic_node): Likewise.
13071 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13072 (simple_iv_with_niters): Likewise.
13073 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13074 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13075 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13076 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13077 (refine_value_range_using_guard, number_of_iterations_ne_max)
13078 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13079 (get_cst_init_from_scev, record_nonwrapping_iv)
13080 (scev_var_range_cant_overflow): Likewise.
13081 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13082 * tree-ssa-pre.c (compute_avail): Likewise.
13083 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13084 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13085 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13086 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13087 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13088 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13089 (dump_case_nodes, try_switch_expansion): Likewise.
13090 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13091 (vect_do_peeling): Likewise.
13092 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13093 * tree-vect-stmts.c (vectorizable_load): Likewise.
13094 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13095 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13096 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13097 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13098 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13099 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13100 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13101 (evrp_dom_walker::before_dom_children): Likewise.
13102 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13103 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13104 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13105 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13106 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13107 (get_range_pos_neg): Likewise.
13108 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13109 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13110 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13111 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13112 * config/avr/avr.c (avr_fold_builtin): Likewise.
13113 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13114 * config/msp430/msp430.c (msp430_attr): Likewise.
13115 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13116 * config/powerpcspe/powerpcspe-c.c
13117 (altivec_resolve_overloaded_builtin): Likewise.
13118 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13119 (rs6000_expand_ternop_builtin): Likewise.
13120 * config/rs6000/rs6000-c.c
13121 (altivec_resolve_overloaded_builtin): Likewise.
13122 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13123 (rs6000_expand_ternop_builtin): Likewise.
13124 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13126 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13128 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13129 when copying loop nest with only one inner loop.
13131 2017-10-10 Richard Biener <rguenther@suse.de>
13133 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13134 blocks if SCEV is active.
13135 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13137 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13138 (scev_initialize): Assert we are not yet initialized.
13140 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13142 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13143 inner loop's exit stmt by making it always exit the loop, otherwise
13144 we would generate an infinite empty loop.
13146 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13148 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13149 renaming variables in new preheader if it's deleted.
13151 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13153 * tree-loop-distribution.c (struct partition): Remove unused field
13154 loops of the structure.
13155 (partition_alloc, partition_free): Ditto.
13156 (build_rdg_partition_for_vertex): Ditto.
13158 2017-10-09 Jeff Law <law@redhat.com>
13160 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13161 return type to match prototype and documentation.
13163 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13165 * config/rs6000/rs6000.c (processor_costs): Move to ...
13166 * config/rs6000/rs6000.h: ... here.
13167 (rs6000_cost): Declare.
13169 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13171 * except.c (setjmp_fn): New global variable.
13172 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13173 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13174 if DONT_USE_BUILTIN_SETJMP is defined.
13176 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13178 * target.def (insn_cost): New hook.
13179 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13180 * doc/tm.texi: Regenerate.
13181 * rtlanal.c (insn_cost): Use the new hook.
13183 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13185 * combine.c (combine_validate_cost): Compute the new insn_cost,
13186 not just pattern_cost.
13187 (try_combine): Adjust comment.
13189 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13191 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13193 * combine.c (uid_insn_cost): Adjust comment.
13194 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13196 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13197 * dse.c (find_shift_sequence): Ditto.
13198 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13199 (bb_valid_for_noce_process_p): Use pattern_cost.
13200 * rtl.h (insn_rtx_cost): Delete.
13201 (pattern_cost): New prototype.
13202 (insn_cost): New prototype.
13203 * rtlanal.c (insn_rtx_cost): Rename to...
13204 (pattern_cost): ... this.
13207 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13209 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13210 (*jcc<mode>_0_r_i387): Ditto.
13211 (*jccxf_r_i387): Ditto.
13212 (*jcc<mode>_r_i387): Ditto.
13213 (*jccu<mode>_r_i387): Ditto.
13214 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13215 (*jcc): Rename from *jcc_1.
13217 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13219 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13220 deferred rescans after the lvx/stvx recombination pre-pass.
13222 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13224 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13225 memory operation instruction support.
13226 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13227 (rs6000-ibm-aix[789]*): Likewise.
13228 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13229 Document new functions.
13231 2017-10-09 Richard Biener <rguenther@suse.de>
13233 PR tree-optimization/82397
13234 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13235 equality only for semantically equal trees.
13237 2017-10-09 Richard Biener <rguenther@suse.de>
13239 PR tree-optimization/82449
13240 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13241 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13242 allow constant addresses.
13243 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13246 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13248 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13251 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13254 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13257 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13260 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13262 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13265 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13266 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13268 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13270 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13271 (WI_BINARY_PREDICATE_RESULT): Likewise.
13272 (wi::binary_traits::operator_result): New type.
13273 (wi::binary_traits::predicate_result): Likewise.
13274 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13275 (generic_wide_int::operator==, generic_wide_int::operator!=)
13276 (generic_wide_int::operator&, generic_wide_int::and_not)
13277 (generic_wide_int::operator|, generic_wide_int::or_not)
13278 (generic_wide_int::operator^, generic_wide_int::operator+
13279 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13281 (operator~, unary operator-, operator==, operator!=, operator&)
13282 (operator|, operator^, operator+, binary operator-, operator*): New
13284 * expr.c (get_inner_reference): Use wi::bit_and_not.
13285 * fold-const.c (fold_binary_loc): Likewise.
13286 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13287 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13288 (bit_value_binop): Likewise.
13289 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13290 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13291 (extract_range_from_binary_expr_1): Likewise.
13292 (masked_increment): Likewise.
13293 (simplify_bit_ops_using_ranges): Likewise.
13295 2017-10-09 Martin Jambor <mjambor@suse.cz>
13298 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13299 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13300 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13302 (hsa_fixup_mov_insn_type): New function.
13303 (hsa_op_with_type::get_in_type): Use it.
13304 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13305 immediates in an assert.
13306 (hsa_op_with_type::extend_int_to_32bit): New method.
13307 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13308 types. Convert to dest type if necessary.
13309 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13310 (reg_for_gimple_ssa): Pass false as min32int to
13311 hsa_type_for_scalar_tree_type.
13312 (gen_hsa_addr): Fixup type when creating addresable temporary.
13313 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13314 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13315 necessary. Call hsa_fixup_mov_insn_type.
13316 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13317 extend operands and convert to dest type if necessary.
13318 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13319 to dest type if necessary.
13320 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13321 if conversion nt necessary and size matches.
13322 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13323 to dest type if necessary.
13324 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13325 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13327 (gen_hsa_clrsb): Likewise.
13328 (gen_hsa_ffs): Likewise.
13329 (gen_hsa_divmod): Extend operands and convert to dest type if
13331 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13333 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13335 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13336 Remove empty default arguments. Use a brace block as output
13338 (conditional return): Ditto.
13340 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13341 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13343 (group_ending_nop): Ditto.
13344 (doloop_end): Ditto.
13345 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13346 (splitters for those): Ditto.
13348 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13350 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13351 a conditional jump (and the compare for it) so that pc_rtx is the
13353 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13354 for the deleted and renamed ctr<mode>_internal[234] patterns.
13355 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13356 Delete second conditional return pattern.
13357 (ctr<mode>_internal2): Delete this second bdnz pattern.
13358 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13359 (ctr<mode>_internal4): Delete this second bdz pattern.
13361 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13363 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13364 (always_initialized_rtx_for_ssa_name_p): New predicate.
13365 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13366 (finish_out_of_ssa): Free new field of SA.
13367 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13368 * tree-ssa-coalesce.c: Include tree-ssa.h.
13369 (get_parm_default_def_partitions): Remove extern keyword.
13370 (get_undefined_value_partitions): New function.
13371 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13372 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13373 uninitialized bits.
13374 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13376 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13378 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13380 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13382 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13383 for targets that preffer 128bit.
13385 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13387 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13389 2017-10-08 Olivier Hainque <hainque@adacore.com>
13391 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13392 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13394 (thumb_set_return_address): Likewise.
13396 2017-10-08 Olivier Hainque <hainque@adacore.com>
13398 * common/config/arm/arm-common.c (arm_except_unwind_info):
13399 Handle DWARF2_UNWIND_INFO.
13401 2017-10-07 Michael Collison <michael.collison@arm.com>
13403 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13406 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13408 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13409 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13410 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13411 defined, force the creation of a new block for a dispatch label.
13413 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13415 * invoke.texi (Wsuggest-attribute=cold): Document.
13416 * common.opt (Wsuggest-attribute=cold): New
13417 * ipa-pure-const.c (warn_function_cold): New function.
13418 * predict.c (compute_function_frequency): Use it.
13419 * predict.h (warn_function_cold): Declare.
13421 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13423 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13426 2017-10-06 Martin Liska <mliska@suse.cz>
13428 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13429 keyword for member functions.
13430 (struct sanopt_tree_couple): New struct.
13431 (struct sanopt_tree_couple_hash): New function.
13432 (struct sanopt_ctx): Add new hash_map.
13433 (has_dominating_ubsan_ptr_check): New function.
13434 (record_ubsan_ptr_check_stmt): Likewise.
13435 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13436 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13437 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13439 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13442 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13443 aarch64_simd_valid_immediate on CONST_VECTORs.
13444 (aarch64_reg_or_bic_imm): Likewise.
13446 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13448 PR rtl-optimization/82396
13449 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13451 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13453 * graphite-dependences.c (scop_get_reads): Move code to...
13454 (scop_get_must_writes): Move code to...
13455 (scop_get_may_writes): Move code to...
13456 (scop_get_reads_and_writes): ... here.
13457 (scop_get_dependences): Call scop_get_reads_and_writes.
13459 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13461 PR tree-optimization/82434
13462 * fold-const.h (can_native_encode_type_p,
13463 can_native_encode_string_p): Remove.
13464 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13465 don't encode anything, just return what would be otherwise returned.
13466 (native_encode_fixed, native_encode_complex, native_encode_vector):
13468 (native_encode_string): Likewise. Inline by hand
13469 can_native_encode_string_p.
13470 (can_native_encode_type_p): Remove.
13471 (can_native_encode_string_p): Remove.
13472 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13473 STRING_CSTs using can_native_encode_string_p, test all
13474 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13475 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13476 argument from native_encode_expr.
13477 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13478 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13481 2017-10-06 Richard Biener <rguenther@suse.de>
13483 PR tree-optimization/82397
13484 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13485 operand_equal_p but rely on data_ref_compare_tree for detecting
13487 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13488 to match up with dr_group_sort_cmp.
13490 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13493 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13495 * config/s390/s390-builtin-types.def: Regenerate.
13497 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13500 * config/s390/s390-builtin-types.def: Regenerate.
13501 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13502 Change flag from B_VXE to B_VX.
13503 (s390_vec_min_dbl): Remove B_VXE flag.
13505 2017-10-06 Richard Biener <rguenther@suse.de>
13507 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13508 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13509 translate_isl_ast_to_gimple::is_valid_rename,
13510 translate_isl_ast_to_gimple::get_rename,
13511 translate_isl_ast_to_gimple::get_def_bb_for_const,
13512 translate_isl_ast_to_gimple::get_new_name,
13513 translate_isl_ast_to_gimple::collect_all_ssa_names,
13514 translate_isl_ast_to_gimple::copy_loop_phi_args,
13515 translate_isl_ast_to_gimple::collect_all_ssa_names,
13516 translate_isl_ast_to_gimple::copy_loop_phi_args,
13517 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13518 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13519 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13520 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13521 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13522 translate_isl_ast_to_gimple::copy_cond_phi_args,
13523 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13524 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13525 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13526 translate_isl_ast_to_gimple::rename_uses,
13527 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13528 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13529 (set_rename_for_each_def): Likewise.
13530 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13531 here. Handle rewriting SCEV analyzable uses here.
13532 (copy_bb_and_scalar_dependences): Generate code for PHI
13534 (graphite_regenerate_ast_isl): Adjust.
13535 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13536 (add_write, add_read): New functions.
13537 (build_cross_bb_scalars_def): Use it and simplify.
13538 (build_cross_bb_scalars_use): Likewise.
13539 (graphite_find_cross_bb_scalar_vars): Inline into...
13540 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13541 simulating out-of-SSA. Compute liveout and add dependencies.
13542 (build_scops): Force an empty entry block.
13543 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13545 (sese_build_liveouts): Declare.
13546 (sese_trivially_empty_bb_p): Likewise.
13547 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13548 compute liveout and debug_liveout.
13549 (sese_bad_liveouts_use): Remove.
13550 (sese_reset_debug_liveouts_bb): Likewise.
13551 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13552 (sese_build_liveouts): Build liveout and debug_liveout and store
13554 (new_sese_info): Adjust.
13555 (free_sese_info): Likewise.
13556 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13557 do not build liveout here.
13558 (move_sese_in_condition): Adjust region entry.
13559 (scev_analyzable_p): Match up with chrec_apply requirements.
13560 (sese_trivially_empty_bb_p): New.
13561 * tree-into-ssa.c (get_reaching_def): Properly support generating
13562 default-defs for incremental rewrite of anonymous names.
13564 2017-10-06 Richard Biener <rguenther@suse.de>
13566 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13567 precision do not perform modulo reduction.
13569 2017-10-06 Richard Biener <rguenther@suse.de>
13571 PR tree-optimization/82436
13572 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13573 conservatively choose the vectorization factor when checking
13574 whether we can perform the required load permutation.
13575 (vect_transform_slp_perm_load): Assert when we may not fail.
13577 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13579 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13580 message for incompatible -msdata=* and -mcall-* options.
13582 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13584 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13585 rate for post-reload scheduling.
13587 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13589 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13591 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13593 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13594 to improve monte carlo in scimark.
13596 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13598 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13599 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13600 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13601 pentium4_cost, nocona_cost): Set reassociation width to 1.
13602 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13603 width to 2 for fp operations and 1 otherwise.
13604 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13606 (atom_cost): Set reassociation width to 2.
13607 (slm_cost, generic_cost): Set fp reassociation width
13608 to 2 and 1 otherwise.
13609 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13610 (core_cost): Set fp reassociation width to 4 and vector to 2.
13611 (ix86_reassociation_width): Rewrite using cost table; special case
13612 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13613 and TARGET_AVX128_OPTIMAL.
13614 * config/i386/i386.h (processor_costs): Add
13615 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13616 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13617 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13618 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13619 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13620 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13622 2017-10-05 Nathan Sidwell <nathan@acm.org>
13624 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13626 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13628 * config/arm/arm.c (arm_test_fpu_data): New.
13629 (arm_run_selftests): Call arm_test_fpu_data.
13631 2017-10-04 Nathan Sidwell <nathan@acm.org>
13633 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13635 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13637 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13639 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13640 check type for aarch64_simd_valid_immediate.
13641 (aarch64_output_simd_mov_immediate): Update prototype.
13642 (aarch64_simd_valid_immediate): Update prototype.
13643 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13644 support for ORR-immediate.
13645 (and<mode>3): modified pattern to add support for BIC-immediate.
13646 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13647 now checks for valid immediate for BIC and ORR based on new enum
13649 (aarch64_output_simd_mov_immediate): Function now used to output
13650 BIC/ORR imm as well based on new enum argument.
13651 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13653 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13654 (aarch64_reg_or_bic_imm): Likewise.
13656 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13658 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13659 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13661 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13665 PR rtl-optimization/82396
13666 * haifa-sched.c (autopref_multipass_init): Simplify
13668 (autopref_rank_data): Simplify sort order.
13669 * sched-int.h (autopref_multipass_data_): Remove
13670 multi_mem_insn_p, min_offset and max_offset.
13672 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13674 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13676 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13678 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13679 vect_doubleint_cvt.
13681 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13683 * doc/sourcebuild.texi: Document vect_long_mult.
13685 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13687 PR tree-optimization/82413
13688 * fold-const.c (build_range_check): Use widest_int when comparing
13689 the maximum ETYPE value with HIGH.
13691 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13693 PR rtl-optimization/82396
13694 * haifa-sched.c (autopref_multipass_init): Simplify
13696 (autopref_rank_data): Simplify sort order.
13697 * sched-int.h (autopref_multipass_data_): Remove
13698 multi_mem_insn_p, min_offset and max_offset.
13700 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13702 PR tree-optimization/82381
13703 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13704 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13707 PR tree-optimization/82374
13708 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13709 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13710 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13711 current_function_decl to the new decl.
13713 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13715 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13716 helper macro for IEEE float128 hardware built-in functions.
13717 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13719 (TRUNCF128_ODD): Likewise.
13720 (ADDF128_ODD): Likewise.
13721 (SUBF128_ODD): Likewise.
13722 (MULF128_ODD): Likewise.
13723 (DIVF128_ODD): Likewise.
13724 (FMAF128_ODD): Likewise.
13725 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13726 UNSPEC_TRUNC_ROUND_TO_ODD.
13727 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13728 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13729 floating point round to odd instructions.
13730 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13731 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13732 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13733 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13734 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13735 (trunc<mode>sf2_hw): Change the truncate with round to odd
13736 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13737 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13738 to odd hardware instructions.
13739 (sub<mode>3_odd): Likewise.
13740 (mul<mode>3_odd): Likewise.
13741 (div<mode>3_odd): Likewise.
13742 (sqrt<mode>2_odd): Likewise.
13743 (fma<mode>4_odd): Likewise.
13744 (fms<mode>4_odd): Likewise.
13745 (nfma<mode>4_odd): Likewise.
13746 (nfms<mode>4_odd): Likewise.
13747 (trunc<mode>df2_odd): Change the truncate with round to odd
13748 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13750 * doc/extend.texi (PowerPC built-in functions): Update documentation
13751 for existing IEEE float128-bit built-in functions. Add built-in
13752 functions that generate the IEEE 128-bit floating point round to
13755 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13757 PR rtl-optimization/77729
13758 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13759 to (X&(C1&~C2))|C2 transformations.
13761 2017-10-03 Martin Jambor <mjambor@suse.cz>
13763 PR tree-optimization/82363
13764 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13765 mismatch, mark lacc written regardless of racc.
13767 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13769 PR tree-optimization/82381
13770 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13771 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13772 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13773 fallthrough into reassoc_stmt_dominates_stmt_p.
13776 * combine.c (combine_instructions): Don't combine in unreachable
13779 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13782 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13783 function to not use the have_cpu variable. Do not set cpu_index,
13784 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13785 or the default cpu.
13786 (rs6000_valid_attribute_p): Remove duplicate initializations of
13787 old_optimize and func_optimize.
13788 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13789 (rs6000_activate_target_options): Make global.
13790 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13793 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13795 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13796 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13797 Set *poffset to 0 and *psize and *pmax_size to -1 if
13798 *poffset + *psize overflows in HOST_WIDE_INT.
13800 PR tree-optimization/82387
13801 PR tree-optimization/82388
13802 PR tree-optimization/82389
13803 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13804 instead of live_bytes non-NULL.
13806 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13809 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13812 2017-10-02 Richard Biener <rguenther@suse.de>
13814 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13815 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13816 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13818 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13820 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13821 requested precision matches the type's.
13822 * calls.c (alloc_max_size): Calculate the new candidate size as
13823 a widest_int and use wi::to_widest when comparing it with the
13824 current candidate size.
13825 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13826 zero rather than integer_zero_node.
13827 * match.pd: Check for a no-op conversion before using wi::add
13828 rather than after. Use tree_to_uhwi when summing small shift
13829 counts into an unsigned int.
13831 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13832 Alan Hayward <alan.hayward@arm.com>
13833 David Sherwood <david.sherwood@arm.com>
13836 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13837 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13838 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13839 POINTER_AND_FP_REGS.
13841 2017-10-02 Richard Biener <rguenther@suse.de>
13843 PR tree-optimization/82355
13844 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13845 a mapping for the enclosing loop but avoid generating one for
13846 the loop tree root.
13847 (copy_bb_and_scalar_dependences): Remove premature codegen
13848 error on PHIs in blocks duplicated into multiple places.
13849 * graphite-scop-detection.c
13850 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13851 in the region use it as loop and nest to analyze the DR in.
13852 (try_generate_gimple_bb): Likewise.
13853 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13854 (add_loop_constraints): For blocks in a loop not in the region
13855 create a dimension with a single iteration.
13856 * sese.h (gbb_loop_at_index): Remove assert.
13858 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13860 * omp-expand.c (adjust_context_scope): New function.
13861 (expand_parallel_call): Call adjust_context_scope.
13863 2017-10-01 Jeff Law <law@redhat.com>
13865 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13866 dom_opt_dom_walker class with direct access to private members.
13867 Add comments. Call test_for_singularity.
13868 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13869 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13870 m_dummy_cond anymore.
13871 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13873 (pass_dominator:execute): Build the dummy_cond here and pass it
13874 to the dom_opt_dom_walker ctor.
13875 (test_for_singularity): New function.
13877 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13878 Maya Rashish <coypu@sdf.org>
13880 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13881 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13882 (alpha*-*-netbsd*) Use nbsd_tm_file.
13883 (arm*-*-netbsdelf*) Likewise.
13884 (i[34567]86-*-netbsdelf*) Likewise.
13885 (x86_64-*-netbsd*) Likewise.
13886 (mips*-*-netbsd*) Likewise.
13887 (powerpc-*-netbsd*) Likewise.
13888 (sh*-*-netbsd*) Likewise.
13889 (sparc-*-netbsdelf*) Likewise.
13890 (sparc64-*-netbsd*) Likewise.
13891 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13893 (vax-*-netbsdelf*) Likewise.
13894 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13895 (UINT_FAST8_TYPE) Likewise.
13896 (INT_FAST16_TYPE) Check CHAR_FAST16.
13897 (UINT_FAST16_TYPE) Likewise.
13899 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13902 * config/i386/i386.md
13903 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13904 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13905 *divmodsi4_zext_2): New define_insn_and_split.
13906 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13907 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13908 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13909 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13910 New define_insn_and_split.
13911 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13912 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13913 operands[1] having DImode when mode is SImode.
13915 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13916 always SImode for DIV and MOD in REG_EQUAL notes.
13918 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13920 PR middle-end/82319
13921 * match.pd: Fix handling of NaNs in pattern.
13923 2017-09-29 Jeff Law <law@redhat.com>
13925 * sbitmap.c (bitmap_bit_in_range_p): New function.
13926 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13927 * tree-ssa-dse.c (live_bytes_read): New function.
13928 (dse_classify_store): Ignore reads of dead bytes.
13930 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13931 typos and whitespace errors.
13932 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13933 * config/s390/s390.c (s390_emit_prologue): Likewise.
13935 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13938 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13939 with a symbol for LRA.
13941 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13943 PR rtl-optimization/82338
13944 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13946 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13948 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13949 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13950 (qsort_chk): Declare.
13951 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13952 (qsort_chk): New function.
13954 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13956 PR tree-optimization/82337
13957 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13958 phi definition if the PHI result appears in an abnormal PHI.
13959 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13960 basis appears in an abnormal PHI.
13962 2017-09-29 Richard Biener <rguenther@suse.de>
13964 * graphite-isl-ast-to-gimple.c
13965 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13966 (binary_op_to_tree): Use it.
13967 (get_rename_from_scev): Likewise.
13968 (copy_loop_phi_nodes): Likewise.
13969 (copy_bb_and_scalar_dependences): Likewise.
13970 (translate_pending_phi_nodes): Likewise.
13972 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13975 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13976 for movabsq $(i32 << shift), r64.
13978 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13980 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13981 index when encoding %esp as %rsp to avoid 0x67 prefix.
13983 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13985 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13986 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13988 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13990 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13991 Extensions with more than 16 double VFP registers.
13992 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13993 to_clear_mask and all code related to it. Replace the remaining
13994 entry by a sbitmap and adapt code accordingly.
13996 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13998 * brig-builtins.def: Change pure attributes to const.
14000 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14002 * config.gcc (default_gnu_indirect_function): Default to yes for
14003 sparc*-*-linux* with glibc.
14005 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14007 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14008 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14009 when creating .init_array and .fini_array sections with priority
14012 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14015 * config/aarch64/aarch64.c
14016 (aarch64_builtin_support_vector_misalignment): Always return false
14017 when misalignment is unknown.
14019 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14021 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14022 this function to return false if the definition used by the swap
14023 instruction is artificial, or if the memory address from which the
14024 constant value is loaded is not represented by a base address held
14025 in a register or if the base address register is a frame or stack
14026 pointer. Additionally, return false if the base address of the
14027 loaded constant is a SYMBOL_REF but is not considered to be a
14029 (replace_swapped_load_constant): New function.
14030 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14031 loaded constant vector with a load of a swapped constant vector.
14033 2017-09-27 Carl Love <cel@us.ibm.com>
14035 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14036 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14037 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14040 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14042 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14043 first, always call autopref_rank_data otherwise.
14045 2017-09-27 Richard Biener <rguenther@suse.de>
14047 * graphite-scop-detection.c (find_scop_parameters): Move
14048 loop bound handling ...
14049 (gather_bbs::before_dom_children): ... here, avoiding the need
14050 to build scop_info->loop_nest.
14051 (record_loop_in_sese): Remove.
14052 * sese.h (sese_info_t::loop_nest): Remove.
14053 * sese.c (new_sese_info): Do not allocate loop_nest.
14054 (free_sese_info): Do not free loop_nest.
14056 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14059 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14060 lhs from calls if the lhs has addressable type.
14062 2017-09-27 Richard Biener <rguenther@suse.de>
14064 * graphite.h (scop::max_alias_set): New member.
14065 * graphite-scop-detection.c: Remove references to non-existing
14066 --param in comments.
14067 (build_alias_sets): Record the maximum alias set used for drs.
14068 (build_scops): Support zero as unlimited for
14069 --param graphite-max-arrays-per-scop.
14070 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14071 and inline into ...
14072 (build_poly_sr_1): ... here. Compute alias set based on the
14073 maximum alias set used for drs rather than
14074 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14076 2017-09-27 Richard Biener <rguenther@suse.de>
14078 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14079 --param loop-block-tile-size=0 to disable tiling.
14081 2017-09-27 Richard Biener <rguenther@suse.de>
14083 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14084 (graphite-max-nb-scop-params): Document special value zero.
14085 * domwalk.h (dom_walker::STOP): New symbolical constant.
14086 (dom_walker::dom_walker): Add optional parameter for bb to
14088 (dom_walker::~dom_walker): Declare.
14089 (dom_walker::before_dom_children): Document STOP return value.
14090 (dom_walker::m_user_bb_to_rpo): New member.
14091 (dom_walker::m_bb_to_rpo): Likewise.
14092 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14093 mapping here if not provided by the user.
14094 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14095 provided by the user.
14096 (dom_walker::STOP): Define.
14097 (dom_walker::walk): Do not compute bb to RPO mapping here.
14098 Support STOP return value from before_dom_children to stop
14100 * graphite-optimize-isl.c (optimize_isl): If the schedule
14101 is the same still generate code if -fgraphite-identity
14102 or -floop-parallelize-all are given.
14103 * graphite-scop-detection.c: Include cfganal.h.
14104 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14106 (gather_bbs::before_dom_children): Return STOP for BBs
14108 (build_scops): Compute bb to RPO mapping and pass it to
14109 the domwalk. Treat --param graphite-max-nb-scop-params=0
14110 as not limiting the number of params.
14111 * graphite.c (graphite_initialize): Remove limit on the
14112 number of basic-blocks in a function.
14113 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14114 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14115 default value of 10.
14117 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14119 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14120 Adjust code to eliminate needing to do the shift right 32-bits
14121 operation after XSCVDPSPN.
14123 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14125 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14126 ((X / Y) != 0 -> X >= Y): Likewise.
14128 2017-09-26 Carl Love <cel@us.ibm.com>
14130 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14131 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14132 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14133 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14134 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14135 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14136 definitions and overloading.
14137 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14138 statement for P9V_BUILTIN_XST_LEN_R.
14139 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14140 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14141 define_expand and define_insn for the instructions and builtins.
14142 * doc/extend.texi: Update the built-in documentation file for the new
14143 built-in functions.
14144 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14145 define_insn for the instructions
14147 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14150 * gcc/config/netbsd-protos.h: New file.
14151 * gcc/config/netbsd.c: New file.
14152 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14153 * gcc/config/t-netbsd: New file.
14154 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14155 (tmake_file) Add t-netbsd.
14156 (extra_objs) Add netbsd.o.
14158 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14162 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14164 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14166 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14167 sign extension from a vector register to a GPR by doing a 32-bit
14168 direct move and then an EXTSW.
14169 (extendsi<mode>2 splitter): Likewise.
14170 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14171 right or vector extract after doing XSCVDPSPN. Use
14172 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14174 (movdi_from_sf_zero_ext): Likewise.
14175 (reload_gpr_from_vsxsf): Likewise.
14176 (p8_mfvsrd_4_disf): Delete, no longer used.
14177 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14178 then needing to move the SFmode to a GPR to use the XSCVDPSP
14179 instruction instead of FRSP and XSCVDPSPN.
14180 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14181 it is adjacent to the other XSCVSPDP insns.
14182 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14183 SFmode to be in traditional Altivec registers.
14184 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14185 (vsx_xscvspdpn): Likewise.
14186 (vsx_xscvspdpn_scalar): Likewise.
14188 2017-09-26 Martin Jambor <mjambor@suse.cz>
14190 * tree-sra.c (compare_access_positions): Put integral types first,
14191 stabilize sorting of integral types, remove conditions putting
14192 non-full-precision integers last.
14193 (sort_and_splice_var_accesses): Disable scalarization if a
14194 non-integert would be represented by a non-full-precision integer.
14196 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14198 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14199 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14200 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14201 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14202 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14203 conditionals inside the function instead of around it. Call
14204 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14205 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14207 2017-09-26 Richard Biener <rguenther@suse.de>
14209 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14211 (scop_detection::build_scop_breadth): ... this. Removed.
14212 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14213 (scop_detection::harmful_stmt_in_bb): Likewise.
14214 (scop_detection::graphite_can_represent_stmt): Likewise.
14215 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14216 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14217 (scop_detection::can_represent_loop_1): ... this. Removed.
14218 (scop_detection::harmful_loop_in_region): Simplify after inlining
14219 the above and remove more quadraticness.
14220 (build_scops): Adjust.
14221 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14224 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14227 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14228 REGNO (base) == SP_REG if base is a REG.
14230 PR middle-end/35691
14231 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14232 if it is different SSA_NAME.
14233 (optimize_range_tests_cmp_bitwise): New function.
14234 (optimize_range_tests): Call it.
14236 2017-09-26 Richard Biener <rguenther@suse.de>
14238 PR tree-optimization/82321
14239 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14240 for the def being inside the loop.
14242 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14244 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14246 * config/s390/s390-builtins.def: Fix constraint on op4.
14248 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14250 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14251 independent expanders.
14252 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14253 ("vec_ordered", "vec_unordered"): New expanders.
14255 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14257 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14260 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14262 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14263 vec_unpacks_lo_v16qi.
14264 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14266 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14268 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14269 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14270 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14272 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14274 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14276 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14278 ("*vec_slb<mode>"): New insn pattern.
14279 ("vec_shr_<mode>"): New expander.
14280 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14281 and force the shift count operand to V16QImode.
14282 ("vec_srb<mode>"): Set shift count mode to V16QI.
14284 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14286 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14287 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14288 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14290 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14293 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14295 2017-09-26 Richard Biener <rguenther@suse.de>
14297 PR tree-optimization/82320
14298 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14301 2017-09-25 Jeff Law <law@redhat.com>
14303 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14304 prototype for new argument.
14305 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14306 mostly extracted from rs6000_emit_allocate_stack.
14307 (rs6000_emit_probe_stack_range_stack_clash): New function.
14308 (rs6000_emit_allocate_stack): Call
14309 rs6000_emit_probe_stack_range_stack_clash as needed.
14310 (rs6000_emit_probe_stack_range): Add additional argument
14311 to call to gen_probe_stack_range{si,di}.
14312 (output_probe_stack_range): New.
14313 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14314 (output_probe_stack_range_stack_clash): New.
14315 (rs6000_emit_prologue): Emit notes into dump file as requested.
14316 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14317 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14318 Add additional operand and pass it to output_probe_stack_range.
14320 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14322 PR tree-optimization/82163
14323 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14324 (checking_verify_loop_closed_ssa): New parameter.
14325 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14326 (check_loop_closed_ssa_stmt): Delete.
14327 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14328 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14329 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14332 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14334 * brig-builtins.def: Treat HSAIL barrier builtins as
14335 setjmp/longjump style functions.
14337 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14339 * target.def (constant_alignment): New hook.
14340 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14341 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14342 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14343 * doc/tm.texi: Regenerate.
14344 * targhooks.h (default_constant_alignment): Declare.
14345 (constant_alignment_word_strings): Likewise.
14346 * targhooks.c (default_constant_alignment): New function.
14347 (constant_alignment_word_strings): Likewise.
14348 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14349 instead of CONSTANT_ALIGNMENT.
14350 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14351 (force_const_mem): Likewise.
14352 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14353 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14354 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14355 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14356 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14358 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14359 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14360 constant_alignment_word_strings.
14361 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14362 (CONSTANT_ALIGNMENT): Likewise.
14363 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14364 (arm_constant_alignment): New function.
14365 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14366 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14367 constant_alignment_word_strings.
14368 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14369 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14370 constant_alignment_word_strings.
14371 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14372 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14373 (cris_constant_alignment): New function.
14374 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14375 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14376 (epiphany_constant_alignment): New function.
14377 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14378 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14379 constant_alignment_word_strings.
14380 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14381 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14382 constant_alignment_word_strings.
14383 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14384 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14385 constant_alignment_word_strings.
14386 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14387 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14388 * config/i386/i386.c (ix86_constant_alignment): Make static.
14389 Use the same interface as the target hook.
14390 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14391 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14392 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14393 constant_alignment_word_strings.
14394 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14395 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14396 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14397 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14398 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14399 constant_alignment_word_strings.
14400 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14401 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14402 constant_alignment_word_strings.
14403 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14404 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14405 constant_alignment_word_strings.
14406 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14407 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14409 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14410 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14411 * config/mips/mips.c (mips_constant_alignment): New function.
14412 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14413 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14414 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14415 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14416 (mmix_constant_alignment): Make static. Use the same interface
14417 as the target hook.
14418 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14419 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14420 constant_alignment_word_strings.
14421 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14422 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14423 constant_alignment_word_strings.
14424 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14425 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14426 constant_alignment_word_strings.
14427 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14428 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14429 (rs6000_constant_alignment): New function.
14430 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14431 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14432 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14433 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14434 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14435 (rs6000_constant_alignment): New function.
14436 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14437 * config/s390/s390.c (s390_constant_alignment): New function.
14438 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14439 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14440 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14441 constant_alignment_word_strings.
14442 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14443 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14444 (sparc_constant_alignment): New function.
14445 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14446 * config/spu/spu.c (spu_constant_alignment): New function.
14447 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14448 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14449 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14450 constant_alignment_word_strings.
14451 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14452 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14453 constant_alignment_word_strings.
14454 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14455 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14456 constant_alignment_word_strings.
14457 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14458 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14459 (visium_constant_alignment): New function.
14460 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14461 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14462 (xtensa_constant_alignment): New function.
14463 * system.h (CONSTANT_ALIGNMENT): Poison.
14465 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14468 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14469 (rs6000_builtin_valid_without_lhs): New helper function.
14470 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14471 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14473 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14475 * target.h (vec_perm_indices): Use unsigned short rather than
14477 (auto_vec_perm_indices): Likewise.
14478 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14479 Use unsigned int rather than unsigned char.
14480 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14482 2017-09-25 Richard Biener <rguenther@suse.de>
14484 * cfgloop.h (sort_sibling_loops): Declare.
14485 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14486 (sort_sibling_loops): New function sorting the sibling loop list
14488 * graphite.c (graphite_transform_loops): Sort sibling loops.
14490 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14492 * target.def (vec_perm_const_ok): Change sel parameter to
14494 * optabs-query.c (can_vec_perm_p): Update accordingly.
14495 * doc/tm.texi: Regenerate.
14496 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14497 auto_vec_perm_indices and remove separate nelt field.
14498 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14499 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14500 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14501 (aarch64_expand_vec_perm_const): Update accordingly.
14502 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14503 to vec_perm_indices.
14504 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14505 auto_vec_perm_indices and remove separate nelt field.
14506 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14507 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14508 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14510 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14511 to vec_perm_indices.
14512 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14513 sel to vec_perm_indices.
14514 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14515 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14516 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14518 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14521 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14524 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14525 on the FUNCTION_DECL function context if it has a DIE that is a
14528 2017-09-25 Richard Biener <rguenther@suse.de>
14530 PR tree-optimization/82285
14531 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14534 2017-09-25 Tom de Vries <tom@codesourcery.com>
14538 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14541 2017-09-25 Richard Biener <rguenther@suse.de>
14543 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14544 ISL errors other than isl_error_quota happen. Dump if the
14545 schedule is the same.
14546 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14547 errors instead of aborting inside ISL.
14549 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14552 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14553 of libgcc_eh for m64.
14554 * config/i386/darwin64.h: Likewise.
14556 2017-09-25 Richard Biener <rguenther@suse.de>
14558 PR middle-end/82144
14559 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14560 attribute for incomplete types nor twice for complete ones.
14562 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14565 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14566 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14569 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14572 * config/i386/i386.opt (mprefer-avx256): Use
14573 ix86_target_flags variable.
14574 * config/i386/i386.c (ix86_target_string): Move
14575 -mprefer-avx256 to flag2_opts.
14577 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14579 PR middle-end/35691
14580 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14581 and x != -1 | y != -1 into (x & y) != -1.
14583 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14585 * config.gcc: Add new case statement to set
14586 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14587 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14588 s390x-*-linux* case statements. Added aarch64 to the list of
14589 supported architectures.
14591 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14593 PR tree-optimization/82289
14594 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14595 STMT_VINFO_RELEVANT_P.
14597 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14602 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14603 (extract_range_from_binary_expr_1): Don't call
14604 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14606 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14607 Alan Hayward <alan.hayward@arm.com>
14608 David Sherwood <david.sherwood@arm.com>
14610 * target.def (preferred_vector_alignment): New hook.
14611 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14613 * doc/tm.texi: Regenerate.
14614 * targhooks.h (default_preferred_vector_alignment): Declare.
14615 * targhooks.c (default_preferred_vector_alignment): New function.
14616 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14618 (DR_TARGET_ALIGNMENT): New macro.
14619 (aligned_access_p): Update commentary.
14620 (vect_known_alignment_in_bytes): New function.
14621 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14623 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14624 Calculate the misalignment based on the target alignment rather than
14626 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14627 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14628 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14629 the target alignment, rather than masking the element misalignment
14630 with the number of elements in a vector. Also use the target
14631 alignment when calculating the maximum number of peels.
14632 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14633 instead of TYPE_ALIGN_UNIT.
14634 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14635 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14636 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14637 (vect_create_data_ref_ptr): Likewise.
14638 (vect_setup_realignment): Realign by ANDing with
14639 -DR_TARGET_MISALIGNMENT.
14640 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14641 the number of peels based on DR_TARGET_ALIGNMENT.
14642 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14643 with the guaranteed alignment boundary when deciding whether
14645 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14646 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14647 (ensure_base_align): Remove stmt_info parameter. Get the
14648 target base alignment from DR_TARGET_ALIGNMENT.
14649 (vectorizable_store): Update call accordingly. Interpret
14650 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14652 (vectorizable_load): Likewise.
14654 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14655 Alan Hayward <alan.hayward@arm.com>
14656 David Sherwood <david.sherwood@arm.com>
14658 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14659 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14660 (vect_enhance_data_refs_alignment): Likewise.
14662 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14664 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14665 error. Only quit immediately if parsing is complete.
14666 (BEGIN): Initialize fatal_err and parse_done.
14667 (begin fpu, end fpu): Check number of arguments.
14668 (begin arch, end arch): Likewise.
14669 (begin cpu, end cpu): Likewise.
14670 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14671 (optalias): Likewise.
14673 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14675 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14676 * config/arm/arm-isa.h: Delete. Move definitions to ...
14677 * arm-cpus.in: ... here. Use new feature and fgroup values.
14678 * config/arm/arm.c (arm_option_override): Use lower case for feature
14680 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14681 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14682 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14684 (print_isa_bits_for): New function.
14685 (gen_isa): New function.
14686 (gen_comm_data): Use print_isa_bits_for.
14687 (define feature): New keyword.
14688 (define fgroup): New keyword.
14689 * config/arm/t-arm (TM_H): Remove.
14690 (GTM_H): Add arm-isa.h.
14691 (arm-isa.h): Add rule to generate file.
14692 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14693 case for feature bit names.
14695 2017-09-22 Richard Biener <rguenther@suse.de>
14697 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14699 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14700 print of no dependency loops ...
14701 * graphite.c (graphite_transform_loops): ... here.
14702 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14704 (same_close_phi_node, remove_duplicate_close_phi,
14705 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14706 (canonicalize_loop_closed_ssa): ... here and simplify.
14707 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14708 (optimize_isl): Use dump_printf_loc to tell when we stopped
14709 optimizing because of an ISL timeout.
14711 2017-09-22 Richard Biener <rguenther@suse.de>
14713 PR tree-optimization/82291
14714 * tree-if-conv.c (predicate_mem_writes): Make sure to
14715 remove writes in blocks predicated with false.
14717 2017-09-22 Richard Biener <rguenther@suse.de>
14719 * sese.c: Include cfganal.h.
14720 (if_region_set_false_region): Remove.
14721 (create_if_region_on_edge): Likewise.
14722 (move_sese_in_condition): Re-implement without destroying
14725 2017-09-22 Richard Biener <rguenther@suse.de>
14727 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14728 Verify both BBs contain loop PHI nodes before dispatching to
14729 copy_loop_phi_args.
14730 (graphite_regenerate_ast_isl): Do not recompute dominators,
14731 do not verify three times. Restructure for clarity.
14732 * graphite-scop-detection.c (same_close_phi_node,
14733 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14734 defined_in_loop_p, canonicalize_loop_closed_ssa,
14735 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14736 checking and SSA rewrite, move to ...
14737 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14738 (graphite_initialize): Do not pass in ctx, do not reset the
14739 SCEV cache, compute only dominators.
14740 (graphite_transform_loops): Allocate ISL ctx after
14741 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14742 Maintain post-dominators only around build_scops.
14743 * sese.c (if_region_set_false_region): Make static. Free
14744 and recompute dominators.
14745 (move_sese_in_condition): Assert we don't get called with
14746 post-dominators computed.
14747 * sese.h (if_region_set_false_region): Remove.
14749 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14751 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14752 mode attribute for TARGET_AVX512VL.
14754 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14756 * config/i386/i386.opt (mprefer-avx256): New option.
14757 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14759 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14760 for TARGET_PREFER_AVX256.
14761 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14763 2017-09-21 Jeff Law <law@redhat.com>
14765 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14766 Fix dump output if the only stack space is for pushed registers.
14768 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14770 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14773 2017-09-21 Martin Sebor <msebor@redhat.com>
14776 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14777 code (in C++) or code that triggers warnings.
14779 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14781 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14783 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14785 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14786 * sched-rgn.c: Ditto.
14787 * sel-sched-ir.c: Ditto.
14789 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14791 * toplev.h (set_random_seed): Adjust return type.
14792 * toplev.c (init_local_tick): Move eager initialization of random_seed
14793 to get_random_seed. Adjust comment.
14794 (init_random_seed): Inline to get_random_seed, delete.
14795 (get_random_seed): Initialize random_seed lazily.
14796 (set_random_seed): Do not return previous value.
14797 (print_switch_value): Do not call get_random_seed.
14799 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14801 * cgraph.c (delete_function_version): New, broken out from...
14802 (cgraph_node::delete_function_version): ...here. Rename to
14803 cgraph_node::delete_function_version_by_decl. Update all uses.
14804 (cgraph_node::remove): Call delete_function_version.
14806 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14809 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14810 VAR_DECLs to which addressable non-volatile parameters are mapped
14811 and for id->retvar after the return value assignment. Clear
14812 id->retval and id->retbnd after inlining.
14814 2017-09-21 Richard Biener <rguenther@suse.de>
14816 PR tree-optimization/82276
14817 PR tree-optimization/82244
14818 * tree-vrp.c (build_assert_expr_for): Set
14819 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14821 (remove_range_assertions): Revert earlier change.
14823 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14826 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14828 2017-09-21 Richard Biener <rguenther@suse.de>
14830 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14831 Restore valid IL after code generation errors.
14832 * graphite.c (graphite_transform_loops): Diagnose code
14833 generation issues as MSG_MISSED_OPTIMIZATION and continue
14834 with processing SCOPs.
14836 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14837 Alan Hayward <alan.hayward@arm.com>
14838 David Sherwood <david.sherwood@arm.com>
14840 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14841 rather than choosing between plus_constant and gen_rtx_<CODE>.
14842 * expr.c (emit_push_insn): Likewise.
14843 (expand_expr_real_2): Likewise.
14845 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14846 Alan Hayward <alan.hayward@arm.com>
14847 David Sherwood <david.sherwood@arm.com>
14849 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14851 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14852 Alan Hayward <alan.hayward@arm.com>
14853 David Sherwood <david.sherwood@arm.com>
14855 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14856 calling tree_to_uhwi.
14858 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14859 Alan Hayward <alan.hayward@arm.com>
14860 David Sherwood <david.sherwood@arm.com>
14862 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14863 INTEGER_CST rather than a negative test for ADDR_EXPR.
14865 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14866 Alan Hayward <alan.hayward@arm.com>
14867 David Sherwood <david.sherwood@arm.com>
14869 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14870 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14872 2017-09-21 Richard Biener <rguenther@suse.de>
14874 PR tree-optimization/71351
14875 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14876 graphite_create_new_loop_guard): Remove, fold remaining parts
14878 (translate_isl_ast_node_for): ... here and simplify.
14880 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14883 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14884 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14885 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14886 alternative always use QI mode, for -Os imov (=R,R) alternative
14887 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14890 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14891 Jeff Law <law@redhat.com>
14893 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14894 (allocate_stack_space): New function, partially extracted from
14895 s390_emit_prologue.
14896 (s390_emit_prologue): Track offset to most recent stack probe.
14897 Code to allocate space moved into allocate_stack_space.
14898 Dump actions when no stack is allocated.
14899 (s390_prologue_plus_offset): New function.
14900 (s390_emit_stack_probe): Likewise.
14902 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14904 * common.opt (Wa, Wl, Wp, g, gz=): Add
14906 (gno-column-info): Remove.
14907 (gcolumn-info): Drop RejectNegative.
14908 (gno-): New prefix.
14909 (gno-record-gcc-switches): Remove.
14910 (grecord-gcc-switches): Drop RejectNegative.
14911 (gno-split-dwarf): Remove.
14912 (gsplit-dwarf): Drop RejectNegative.
14913 (gno-strict-dwarf): Remove.
14914 (gstrict-dwarf): Drop RejectNegative.
14915 * config/darwin.opt (gfull, gused): Add RejectNegative.
14916 * dwarf2out.c (gen_producer_string): Drop
14917 gno-record-gcc-switches handler.
14918 * optc-gen.awk: Add g to prefixes with negative forms.
14919 * opts-common.c (remapping_prefix_p): New.
14920 (find_opt): Check it.
14921 (generate_canonical_option): Test g prefix.
14922 (option_map): Add -gno- mapping.
14923 (add_misspelling_candidates): Check remapping_prefix_p.
14925 2017-09-20 Jeff Law <law@redhat.com>
14927 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14928 thinko in stack clash protection support.
14930 * explow.c (compute_stack_clash_protection_loop_data): Use
14931 CONST_INT_P instead of explicit test. Verify object is a
14932 CONST_INT_P before looking at INTVAL.
14933 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14934 instead of explicit test.
14936 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14939 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14940 address instead of to r1 and r11.
14942 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14944 * config.gcc: Support "knm".
14945 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14946 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14948 * config/i386/i386.c (m_KNM): Define.
14949 (processor_target_table): Add "knm".
14951 (ix86_option_override_internal): Add "knm".
14952 (ix86_issue_rate): Add PROCESSOR_KNM.
14953 (ix86_adjust_cost): Ditto.
14954 (ia32_multipass_dfa_lookahead): Ditto.
14955 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14956 (fold_builtin_cpu): Add M_INTEL_KNM.
14957 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14958 (processor_type): Add PROCESSOR_KNM.
14959 * config/i386/x86-tune.def: Add m_KNM.
14960 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14962 2017-09-20 Richard Biener <rguenther@suse.de>
14964 PR tree-optimization/80213
14965 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14966 are allowed in empty BBs as well.
14967 (canonicalize_loop_closed_ssa): Also look for other complex
14969 (scop_detection::get_sese): Include the loop-closed PHI block
14971 (scop_detection::merge_sese): Remove code adding extra blocks.
14972 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14973 (build_scops): Assert the final returned scop is invalid.
14975 2017-09-20 Richard Biener <rguenther@suse.de>
14977 PR tree-optimization/82264
14978 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14980 (vn_phi_lookup): Likewise.
14981 (vn_phi_insert): Likewise.
14983 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14985 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14986 that fits into uhwi or shwi, add DW_AT_const_value regardless
14987 of early_dwarf without going through RTL, using add_AT_unsigned
14990 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14991 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14992 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14993 (DEBUG_MACRO_SECTION): Likewise.
14994 (DEBUG_LTO_MACRO_SECTION): Likewise.
14995 (DEBUG_STR_DWO_SECTION): Likewise.
14996 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14997 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14998 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14999 (DEBUG_LTO_LINE_STR_SECTION): Define.
15000 (init_sections_and_labels): Initialize debug_line_str_section
15001 variable. Initialize debug_loc_section for -gdwarf-5 to
15002 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15004 2017-09-20 Richard Biener <rguenther@suse.de>
15006 * graphite-sese-to-poly.c (extract_affine): Properly handle
15007 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15009 2017-09-20 Richard Biener <rguenther@suse.de>
15011 PR tree-optimization/81373
15012 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15013 Force SESE live-out defs to be handled even if they are
15016 2017-09-19 Jeff Law <law@redhat.com>
15018 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15019 nothing for stack adjustments with REG_STACK_CHECK.
15020 * sched-deps.c (parse_add_or_inc): Reject insns with
15021 REG_STACK_CHECK from dependency breaking.
15022 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15023 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15024 * reg-notes.def (STACK_CHECK): New note.
15026 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15027 (ix86_expand_prologue): Dump stack clash info as needed.
15028 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15030 * function.c (dump_stack_clash_frame_info): New function.
15031 * function.h (dump_stack_clash_frame_info): Prototype.
15032 (enum stack_clash_probes): New enum.
15034 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15035 flag_stack_clash_protection.
15036 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15037 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15038 (arm_frame_pointer_required): Likewise.
15039 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15040 (ia64_expand_prologue): Likewise.
15041 * config/mips/mips.c (mips_expand_prologue): Likewise.
15042 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15043 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15044 (sparc_flat_expand_prologue): Likewise.
15045 * config/spu/spu.c (spu_expand_prologue): Likewise.
15047 * explow.c: Include "params.h".
15048 (anti_adjust_stack_and_probe_stack_clash): New function.
15049 (get_stack_check_protect): Likewise.
15050 (compute_stack_clash_protection_loop_data): Likewise.
15051 (emit_stack_clash_protection_loop_start): Likewise.
15052 (emit_stack_clash_protection_loop_end): Likewise.
15053 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15054 Use anti_adjust_stack_and_probe_stack_clash.
15055 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15056 (emit_stack_clash_protection_loop_start): Likewise.
15057 (emit_stack_clash_protection_loop_end): Likewise.
15058 * rtl.h (get_stack_check_protect): Prototype.
15059 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15060 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15061 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15063 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15065 * doc/tm.texi: Rebuilt.
15066 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15067 get_stack_check_protect.
15068 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15069 * config/arm/arm.c (arm_expand_prologue): Likewise.
15070 (arm_frame_pointer_required): Likewise.
15071 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15072 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15073 * config/mips/mips.c (mips_expand_prologue): Likewise.
15074 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15075 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15076 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15077 (sparc_flat_expand_prologue): Likewise.
15079 * common.opt (-fstack-clash-protection): New option.
15080 * flag-types.h (enum stack_check_type): Note difference between
15081 -fstack-check= and -fstack-clash-protection.
15082 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15083 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15084 * toplev.c (process_options): Issue warnings/errors for cases
15085 not handled with -fstack-clash-protection.
15086 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15087 (-fstack-check): Note additional problem with -fstack-check=generic.
15088 Note that -fstack-check is primarily for Ada and refer users
15089 to -fstack-clash-protection for stack-clash-protection.
15090 Document new params for stack clash protection.
15092 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15094 * config/i386/i386.c (ix86_split_long_move): Do not handle
15095 address used for LEA in a special way.
15097 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15099 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15101 2017-09-19 Martin Sebor <msebor@redhat.com>
15104 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15105 of incompatible types.
15107 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15109 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15110 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15111 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15112 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15114 2017-09-19 Richard Biener <rguenther@suse.de>
15116 PR tree-optimization/82244
15117 * tree-vrp.c (remove_range_assertions): Do not propagate
15118 a constant to abnormals but replace the assert with a copy.
15120 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15122 PR rtl-optimization/57878
15123 PR rtl-optimization/68988
15124 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15125 avoidance test involving non_reload_pseudos. Move frequency test
15126 below the general fragmentation avoidance test.
15128 2017-09-19 Richard Biener <rguenther@suse.de>
15130 PR tree-optimization/69728
15131 * graphite-sese-to-poly.c (schedule_error): New global.
15132 (add_loop_schedule): Handle empty domain by failing the
15134 (build_original_schedule): Handle schedule_error.
15136 2017-09-19 Richard Biener <rguenther@suse.de>
15138 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15139 Do not iterate to sibling loops but only to siblings of inner
15142 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15145 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15146 that the registers are different.
15148 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15150 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15151 to processor_model and "amdfam17h" to arch_names_table.
15152 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15154 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15157 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15159 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15160 Alan Hayward <alan.hayward@arm.com>
15161 David Sherwood <david.sherwood@arm.com>
15163 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15165 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15167 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15168 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15169 vect_schedule_slp_instance.
15170 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15171 Update call to vect_slp_analyze_node_operations. Simplify return
15173 (vect_slp_analyze_bb_1): Update call accordingly.
15174 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15175 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15176 (vect_schedule_slp): Update call accordingly.
15178 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15179 Alan Hayward <alan.hayward@arm.com>
15180 David Sherwood <david.sherwood@arm.com>
15182 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15183 with types that aren't in fact scalar.
15185 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15187 * tree-vect-slp.c (vect_record_max_nunits): New function,
15189 (vect_build_slp_tree_1): ...here.
15190 (vect_build_slp_tree_2): Call it for phis too.
15192 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15194 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15195 to vect_get_vec_def_for_operand when getting the mask operand.
15197 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15198 Alan Hayward <alan.hayward@arm.com>
15199 David Sherwood <david.sherwood@arm.com>
15201 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15204 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15205 Alan Hayward <alan.hayward@arm.com>
15206 David Sherwood <david.sherwood@arm.com>
15208 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15209 calculation for vector booleans.
15211 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15212 Alan Hayward <alan.hayward@arm.com>
15213 David Sherwood <david.sherwood@arm.com>
15215 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15217 (vect_transform_stmt): ...here.
15218 (vect_analyze_stmt): Use it instead of calling
15219 vectorizable_live_operation directly.
15221 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15223 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15224 non-SIMT targets in acc vector loops.
15226 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15228 * configure.ac: Add arc and check if assembler supports gdwarf2.
15229 * configure: Regenerate.
15231 2017-09-18 Richard Biener <rguenther@suse.de>
15233 PR tree-optimization/82220
15234 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15235 epilogue niters from the min_profitable_iters compute.
15237 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15240 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15241 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15242 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15244 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15247 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15248 switching to a new text section.
15250 2017-09-18 Richard Biener <rguenther@suse.de>
15252 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15254 (build_alias_set): Reject aliases with no access function.
15256 2017-09-18 Richard Biener <rguenther@suse.de>
15258 PR tree-optimization/79622
15259 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15261 (build_cross_bb_scalars_use): Likewise.
15263 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15265 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15267 2017-09-18 Alan Modra <amodra@gmail.com>
15270 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15271 stack_pointer_rtx for count 0. Update comments. Break up
15272 large rtl expression.
15274 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15276 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15277 Increase to 20 bytes.
15278 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15279 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15280 or avx version of the stub.
15282 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15285 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15286 compute the minimum stack alignment. Also update preferred stack
15287 boundary for leaf functions.
15289 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15291 PR tree-optimization/82228
15292 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15295 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15297 * common/config/nds32/nds32-common.c
15298 (nds32_option_optimization_table): Refine formatting.
15299 (nds32_option_optimization_table): Use -fsched-pressure and
15300 -fomit-frame-pointer for specific optimization level.
15302 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15304 * config/nds32/nds32.c: Refine formatting and comments.
15305 * config/nds32/nds32.h: Likewise.
15306 * config/nds32/nds32.md: Likewise.
15307 * config/nds32/nds32-cost.c: Likewise.
15308 * config/nds32/nds32-isr.c: Likewise.
15309 * config/nds32/nds32-md-auxiliary.c: Likewise.
15310 * config/nds32/nds32-multiple.md: Likewise.
15311 * config/nds32/nds32-predicates.c: Likewise.
15313 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15314 Jakub Jelinek <jakub@redhat.com>
15316 Add support for -std=c++2a.
15317 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15319 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15321 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15324 * doc/extend.texi (Common Function Attributes): Add
15325 references to ARM, AArch64, and S/390 specific attributes.
15326 (Function Specific Option Pragmas): Add AArch64 and S/390
15327 to list of back ends that support the target pragma.
15329 2017-09-15 Nathan Sidwell <nathan@acm.org>
15331 * doc/standards.texi: Fix C++17 description. Update URLs for
15334 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15336 * common.opt (Wcast-align=strict): New warning option.
15337 * doc/invoke.texi: Document -Wcast-align=strict.
15339 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15341 * cgraph.h (cgraph_thunk_info): Add comments.
15342 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15343 assert for VIRTUAL_* arguments stricter.
15345 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15347 PR tree-optimization/71026
15348 * match.pd: Move RDIV patterns from fold-const.c
15349 * fold-const.c (distribute_real_division): Removed.
15350 (fold_binary_loc): Remove calls to distribute_real_divison.
15352 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15354 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15355 c++1z and gnu++1z as deprecated. Change other references to
15356 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15357 Change -Wc++1z-compat to -Wc++17-compat.
15358 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15359 * dwarf2out.c (highest_c_language): Handle C++17.
15360 (gen_compile_unit_die): Likewise.
15362 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15364 PR rtl-optimization/82192
15365 * combine.c (make_extraction): Don't look through non-paradoxical
15366 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15369 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15370 Alan Hayward <alan.hayward@arm.com>
15371 David Sherwood <david.sherwood@arm.com>
15373 * target.def (function_arg_offset): New hook.
15374 * targhooks.h (default_function_arg_offset): Declare.
15375 * targhooks.c (default_function_arg_offset): New function.
15376 * function.c (locate_and_pad_parm): Use
15377 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15378 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15379 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15380 * doc/tm.texi: Regenerate.
15381 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15382 * config/spu/spu.c (spu_function_arg_offset): New function.
15383 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15384 * system.h (FUNCTION_ARG_OFFSET): Poison.
15386 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15387 Alan Hayard <alan.hayward@arm.com>
15388 David Sherwood <david.sherwood@arm.com>
15390 * target.def (truly_noop_truncation): New hook.
15391 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15392 than TRULY_NOOP_TRUNCATION.
15393 * hooks.h (hook_bool_uint_uint_true): Declare.
15394 * hooks.c (hook_bool_uint_uint_true): New function.
15395 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15396 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15397 * doc/tm.texi: Regenerate.
15398 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15399 rather than TRULY_NOOP_TRUNCATION in comments.
15400 (simplify_comparison): Likewise.
15401 (record_truncated_value): Likewise.
15402 * expmed.c (extract_bit_field_1): Likewise.
15403 (extract_split_bit_field): Likewise.
15404 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15405 instead of TRULY_NOOP_TRUNCATION.
15406 * function.c (assign_parm_setup_block): Likewise.
15407 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15408 * rtlhooks.c: Include target.h.
15409 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15410 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15411 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15412 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15413 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15414 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15415 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15416 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15417 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15418 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15419 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15420 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15421 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15422 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15423 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15424 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15425 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15426 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15427 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15428 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15429 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15430 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15431 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15432 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15433 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15434 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15435 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15436 rather than TRULY_NOOP_TRUNCATION in comments.
15437 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15438 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15439 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15440 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15441 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15442 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15443 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15444 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15445 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15446 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15447 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15448 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15449 rather than TRULY_NOOP_TRUNCATION in comments.
15450 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15451 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15452 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15453 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15454 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15455 TRULY_NOOP_TRUNCATION condition.
15456 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15457 (TRULY_NOOP_TRUNCATION): Delete.
15458 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15459 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15460 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15461 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15462 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15463 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15464 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15465 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15466 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15467 rather than TRULY_NOOP_TRUNCATION in comments.
15468 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15469 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15470 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15471 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15472 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15473 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15475 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15478 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15479 (*cmp_ior): Likewise.
15480 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15481 (*ior_scc_scc_cmp): Likewise.
15482 (*and_scc_scc): Likewise.
15483 (*and_scc_scc_cmp): Likewise.
15485 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15486 Alan Hayard <alan.hayward@arm.com>
15487 David Sherwood <david.sherwood@arm.com>
15489 * target.def (can_change_mode_class): New hook.
15490 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15491 (hard_regno_nregs): Likewise.
15492 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15493 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15494 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15495 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15496 (LOAD_EXTEND_OP): Update accordingly.
15497 * doc/tm.texi: Regenerate.
15498 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15499 CANNOT_CHANGE_MODE_CLASS.
15500 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15501 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15502 * combine.c (simplify_set): Update accordingly.
15503 * emit-rtl.c (validate_subreg): Likewise.
15504 * recog.c (general_operand): Likewise.
15505 * regcprop.c (mode_change_ok): Likewise.
15506 * reload1.c (choose_reload_regs): Likewise.
15507 (inherit_piecemeal_p): Likewise.
15508 * rtlanal.c (simplify_subreg_regno): Likewise.
15509 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15510 instead of CANNOT_CHANGE_MODE_CLASS.
15511 (reload_cse_simplify_operands): Likewise.
15512 * reload.c (push_reload): Use targetm.can_change_mode_class
15513 instead of CANNOT_CHANGE_MODE_CLASS.
15514 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15515 REG_CANNOT_CHANGE_MODE_P.
15516 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15517 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15518 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15519 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15520 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15521 (arm_can_change_mode_class): New function.
15522 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15523 than CANNOT_CHANGE_MODE_CLASS in comments.
15524 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15525 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15526 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15527 (ix86_can_change_mode_class): ...this new function, inverting the
15528 sense of the return value.
15529 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15530 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15531 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15532 (ia64_can_change_mode_class): New function.
15533 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15534 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15535 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15536 (m32c_can_change_mode_class): ...this new function, inverting the
15537 sense of the return value.
15538 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15539 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15540 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15541 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15542 (mips_can_change_mode_class): ...this new function, inverting the
15543 sense of the return value.
15544 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15545 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15546 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15547 (msp430_can_change_mode_class): New function.
15548 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15549 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15550 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15551 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15552 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15553 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15554 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15555 (pa_cannot_change_mode_class): Replace with...
15556 (pa_can_change_mode_class): ...this new function, inverting the
15557 sense of the return value.
15558 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15559 than CANNOT_CHANGE_MODE_CLASS in comments.
15560 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15561 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15562 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15563 (pdp11_cannot_change_mode_class): Replace with...
15564 (pdp11_can_change_mode_class): ...this new function, inverting the
15565 sense of the return value.
15566 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15567 * config/powerpcspe/powerpcspe-protos.h
15568 (rs6000_cannot_change_mode_class_ptr): Delete.
15569 * config/powerpcspe/powerpcspe.c
15570 (rs6000_cannot_change_mode_class_ptr): Delete.
15571 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15572 (rs6000_option_override_internal): Assign to
15573 targetm.can_change_mode_class instead of
15574 rs6000_cannot_change_mode_class_ptr.
15575 (rs6000_cannot_change_mode_class): Replace with...
15576 (rs6000_can_change_mode_class): ...this new function, inverting the
15577 sense of the return value.
15578 (rs6000_debug_cannot_change_mode_class): Replace with...
15579 (rs6000_debug_can_change_mode_class): ...this new function.
15580 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15581 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15582 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15583 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15584 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15586 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15587 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15588 (rs6000_option_override_internal): Assign to
15589 targetm.can_change_mode_class instead of
15590 rs6000_cannot_change_mode_class_ptr.
15591 (rs6000_cannot_change_mode_class): Replace with...
15592 (rs6000_can_change_mode_class): ...this new function, inverting the
15593 sense of the return value.
15594 (rs6000_debug_cannot_change_mode_class): Replace with...
15595 (rs6000_debug_can_change_mode_class): ...this new function.
15596 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15597 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15598 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15599 (s390_can_change_mode_class): ...this new function, inverting the
15600 sense of the return value.
15601 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15602 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15603 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15604 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15605 (sh_cannot_change_mode_class): Replace with...
15606 (sh_can_change_mode_class): ...this new function, inverting the
15607 sense of the return value.
15608 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15609 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15610 (sparc_can_change_mode_class): New function.
15611 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15612 * config/spu/spu.c (spu_can_change_mode_class): New function.
15613 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15614 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15615 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15616 (visium_can_change_mode_class): New function.
15617 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15619 2017-09-15 Richard Biener <rguenther@suse.de>
15621 PR tree-optimization/82217
15622 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15623 but not undefined case.
15625 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15628 * postreload.c (reload_cse_simplify_operands): Skip
15629 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15631 2017-09-15 Richard Biener <rguenther@suse.de>
15633 PR tree-optimization/68823
15634 * graphite-scop-detection.c (build_alias_set): If we have a
15635 possible dependence check whether we can handle them by just
15636 looking at the DRs DR_ACCESS_FNs.
15637 (build_scops): If build_alias_set fails, fail the SCOP.
15639 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15641 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15642 to support float128 built-in functions that require the ISA 3.0
15644 (BU_FLOAT128_3_HW): Likewise.
15645 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15646 built-in functions.
15647 (FMAF128): Likewise.
15649 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15650 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15651 floating point instructions.
15652 (rs6000_invalid_builtin): Likewise.
15653 (rs6000_builtin_mask_names): Likewise.
15654 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15655 (RS6000_BTM_FLOAT128_HW): Likewise.
15656 (RS6000_BTM_COMMON): Likewise.
15657 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15659 * doc/extend.texi (RS/6000 built-in functions): Document the
15660 IEEE 128-bit floating point square root and fused multiply-add
15661 built-in functions.
15663 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15665 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15666 reg (r2) isn't in the set of registers defined in the prologue.
15668 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15669 Alan Hayward <alan.hayward@arm.com>
15670 David Sherwood <david.sherwood@arm.com>
15672 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15673 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15674 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15675 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15676 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15678 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15679 max_vectorization_factor.
15680 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15682 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15683 Alan Hayward <alan.hayward@arm.com>
15684 David Sherwood <david.sherwood@arm.com>
15686 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15687 (vect_worthwhile_without_simd_p): Declare.
15688 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15689 (vectorizable_reduction): Use it.
15690 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15691 (vectorizable_operation): Likewise.
15693 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15694 Alan Hayward <alan.hayward@arm.com>
15695 David Sherwood <david.sherwood@arm.com>
15697 * tree-vectorizer.h (vect_get_num_copies): New function.
15698 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15699 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15700 (vectorizable_induction): Likewise.
15701 (vectorizable_live_operation): Likewise.
15702 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15703 (vectorizable_bswap): Likewise.
15704 (vectorizable_call): Likewise.
15705 (vectorizable_conversion): Likewise.
15706 (vectorizable_assignment): Likewise.
15707 (vectorizable_shift): Likewise.
15708 (vectorizable_operation): Likewise.
15709 (vectorizable_store): Likewise.
15710 (vectorizable_load): Likewise.
15711 (vectorizable_condition): Likewise.
15712 (vectorizable_comparison): Likewise.
15713 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15715 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15716 Alan Hayward <alan.hayward@arm.com>
15717 David Sherwood <david.sherwood@arm.com>
15719 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15720 of vect_init_vector.
15722 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15723 Alan Hayward <alan.hayward@arm.com>
15724 David Sherwood <david.sherwood@arm.com>
15726 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15727 an inline wrapper that provides a location.
15728 (gimple_build_vector): Likewise.
15729 * gimple-fold.c (gimple_build_vector_from_val): New function.
15730 (gimple_build_vector): Likewise.
15731 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15732 functions to build the initial value. Always return a gimple value.
15733 (get_initial_defs_for_reduction): Likewise. Only compute
15735 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15736 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15737 (vectorizable_induction): Use gimple_build_vector rather than
15740 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15741 Alan Hayward <alan.hayward@arm.com>
15742 David Sherwood <david.sherwood@arm.com>
15744 * target.h (vec_perm_indices): New typedef.
15745 (auto_vec_perm_indices): Likewise.
15746 * optabs-query.h: Include target.h
15747 (can_vec_perm_p): Take a vec_perm_indices *.
15748 * optabs-query.c (can_vec_perm_p): Likewise.
15749 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15750 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15751 * tree-vect-generic.c (lower_vec_perm): Likewise.
15752 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15753 (vect_grouped_load_supported): Likewise.
15754 (vect_shift_permute_load_chain): Likewise.
15755 (vect_permute_store_chain): Use auto_vec_perm_indices.
15756 (vect_permute_load_chain): Likewise.
15757 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15758 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15759 Update uses of can_vec_perm_p.
15760 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15761 mode with a number of elements. Take a vec_perm_indices *.
15762 (vect_create_epilog_for_reduction): Update accordingly.
15763 Use auto_vec_perm_indices.
15764 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15765 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15766 (vect_transform_slp_perm_load): Likewise.
15767 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15768 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15769 (vect_gen_perm_mask_checked): Likewise.
15770 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15771 (vect_gen_perm_mask_checked): Likewise.
15772 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15773 (vectorizable_store): Likewise.
15774 (vectorizable_load): Likewise.
15775 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15776 (vectorizable_bswap): Likewise.
15778 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15779 Alan Hayward <alan.hayward@arm.com>
15780 David Sherwood <david.sherwood@arm.com>
15782 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15783 * tree.c (build_vector): Likewise.
15784 (build_vector_from_ctor): Update accordingly.
15785 (build_vector_from_val): Likewise.
15786 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15787 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15788 * tree-vect-generic.c (add_rshift): Likewise.
15789 (expand_vector_divmod): Likewise.
15790 (optimize_vector_constructor): Likewise.
15791 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15792 (vect_transform_slp_perm_load): Likewise.
15793 (vect_schedule_slp_instance): Likewise.
15794 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15795 (vectorizable_call): Likewise.
15796 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15797 * expmed.c (make_tree): Likewise.
15798 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15799 a vector passed to build_vector.
15800 (fold_convert_const): Likewise.
15801 (exact_inverse): Likewise.
15802 (fold_ternary_loc): Likewise.
15803 (fold_relational_const): Likewise.
15804 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15805 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15806 (const_unop): Likewise. Store the reduction accumulator in a
15807 variable rather than an array.
15808 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15809 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15810 the new vector, rather than constructing it after the input arrays.
15811 (native_interpret_vector): Use auto_vec<tree> when building
15812 a vector passed to build_vector. Add elements in order.
15813 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15814 auto_vec<tree> when building a vector passed to build_vector.
15815 (vect_create_epilog_for_reduction): Likewise.
15816 (vectorizable_induction): Likewise.
15817 (get_initial_def_for_reduction): Likewise. Fix indentation of
15819 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15821 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15822 passed to build_vector.
15824 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15825 Alan Hayward <alan.hayward@arm.com>
15826 David Sherwood <david.sherwood@arm.com>
15828 * tree-core.h (tree_base::u): Add an "nelts" field.
15829 (tree_vector): Use VECTOR_CST_NELTS as the length.
15830 * tree.c (tree_size): Likewise.
15831 (make_vector): Initialize VECTOR_CST_NELTS.
15832 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15833 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15834 TYPE_VECTOR_SUBPARTS.
15835 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15836 as the number of units, setting it from VECTOR_CST_NELTS.
15837 (const_vector_from_tree): Likewise.
15838 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15839 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15840 (fold_negate_expr_1): Likewise.
15841 (fold_convert_const): Likewise.
15842 (const_binop): Likewise. Differentiate the number of output and
15844 (const_unop): Likewise.
15845 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15846 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15847 in cases that did the opposite.
15849 2017-09-14 Richard Biener <rguenther@suse.de>
15851 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15854 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15856 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15858 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15861 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15862 if and where to split a bb, except for splitting before debug insn
15863 sequences followed by non-label real insn. Delete debug insns
15864 in between basic blocks.
15866 * combine.c (make_compound_operation_int): Formatting fixes.
15868 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15869 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15870 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15871 * config/sol2.h (LINK_EH_SPEC): Likewise.
15872 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15873 * config/s390/linux.h (LINK_SPEC): Likewise.
15874 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15875 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15876 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15877 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15878 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15879 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15880 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15881 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15883 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15885 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15887 (ENDFILE_LINUX_SPEC): Likewise.
15888 (LINK_EH_SPEC): Likewise.
15889 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15890 (LINK_OS_LINUX_SPEC32): Likewise.
15891 (LINK_OS_LINUX_SPEC64): Likewise.
15892 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15893 (LINK_OS_LINUX_SPEC): Likewise.
15895 2017-09-13 Martin Liska <mliska@suse.cz>
15897 PR middle-end/82154
15898 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15899 CASE_HIGH is NULL_TREE.
15901 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15902 Alan Hayward <alan.hayward@arm.com>
15903 David Sherwood <david.sherwood@arm.com>
15905 * target.def (secondary_memory_needed): New hook.
15906 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15907 instead of SECONDARY_MEMORY_NEEDED.
15908 (secondary_memory_needed_mode): Likewise.
15909 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15910 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15911 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15912 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15913 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15914 * doc/tm.texi: Regenerate.
15915 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15916 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15917 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15918 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15919 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15920 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15921 mode argument first and change the reg_class arguments to reg_class_t.
15922 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15923 Make static. Update the call to inline_secondary_memory_needed.
15924 (ix86_register_move_cost): Update the call to
15925 inline_secondary_memory_needed.
15926 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15927 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15929 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15930 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15932 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15933 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15934 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15935 and match hook interface. Add comment from mips.h.
15936 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15937 * config/mmix/mmix.md (truncdfsf2): Refer to
15938 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15940 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15941 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15942 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15943 (pa_secondary_memory_needed): New function.
15944 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15945 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15946 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15947 (pdp11_secondary_memory_needed): Make static and match hook interface.
15948 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15949 * config/powerpcspe/powerpcspe-protos.h
15950 (rs6000_secondary_memory_needed_ptr): Delete.
15951 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15953 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15954 (rs6000_option_override_internal): Assign to
15955 targetm.secondary_memory_needed rather than
15956 rs6000_secondary_memory_needed_ptr.
15957 (rs6000_secondary_memory_needed): Match hook interface.
15958 (rs6000_debug_secondary_memory_needed): Likewise.
15959 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15960 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15961 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15962 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15963 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15964 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15966 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15967 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15968 (rs6000_option_override_internal): Assign to
15969 targetm.secondary_memory_needed rather than
15970 rs6000_secondary_memory_needed_ptr.
15971 (rs6000_secondary_memory_needed): Match hook interface.
15972 (rs6000_debug_secondary_memory_needed): Likewise.
15973 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15974 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15975 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15976 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15977 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15978 (sparc_secondary_memory_needed): New function.
15979 * lra-constraints.c (check_and_process_move): Refer to
15980 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15982 (curr_insn_transform): Likewise.
15983 (process_alt_operands): Use targetm.secondary_memory_needed
15984 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15985 (check_secondary_memory_needed_p): Likewise.
15986 (choose_split_class): Likewise.
15987 * reload.c: Unconditionally include code that was previously
15988 conditional on SECONDARY_MEMORY_NEEDED.
15989 (push_secondary_reload): Use targetm.secondary_memory_needed
15990 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15991 (push_reload): Likewise.
15992 * reload1.c: Unconditionally include code that was previously
15993 conditional on SECONDARY_MEMORY_NEEDED.
15994 (choose_reload_regs): Use targetm.secondary_memory_needed
15995 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15996 (gen_reload): Likewise.
15997 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15999 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16000 Alan Hayward <alan.hayward@arm.com>
16001 David Sherwood <david.sherwood@arm.com>
16003 * target.def (secondary_memory_needed_mode): New hook:
16004 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16005 * targhooks.h (default_secondary_memory_needed_mode): New function.
16006 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16007 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16008 * doc/tm.texi: Regenerate.
16009 * lra-constraints.c (check_and_process_move): Use
16010 targetm.secondary_memory_needed_mode instead of
16011 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16012 (curr_insn_transform): Likewise.
16013 * reload.c (get_secondary_mem): Likewise.
16014 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16015 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16017 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16018 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16019 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16020 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16021 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16023 * config/powerpcspe/powerpcspe-protos.h
16024 (rs6000_secondary_memory_needed_mode): Delete.
16025 * config/powerpcspe/powerpcspe.c
16026 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16027 (rs6000_secondary_memory_needed_mode): Make static.
16028 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16029 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16031 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16033 (rs6000_secondary_memory_needed_mode): Make static.
16034 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16035 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16036 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16037 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16038 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16040 (sparc_secondary_memory_needed_mode): New function.
16041 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16043 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16045 * config/aarch64/constraints.md (Umq): New constraint.
16046 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16048 (mov<mode>): Update condition.
16050 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16052 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16053 when bitposition is the same.
16055 2017-09-13 Richard Biener <rguenther@suse.de>
16057 * dwarf2out.c (output_die_symbol): Remove.
16058 (output_die): Do not output a DIEs symbol.
16060 2017-09-13 Richard Biener <rguenther@suse.de>
16062 PR middle-end/82128
16063 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16064 default-def to avoid breaking iterator update with the weird
16065 interaction with cgraph_update_edges_for_call_stmt_node.
16067 2017-09-13 Richard Biener <rguenther@suse.de>
16069 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16070 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16071 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16072 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16074 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16076 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16077 Disable pc relative literal load irrespective of
16078 TARGET_FIX_ERR_A53_84341 for default.
16080 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16082 * config/sparc/sparc.c (output_return): Output the source location of
16083 the insn in the delay slot, if any.
16084 (output_sibcall): Likewise.
16086 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16089 * common.opt (-static-pie): New alias.
16090 (shared): Negate static-pie.
16091 (-no-pie): Update help text.
16093 (static-pie): New option.
16094 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16095 -static-pie support.
16096 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16097 (LINK_EH_SPEC): Likewise.
16098 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16099 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16100 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16101 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16102 (init_gcc_specs): Likewise.
16103 (init_spec): Likewise.
16104 (display_help): Update help message for -pie.
16105 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16108 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16110 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16111 (movdi_aarch64): Likewise.
16112 (movti_aarch64): Likewise.
16114 2017-09-12 Simon Wright <simon@pushface.org>
16117 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16118 calculation of the minor version, always output as 0.
16120 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16123 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16124 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16125 on it early, rather than manual conversion late. For
16126 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16127 instead of performing manual conversion.
16129 2017-09-12 Carl Love <cel@us.ibm.com>
16131 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16132 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16134 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16135 VMULOSW): Add definitions.
16136 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16137 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16138 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16139 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16140 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16142 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16144 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16146 (movti_aarch64): Likewise.
16147 (movdf_aarch64): Likewise.
16148 (movtf_aarch64): Likewise.
16149 (load_pairdi): Likewise.
16150 (store_pairdi): Likewise.
16151 (load_pairdf): Likewise.
16152 (store_pairdf): Likewise.
16153 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16154 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16155 (ldr_got_small_<mode>): Likewise.
16156 (ldr_got_small_28k_<mode>): Likewise.
16157 (ldr_got_tiny): Likewise.
16158 * config/aarch64/iterators.md (ldst_sz): New.
16159 (ldpstp_sz): Likewise.
16160 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16162 (thunderx_load): Split load_8 to load_16.
16163 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16165 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16166 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16167 (xgene1_store_pair): Split store_8 to store_16.
16168 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16169 (falkor_st_0_st_sd): Split store_8 to store_16.
16171 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16173 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16174 and store1/2/3/4 to store_4/8/12/16.
16175 * config/aarch64/aarch64.md: Update for rename.
16176 * config/arm/arm.md: Likewise.: Likewise.
16177 * config/arm/arm.c: Likewise.
16178 * config/arm/thumb1.md: Likewise.
16179 * config/arm/thumb2.md: Likewise.
16180 * config/arm/vfp.md: Likewise.
16181 * config/arm/arm-generic.md: Likewise.
16182 * config/arm/arm1020e.md: Likewise.
16183 * config/arm/arm1026ejs.md: Likewise.
16184 * config/arm/arm1136jfs.md: Likewise.
16185 * config/arm/arm926ejs.md: Likewise.
16186 * config/arm/cortex-a15.md: Likewise.
16187 * config/arm/cortex-a17.md: Likewise.
16188 * config/arm/cortex-a5.md: Likewise.
16189 * config/arm/cortex-a53.md: Likewise.
16190 * config/arm/cortex-a57.md: Likewise.
16191 * config/arm/cortex-a7.md: Likewise.
16192 * config/arm/cortex-a8.md: Likewise.
16193 * config/arm/cortex-a9.md: Likewise.
16194 * config/arm/cortex-m4.md: Likewise.
16195 * config/arm/cortex-m7.md: Likewise.
16196 * config/arm/cortex-r4.md: Likewise.
16197 * config/arm/exynos-m1.md: Likewise.
16198 * config/arm/fa526.md: Likewise.
16199 * config/arm/fa606te.md: Likewise.
16200 * config/arm/fa626te.md: Likewise.
16201 * config/arm/fa726te.md: Likewise.
16202 * config/arm/fmp626.md: Likewise.
16203 * config/arm/iwmmxt.md: Likewise.
16204 * config/arm/ldmstm.md: Likewise.
16205 * config/arm/marvell-pj4.md: Likewise.
16206 * config/arm/xgene1.md: Likewise.
16207 * config/aarch64/thunderx.md: Likewise.
16208 * config/aarch64/thunderx2t99.md: Likewise.
16209 * config/aarch64/falkor.md: Likewise.
16211 2017-09-12 Martin Liska <mliska@suse.cz>
16213 * attribs.c (private_lookup_attribute): New function.
16214 * attribs.h (private_lookup_attribute): Declared here.
16215 (lookup_attribute): Called from this place.
16217 2017-09-12 Richard Biener <rguenther@suse.de>
16219 PR tree-optimization/82157
16220 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16221 stmts with side-effects.
16223 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16224 Alan Hayward <alan.hayward@arm.com>
16225 David Sherwood <david.sherwood@arm.com>
16227 * target.def (hard_regno_nregs): New hook.
16228 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16229 * targhooks.h (default_hard_regno_nregs): Declare.
16230 * targhooks.c (default_hard_regno_nregs): New function.
16231 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16232 (TARGET_HARD_REGNO_NREGS): ...this hook.
16233 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16234 (CLASS_MAX_NREGS): Likewise.
16235 * doc/tm.texi: Regenerate.
16236 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16237 instead of HARD_REGNO_NREGS.
16238 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16239 HARD_REGNO_NREGS in the comment.
16240 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16241 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16242 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16243 Return an unsigned int.
16244 (TARGET_HARD_REGNO_NREGS): Redefine.
16245 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16246 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16247 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16248 (arc_hard_regno_nregs): New function.
16249 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16250 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16251 (arm_hard_regno_nregs): New function.
16252 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16253 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16254 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16255 (TARGET_HARD_REGNO_NREGS): Redefine.
16256 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16257 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16258 (HARD_REGNO_NREGS): Delete.
16259 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16260 (cr16_hard_regno_nregs): New function.
16261 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16262 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16263 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16264 (cris_hard_regno_nregs): New function.
16265 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16266 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16267 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16268 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16269 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16270 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16271 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16272 (frv_hard_regno_nregs): Make static. Take and return an
16274 (frv_class_max_nregs): Remove outdated copy of documentation.
16275 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16276 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16277 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16278 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16279 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16280 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16281 (TARGET_HARD_REGNO_NREGS): Redefine.
16282 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16283 (CLASS_MAX_NREGS): Update comment.
16284 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16285 (ia64_hard_regno_nregs): New function.
16286 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16287 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16288 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16289 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16290 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16292 (m32c_hard_regno_nregs): Likewise. Make static.
16293 (TARGET_HARD_REGNO_NREGS): Redefine.
16294 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16295 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16296 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16297 (m68k_hard_regno_nregs): New function.
16298 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16299 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16300 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16301 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16302 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16303 Take and return an unsigned int.
16304 (TARGET_HARD_REGNO_NREGS): Redefine.
16305 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16306 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16307 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16308 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16309 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16310 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16311 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16312 (msp430_hard_regno_nregs): Make static. Take and return an
16314 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16315 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16316 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16317 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16318 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16319 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16320 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16321 (TARGET_HARD_REGNO_NREGS): Redefine.
16322 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16323 (PA_HARD_REGNO_NREGS): ...this.
16324 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16325 (PA_HARD_REGNO_NREGS): ...this.
16326 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16327 (pa_hard_regno_nregs): New function.
16328 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16329 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16330 (pdp11_hard_regno_nregs): New function.
16331 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16332 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16333 (rs6000_hard_regno_nregs_hook): New function.
16334 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16335 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16336 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16337 Take and return an unsigned int. Move earlier in file.
16338 (TARGET_HARD_REGNO_NREGS): Redefine.
16339 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16340 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16341 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16342 (rl78_hard_regno_nregs): Make static. Take and return an
16344 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16345 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16346 (rs6000_hard_regno_nregs_hook): New function.
16347 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16348 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16349 (TARGET_HARD_REGNO_NREGS): Redefine.
16350 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16351 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16352 instead of HARD_REGNO_NREGS.
16353 (s390_hard_regno_nregs): New function.
16354 (s390_hard_regno_mode_ok): Add comment from s390.h.
16355 (TARGET_HARD_REGNO_NREGS): Redefine.
16356 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16357 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16358 (sh_hard_regno_nregs): New function.
16359 (sh_pass_in_reg_p): Use it.
16360 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16361 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16362 (sparc_hard_regno_nregs): New function.
16363 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16364 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16365 (spu_function_arg_advance): Use it, supplying a valid register number.
16366 (TARGET_HARD_REGNO_NREGS): Redefine.
16367 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16368 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16369 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16370 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16371 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16372 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16373 (CLASS_MAX_NREGS): Remove copy of old documentation.
16374 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16375 (visium_hard_regno_nregs): New function.
16376 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16377 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16378 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16379 xtensa_hard_regno_nregs): New function.
16380 * system.h (HARD_REGNO_NREGS): Poison.
16382 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16384 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16385 hard_regno_nregs instead of HARD_REGNO_NREGS.
16386 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16387 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16388 (c6x_expand_epilogue): Likewise.
16389 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16390 (frv_read_iacc_argument): Likewise.
16391 * config/sh/sh.c: Include regs.h.
16392 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16393 (regs_used): Likewise.
16394 (output_stack_adjust): Likewise.
16395 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16396 * expmed.c: Include regs.h.
16397 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16398 * ree.c: Include regs.h.
16399 (combine_reaching_defs): Use hard_regno_nregs instead of
16401 (add_removable_extension): Likewise.
16403 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16405 * regs.h (hard_regno_nregs): Turn into a function.
16406 (end_hard_regno): Update accordingly.
16407 * caller-save.c (setup_save_areas): Likewise.
16408 (save_call_clobbered_regs): Likewise.
16409 (replace_reg_with_saved_mem): Likewise.
16410 (insert_restore): Likewise.
16411 (insert_save): Likewise.
16412 * combine.c (can_change_dest_mode): Likewise.
16413 (move_deaths): Likewise.
16414 (distribute_notes): Likewise.
16415 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16416 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16417 (rs6000_split_multireg_move): Likewise.
16418 (rs6000_register_move_cost): Likewise.
16419 (rs6000_memory_move_cost): Likewise.
16420 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16421 (rs6000_split_multireg_move): Likewise.
16422 (rs6000_register_move_cost): Likewise.
16423 (rs6000_memory_move_cost): Likewise.
16424 * cselib.c (cselib_reset_table): Likewise.
16425 (cselib_lookup_1): Likewise.
16426 * emit-rtl.c (set_mode_and_regno): Likewise.
16427 * function.c (aggregate_value_p): Likewise.
16428 * ira-color.c (setup_profitable_hard_regs): Likewise.
16429 (check_hard_reg_p): Likewise.
16430 (calculate_saved_nregs): Likewise.
16431 (assign_hard_reg): Likewise.
16432 (improve_allocation): Likewise.
16433 (calculate_spill_cost): Likewise.
16434 * ira-emit.c (modify_move_list): Likewise.
16435 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16436 (ira_hard_reg_in_set_p): Likewise.
16437 * ira.c (setup_reg_mode_hard_regset): Likewise.
16438 (clarify_prohibited_class_mode_regs): Likewise.
16439 (check_allocation): Likewise.
16440 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16441 (lra_setup_reg_renumber): Likewise.
16442 (setup_try_hard_regno_pseudos): Likewise.
16443 (spill_for): Likewise.
16444 (assign_hard_regno): Likewise.
16445 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16446 * lra-constraints.c (in_class_p): Likewise.
16447 (lra_constraint_offset): Likewise.
16448 (simplify_operand_subreg): Likewise.
16449 (lra_constraints): Likewise.
16450 (split_reg): Likewise.
16451 (split_if_necessary): Likewise.
16452 (invariant_p): Likewise.
16453 (inherit_in_ebb): Likewise.
16454 * lra-lives.c (process_bb_lives): Likewise.
16455 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16456 (get_hard_regs): Likewise.
16457 (do_remat): Likewise.
16458 * lra-spills.c (assign_spill_hard_regs): Likewise.
16459 * mode-switching.c (create_pre_exit): Likewise.
16460 * postreload.c (reload_combine_recognize_pattern): Likewise.
16461 * recog.c (peep2_find_free_register): Likewise.
16462 * regcprop.c (kill_value_regno): Likewise.
16463 (set_value_regno): Likewise.
16464 (copy_value): Likewise.
16465 (maybe_mode_change): Likewise.
16466 (find_oldest_value_reg): Likewise.
16467 (copyprop_hardreg_forward_1): Likewise.
16468 * regrename.c (check_new_reg_p): Likewise.
16469 (regrename_do_replace): Likewise.
16470 * reload.c (push_reload): Likewise.
16471 (combine_reloads): Likewise.
16472 (find_dummy_reload): Likewise.
16473 (operands_match_p): Likewise.
16474 (find_reloads): Likewise.
16475 (find_equiv_reg): Likewise.
16476 (reload_adjust_reg_for_mode): Likewise.
16477 * reload1.c (count_pseudo): Likewise.
16478 (count_spilled_pseudo): Likewise.
16479 (find_reg): Likewise.
16480 (clear_reload_reg_in_use): Likewise.
16481 (free_for_value_p): Likewise.
16482 (allocate_reload_reg): Likewise.
16483 (choose_reload_regs): Likewise.
16484 (reload_adjust_reg_for_temp): Likewise.
16485 (emit_reload_insns): Likewise.
16486 (delete_output_reload): Likewise.
16487 * rtlanal.c (subreg_get_info): Likewise.
16488 * sched-deps.c (sched_analyze_reg): Likewise.
16489 * sel-sched.c (init_regs_for_mode): Likewise.
16490 (mark_unavailable_hard_regs): Likewise.
16491 (choose_best_reg_1): Likewise.
16492 (verify_target_availability): Likewise.
16493 * valtrack.c (dead_debug_insert_temp): Likewise.
16494 * var-tracking.c (track_loc_p): Likewise.
16495 (emit_note_insn_var_location): Likewise.
16496 * varasm.c (make_decl_rtl): Likewise.
16497 * reginfo.c (choose_hard_reg_mode): Likewise.
16498 (init_reg_modes_target): Refer directly to
16499 this_target_regs->x_hard_regno_nregs.
16501 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16503 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16504 instead of hard_regno_nregs.
16506 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16508 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16509 end_hard_regno instead of hard_regno_nregs.
16510 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16511 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16512 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16513 * ira-color.c (improve_allocation): Likewise.
16514 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16515 * lra-lives.c (mark_regno_live): Likewise.
16516 (mark_regno_dead): Likewise.
16517 * lra-remat.c (operand_to_remat): Likewise.
16518 * lra.c (collect_non_operand_hard_regs): Likewise.
16519 * postreload.c (reload_combine_note_store): Likewise.
16520 (move2add_valid_value_p): Likewise.
16521 * reload.c (regno_clobbered_p): Likewise.
16523 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16525 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16527 * config/v850/v850.c (v850_reorg): Likewise.
16528 * reload.c (refers_to_regno_for_reload_p): Likewise.
16529 (find_equiv_reg): Likewise.
16530 * reload1.c (reload_reg_reaches_end_p): Likewise.
16532 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16534 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16536 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16537 * config/arm/arm.c (output_move_neon): Likewise.
16538 (arm_attr_length_move_neon): Likewise.
16539 (neon_split_vcombine): Likewise.
16540 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16541 (c6x_mark_reg_written): Likewise.
16542 (c6x_dwarf_register_span): Likewise.
16543 * config/i386/i386.c (ix86_save_reg): Likewise.
16544 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16545 (rws_access_reg): Likewise.
16546 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16547 * mode-switching.c (create_pre_exit): Likewise.
16548 * ree.c (combine_reaching_defs): Likewise.
16549 (add_removable_extension): Likewise.
16550 * regcprop.c (find_oldest_value_reg): Likewise.
16551 (copyprop_hardreg_forward_1): Likewise.
16552 * reload.c (reload_inner_reg_of_subreg): Likewise.
16553 (push_reload): Likewise.
16554 (combine_reloads): Likewise.
16555 (find_dummy_reload): Likewise.
16556 (reload_adjust_reg_for_mode): Likewise.
16557 * reload1.c (find_reload_regs): Likewise.
16558 (forget_old_reloads_1): Likewise.
16559 (reload_reg_free_for_value_p): Likewise.
16560 (reload_adjust_reg_for_temp): Likewise.
16561 (emit_reload_insns): Likewise.
16562 (delete_output_reload): Likewise.
16563 * sel-sched.c (choose_best_reg_1): Likewise.
16564 (choose_best_pseudo_reg): Likewise.
16566 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16567 Alan Hayward <alan.hayward@arm.com>
16568 David Sherwood <david.sherwood@arm.com>
16570 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16571 * target.def (slow_unaligned_access): New hook.
16572 * targhooks.h (default_slow_unaligned_access): Declare.
16573 * targhooks.c (default_slow_unaligned_access): New function.
16574 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16575 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16576 * doc/tm.texi: Regenerate.
16577 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16578 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16579 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16581 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16582 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16584 (rs6000_slow_unaligned_access): New function.
16585 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16586 (expand_block_compare): Likewise.
16587 (expand_strn_compare): Likewise.
16588 (rs6000_rtx_costs): Likewise.
16589 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16590 (riscv_slow_unaligned_access): Likewise.
16591 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16592 (riscv_slow_unaligned_access_p): ...this and make static.
16593 (riscv_option_override): Update accordingly.
16594 (riscv_slow_unaligned_access): New function.
16595 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16596 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16597 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16598 (rs6000_slow_unaligned_access): New function.
16599 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16600 (rs6000_rtx_costs): Likewise.
16601 * config/rs6000/rs6000-string.c (expand_block_compare)
16602 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16603 of SLOW_UNALIGNED_ACCESS.
16604 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16605 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16606 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16607 of SLOW_UNALIGNED_ACCESS.
16608 * expmed.c (simple_mem_bitfield_p): Likewise.
16609 * expr.c (alignment_for_piecewise_move): Likewise.
16610 (emit_group_load_1): Likewise.
16611 (emit_group_store): Likewise.
16612 (copy_blkmode_from_reg): Likewise.
16613 (emit_push_insn): Likewise.
16614 (expand_assignment): Likewise.
16615 (store_field): Likewise.
16616 (expand_expr_real_1): Likewise.
16617 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16618 * lra-constraints.c (simplify_operand_subreg): Likewise.
16619 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16620 * gimple-ssa-store-merging.c: Likewise in block comment at start
16622 * tree-ssa-strlen.c: Include target.h.
16623 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16624 of SLOW_UNALIGNED_ACCESS.
16625 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16627 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16629 PR rtl-optimization/82185
16630 * expmed.c (emit_store_flag_int): Only test tem if it has been
16633 2017-09-12 Richard Biener <rguenther@suse.de>
16635 PR middle-end/82149
16636 * match.pd ((FTYPE) N CMP CST): Fix typo.
16638 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16640 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16642 (mips_near_type_p): Add 'short_call' attribute as a synonym
16644 * doc/extend.texi (short_call): Document new function attribute.
16646 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16649 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16650 assertion check that in the condition.
16651 (get_atomic_generic_size): Likewise. Before testing if parameter
16652 has pointer type, if it has array type, call for C++
16653 default_conversion to perform array-to-pointer conversion.
16655 2017-09-12 Richard Biener <rguenther@suse.de>
16657 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16658 for operations we cannot scalarize.
16660 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16662 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16663 vectors heap vectors. Clean up comments.
16664 Make visited_bbs a reference.
16665 (profitable_jump_thread_path): Make GC
16666 vectors heap vectors. Clean up comments.
16668 (convert_and_register_jump_thread_path): Make GC vectors heap
16670 (check_subpath_and_update_thread_path): Same. Clean up comments.
16671 Make visited_bbs a reference.
16672 (handle_phi): Abstract common code to to
16673 register_jump_thread_path_if_profitable.
16674 Rename VAR_BB to DEF_BB.
16676 Make GC vectors heap vectors.
16677 Make visited_bbs a reference.
16678 (handle_assignment): Same.
16679 (register_jump_thread_path_if_profitable): New.
16680 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16682 Make GC vectors heap vectors. Clean up comments.
16683 Make visited_bbs a reference.
16684 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16685 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16688 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16691 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16692 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16694 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16696 Revert r251800 and r251799.
16698 2017-09-11 Martin Jambor <mjambor@suse.cz>
16701 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16702 arguments in advance.
16703 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16704 use it to find predecessor edges.
16705 (naive_outof_ssa): Collect vector of predecessors.
16707 2017-09-08 Jason Merrill <jason@redhat.com>
16709 PR c++/70029 - ICE with ref-qualifier and -flto
16710 * langhooks.h (struct lang_hooks_for_types): Add
16711 copy_lang_qualifiers.
16712 * attribs.c (build_type_attribute_qual_variant): Use it.
16713 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16715 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16716 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16718 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16721 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16722 (*mulsi3_sp64): New instruction.
16723 (mulsi3): New expander.
16725 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16727 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16729 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16731 * sancov.c: Include memmodel.h.
16733 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16736 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16739 2017-09-07 Carl Love <cel@us.ibm.com>
16741 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16742 the sldi instruction.
16744 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16746 * sancov.c: Include tm_p.h.
16748 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16751 * output.h (switch_to_other_text_partition): New declaration.
16752 * varasm.c (switch_to_other_text_partition): New function.
16753 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16754 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16755 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16756 to the other text partition before emitting LCL label and switch back
16757 after emitting the word after it.
16759 2017-09-07 Richard Biener <rguenther@suse.de>
16761 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16762 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16763 critical edges here, after loop init.
16764 (pass_data_pre): Remove PROP_no_crit_edges flags.
16765 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16766 for valueization of call args to avoid leaking VN_TOP.
16767 (visit_use): Assert we do not visit default defs.
16768 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16769 Use error_mark_node to more easily detect leaking VN_TOP.
16770 All default-defs are varying, not VN_TOP. Mark them visited.
16771 (run_scc_vn): Make code match comment.
16773 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16775 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16776 OPTION_MASK_FLOAT128_KEYWORD.
16777 (POWERPC_MASKS): Likewise.
16778 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16779 support for the -mfloat128-type option, and make -mfloat128
16780 default on PowerPC Linux systems. Define or undefine
16781 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16782 Define __float128 to be __ieee128 if IEEE 128-bit support is
16783 enabled, or undefine it.
16784 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16785 Delete defining __FLOAT128_TYPE__.
16786 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16787 -mfloat128-type option and make -mfloat128 default on PowerPC
16789 (TARGET_FLOAT128_TYPE): Likewise.
16790 (-mfloat128-type): Likewise.
16791 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16792 Delete the -mfloat128-type option and make -mfloat128 default on
16793 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16794 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16795 128-bit floating point is enabled. Change tests from using
16796 -mfloat128-type to -mfloat128.
16797 (rs6000_mangle_type): Use the correct mangling for the __float128
16798 type even if normal long double is restricted to 64-bits.
16799 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16801 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16802 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16803 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16805 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16806 documentation for -mfloat128.
16808 2017-09-06 Olivier Hainque <hainque@adacore.com>
16810 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16812 2017-09-06 Wish Wu <wishwu007@gmail.com>
16813 Jakub Jelinek <jakub@redhat.com>
16815 * asan.c (initialize_sanitizer_builtins): Add
16816 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16817 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16818 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16819 BT_FN_VOID_UINT64_PTR variables.
16820 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16821 (BT_FN_VOID_UINT16_UINT16): Likewise.
16822 (BT_FN_VOID_UINT32_UINT32): Likewise.
16823 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16824 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16825 (BT_FN_VOID_UINT64_PTR): Likewise.
16826 * common.opt (flag_sanitize_coverage): New variable.
16827 (fsanitize-coverage=trace-pc): Remove.
16828 (fsanitize-coverage=): Add.
16829 * flag-types.h (enum sanitize_coverage_code): New enum.
16830 * fold-const.c (fold_range_test): Disable non-short-circuit
16831 optimization if flag_sanitize_coverage.
16832 (fold_truth_andor): Likewise.
16833 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16834 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16835 (coverage_sanitizer_opts): New array.
16836 (get_closest_sanitizer_option): Add OPTS argument, handle also
16837 OPT_fsanitize_coverage_.
16838 (parse_sanitizer_options): Adjusted to also handle
16839 OPT_fsanitize_coverage_.
16840 (common_handle_option): Add OPT_fsanitize_coverage_.
16841 * sancov.c (instrument_comparison, instrument_switch): New function.
16842 (sancov_pass): Add trace-cmp support.
16843 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16844 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16845 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16846 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16847 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16848 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16849 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16850 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16851 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16852 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16854 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16856 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16857 error. Only quit immediately if parsing is complete.
16858 (BEGIN): Initialize fatal_err and parse_done.
16859 (begin fpu, end fpu): Check number of arguments.
16860 (begin arch, end arch): Likewise.
16861 (begin cpu, end cpu): Likewise.
16862 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16863 (optalias): Likewise.
16865 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16867 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16868 * config/arm/arm-isa.h: Delete. Move definitions to ...
16869 * arm-cpus.in: ... here. Use new feature and fgroup values.
16870 * config/arm/arm.c (arm_option_override): Use lower case for feature
16872 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16873 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16874 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16876 (print_isa_bits_for): New function.
16877 (gen_isa): New function.
16878 (gen_comm_data): Use print_isa_bits_for.
16879 (define feature): New keyword.
16880 (define fgroup): New keyword.
16881 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16882 (arm-isa.h): Add rule to generate file.
16883 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16884 case for feature bit names.
16886 2017-09-06 Richard Biener <rguenther@suse.de>
16888 * tree-ssa-pre.c (NECESSARY): Remove.
16889 (create_expression_by_pieces): Do not touch pass-local flags.
16890 (insert_into_preds_of_block): Likewise.
16891 (do_pre_regular_insertion): Likewise.
16892 (eliminate_insert): Likewise.
16893 (eliminate_dom_walker::before_dom_children): Likewise.
16894 (fini_eliminate): Do not look at inserted_exprs.
16895 (mark_operand_necessary): Remove.
16896 (remove_dead_inserted_code): Replace with simple work-list
16897 algorithm based on inserted_exprs and SSA uses.
16898 (pass_pre::execute): Re-order fini_eliminate and
16899 remove_dead_inserted_code.
16901 2017-09-06 Olivier Hainque <hainque@adacore.com>
16903 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16904 for VxWorks 7. Adjust surrounding comments.
16906 2017-09-06 Richard Biener <rguenther@suse.de>
16908 * gimple-ssa-strength-reduction.c
16909 (find_candidates_dom_walker::before_dom_children): Also allow
16912 2017-09-06 Richard Biener <rguenther@suse.de>
16914 PR tree-optimization/82108
16915 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16916 for gap in the non-permutation SLP case.
16918 2017-09-06 Martin Jambor <mjambor@suse.cz>
16920 PR tree-optimization/82078
16921 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16922 add_access_to_work_queue...
16923 (build_accesses_from_assign): ...here.
16924 (propagate_all_subaccesses): Make sure racc is the group
16925 representative, if there is one.
16927 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16929 PR middle-end/82095
16930 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16933 2017-09-06 Richard Biener <rguenther@suse.de>
16935 * gimple-ssa-strength-reduction.c
16936 (find_candidates_dom_walker::before_doom_children): Use a
16937 type and not a mode check.
16939 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16942 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16944 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16946 2017-09-05 Jeff Law <law@redhat.com>
16948 PR tree-optimization/64910
16949 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16950 cases where we have 3 or more operands.
16952 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16954 PR middle-end/81768
16955 * omp-low.c (lower_omp_for): Recompute tree invariant if
16956 gimple_omp_for_initial/final is ADDR_EXPR.
16958 PR middle-end/81768
16959 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16960 into gimple val before gimplification fo the COND_EXPR.
16962 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16964 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16965 REGION_COPY argument.
16966 (thread_through_all_blocks): Remove unused argument to
16967 duplicate_thread_path.
16969 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16970 Alan Hayward <alan.hayward@arm.com>
16971 David Sherwood <david.sherwood@arm.com>
16973 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16974 Take a scalar_mode rather than a machine_mode.
16975 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16976 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16977 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16978 (aarch64_gen_adjusted_ldpstp): Likewise.
16979 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16981 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16982 Alan Hayward <alan.hayward@arm.com>
16983 David Sherwood <david.sherwood@arm.com>
16985 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16986 Take a scalar_int_mode instead of a machine_mode.
16987 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16988 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16989 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16990 (aarch64_simd_attr_length_rglist): Delete.
16991 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16992 a scalar_int_mode instead of a machine_mode.
16993 (aarch64_add_offset): Likewise.
16994 (aarch64_internal_mov_immediate): Likewise
16995 (aarch64_add_constant_internal): Likewise.
16996 (aarch64_add_constant): Likewise.
16997 (aarch64_movw_imm): Likewise.
16998 (aarch64_rtx_arith_op_extract_p): Likewise.
16999 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17000 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17001 Remove assert that the mode isn't a vector.
17002 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17003 (aarch64_expand_mov_immediate): Update calls after above changes.
17004 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17005 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17006 (aarch64_move_imm): Likewise.
17007 (aarch64_can_const_movi_rtx_p): Likewise.
17008 (aarch64_strip_extend): Likewise.
17009 (aarch64_extr_rtx_p): Likewise.
17010 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17011 a CONST_INT when the mode parameter is VOIDmode.
17012 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17014 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17016 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17017 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17018 (bitwise_type_for_mode): Update accordingly.
17020 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17022 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17023 * stor-layout.c (mode_for_size_tree): Likewise.
17024 (mode_for_array): Update accordingly.
17025 (layout_decl): Likewise.
17026 (compute_record_mode): Likewise. Only set the mode once.
17028 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17030 * target.def (get_mask_mode): Change return type to opt_mode.
17032 * doc/tm.texi: Regenerate.
17033 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17034 * targhooks.c (default_get_mask_mode): Likewise.
17035 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17036 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17037 targetm.get_mask_mode.
17038 * tree.c (build_truth_vector_type): Likewise.
17040 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17042 * machmode.h (mode_for_vector): Return an opt_mode.
17043 * stor-layout.c (mode_for_vector): Likewise.
17044 (mode_for_int_vector): Update accordingly.
17045 (layout_type): Likewise.
17046 * config/i386/i386.c (emit_memmov): Likewise.
17047 (ix86_expand_set_or_movmem): Likewise.
17048 (ix86_expand_vector_init): Likewise.
17049 (ix86_get_mask_mode): Likewise.
17050 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17052 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17053 * expmed.c (extract_bit_field_1): Likewise.
17054 * expr.c (expand_expr_real_2): Likewise.
17055 * optabs-query.c (can_vec_perm_p): Likewise.
17056 (can_vec_mask_load_store_p): Likewise.
17057 * optabs.c (expand_vec_perm): Likewise.
17058 * targhooks.c (default_get_mask_mode): Likewise.
17059 * tree-vect-stmts.c (vectorizable_store): Likewise.
17060 (vectorizable_load): Likewise.
17061 (get_vectype_for_scalar_type_and_size): Likewise.
17063 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17065 * machmode.h (mode_for_int_vector): New function.
17066 * stor-layout.c (mode_for_int_vector): Likewise.
17067 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17068 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17069 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17070 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17071 (s390_expand_vcond): Likewise.
17073 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17075 * machmode.h (opt_machine_mode): New type.
17076 (opt_mode<T>): Allow construction from anything that can be
17078 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17079 (mode_for_size): Return an opt_machine_mode.
17080 * stor-layout.c (mode_for_size): Likewise.
17081 (mode_for_size_tree): Update call accordingly.
17082 (bitwise_mode_for_mode): Likewise.
17083 (make_fract_type): Likewise.
17084 (make_accum_type): Likewise.
17085 * caller-save.c (replace_reg_with_saved_mem): Update call
17087 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17088 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17089 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17090 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17091 * expmed.c (extract_bit_field_1): Likewise.
17092 * reload.c (get_secondary_mem): Likewise.
17093 * varasm.c (assemble_integer): Likewise.
17094 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17097 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17099 * machmode.h (decimal_float_mode_for_size): New function.
17100 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17101 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17102 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17103 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17104 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17105 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17107 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17109 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17110 (get_builtin_sync_mode): Likewise.
17111 (expand_ifn_atomic_compare_exchange): Likewise.
17112 (expand_builtin_atomic_clear): Likewise.
17113 (expand_builtin_atomic_test_and_set): Likewise.
17114 (fold_builtin_atomic_always_lock_free): Likewise.
17115 * calls.c (compute_argument_addresses): Likewise.
17116 (emit_library_call_value_1): Likewise.
17117 (store_one_arg): Likewise.
17118 * combine.c (combine_instructions): Likewise.
17119 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17120 * config/arm/arm.c (arm_function_value): Likewise.
17121 (aapcs_allocate_return_reg): Likewise.
17122 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17123 * config/i386/i386.c (construct_container): Likewise.
17124 (ix86_gimplify_va_arg): Likewise.
17125 (ix86_expand_sse_cmp): Likewise.
17126 (emit_memmov): Likewise.
17127 (emit_memset): Likewise.
17128 (expand_small_movmem_or_setmem): Likewise.
17129 (ix86_expand_pextr): Likewise.
17130 (ix86_expand_pinsr): Likewise.
17131 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17132 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17134 * config/mips/mips.c (mips_function_value_1) Likewise.
17135 (mips_block_move_straight): Likewise.
17136 (mips_expand_ins_as_unaligned_store): Likewise.
17137 * config/powerpcspe/powerpcspe.c
17138 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17139 (rs6000_darwin64_record_arg_flush): Likewise.
17140 * config/rs6000/rs6000.c
17141 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17142 (rs6000_darwin64_record_arg_flush): Likewise.
17143 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17144 (sparc_function_value_1): Likewise.
17145 * config/spu/spu.c (adjust_operand): Likewise.
17146 (spu_emit_branch_or_set): Likewise.
17147 (arith_immediate_p): Likewise.
17148 * emit-rtl.c (gen_lowpart_common): Likewise.
17149 * expr.c (expand_expr_real_1): Likewise.
17150 * function.c (assign_parm_setup_block): Likewise.
17151 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17152 * reload1.c (alter_reg): Likewise.
17153 * stor-layout.c (mode_for_vector): Likewise.
17154 (layout_type): Likewise.
17156 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17158 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17159 (spu_convert_move): Likewise.
17160 * lower-subreg.c (resolve_simple_move): Likewise.
17162 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17165 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17166 define_insn to a define_expand.
17167 (altivec_vsum2sws_direct): New define_insn.
17168 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17170 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17172 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17175 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17180 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17181 realign stack if stack alignment needed is less than incoming
17184 2017-09-05 Marek Polacek <polacek@redhat.com>
17187 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17190 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17192 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17194 2017-09-05 Richard Biener <rguenther@suse.de>
17196 PR tree-optimization/82084
17197 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17199 2017-09-05 Richard Biener <rguenther@suse.de>
17201 PR tree-optimization/82102
17202 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17204 2017-09-05 Martin Liska <mliska@suse.cz>
17206 PR tree-optimization/82032
17207 * tree-cfg.c (generate_range_test): New function.
17208 * tree-cfg.h (generate_range_test): Declared here.
17209 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17210 (cleanup_control_expr_graph): Use it.
17211 * tree-switch-conversion.c (try_switch_expansion): Remove
17213 (emit_case_nodes): Use generate_range_test.
17215 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17218 * config/i386/i386.md (*<btsc><mode>_mask): Add
17219 TARGET_USE_BT to insn constraint.
17220 (*btr<mode>_mask): Ditto.
17222 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17224 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17225 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17227 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17230 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17231 TARGET_NEON and TARGET_IWMMXT.
17232 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17233 TARGET_NEON and TARGET_IWMMXT.
17234 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17236 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17238 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17239 (ix86_rewrite_tls_address): Ditto.
17240 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17241 (ix86_rewrite_tls_address_1): Ditto.
17242 (ix86_rewrite_tls_address): Ditto.
17243 * config/i386/predicates.md (tls_address_pattern): New predicate.
17244 * config/i386/i386.md (TLS address splitter): New splitter.
17246 2017-09-04 Richard Biener <rguenther@suse.de>
17248 PR tree-optimization/82084
17249 * fold-const.h (can_native_encode_string_p): Declare.
17250 * fold-const.c (can_native_encode_string_p): Factor out from ...
17251 (native_encode_string): ... here.
17252 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17253 vectorizing stores from constants we later cannot handle.
17255 2017-09-04 Marek Polacek <polacek@redhat.com>
17258 * doc/invoke.texi: Update -Wtautological-compare documentation.
17260 2017-09-04 Jeff Law <law@redhat.com>
17262 PR tree-optimization/64910
17263 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17264 swap the first and last operand if the last is a constant.
17266 2017-09-04 Marek Polacek <polacek@redhat.com>
17269 * convert.c (do_narrow): When sanitizing signed integer overflows,
17270 bail out for signed types.
17271 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17273 2017-09-04 Richard Biener <rguenther@suse.de>
17275 PR tree-optimization/82060
17276 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17277 Move devirtualization after stmt folding and before EH/AB/noreturn
17278 cleanup to get the stmt refs canonicalized. Use a bool instead
17279 of gimple_modified_p since that doesn't work for NOPs. Schedule
17280 NOPs generated by folding for removal.
17282 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17283 Alan Hayward <alan.hayward@arm.com>
17284 David Sherwood <david.sherwood@arm.com>
17286 * coretypes.h (pad_direction): New enum.
17287 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17288 (FUNCTION_ARG_PADDING): Likewise.
17289 * target.def (function_arg_padding): New hook.
17290 * targhooks.h (default_function_arg_padding): Declare.
17291 * targhooks.c (default_function_arg_padding): New function.
17292 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17293 (TARGET_FUNCTION_ARG_PADDING): ...this.
17294 * doc/tm.texi: Regenerate.
17295 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17296 instead of direction.
17297 (compute_argument_addresses): Likewise.
17298 (load_register_parameters): Likewise.
17299 (emit_library_call_value_1): Likewise.
17300 (store_one_arg): Use targetm.calls.function_arg_padding instead
17301 of FUNCTION_ARG_PADDING.
17302 (must_pass_in_stack_var_size_or_pad): Likewise.
17303 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17304 (emit_group_store): Likewise.
17305 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17306 instead of FUNCTION_ARG_PADDING.
17307 (emit_push_insn): Likewise, and propagate enum change throughout
17309 * function.h (direction): Delete.
17310 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17312 * function.c (assign_parm_find_stack_rtl): Likewise.
17313 (assign_parm_setup_block_p): Likewise.
17314 (assign_parm_setup_block): Likewise.
17315 (gimplify_parameters): Likewise.
17316 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17317 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17319 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17320 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17321 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17322 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17323 (aarch64_function_arg_padding): ...this new function.
17324 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17325 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17326 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17327 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17328 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17329 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17330 (arm_pad_arg_upward): Replace with...
17331 (arm_function_arg_padding): ...this new function.
17332 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17334 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17335 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17336 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17337 (ia64_hpux_function_arg_padding): Replace with...
17338 (ia64_function_arg_padding): ...this new function. Use pad_direction
17339 instead of direction. Check for TARGET_HPUX.
17340 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17341 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17342 (iq2000_function_arg_padding): New function.
17343 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17344 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17345 (mips_function_arg_padding): ...this new function.
17346 (mips_pad_reg_upward): Update accordingly.
17347 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17348 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17349 targetm.calls.function_arg_padding.
17350 (FUNCTION_ARG_PADDING): Delete.
17351 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17352 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17353 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17354 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17355 (nios2_block_reg_padding): Return pad_direction instead of direction.
17356 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17357 instead of direction.
17358 (nios2_function_arg_padding): Likewise. Make static.
17359 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17360 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17361 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17362 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17363 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17364 (pa_function_arg_padding): Make static. Return pad_direction instead
17366 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17367 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17368 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17369 instead of direction. Use targetm.calls.function_arg_padding.
17370 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17371 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17372 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17373 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17374 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17376 (function_arg_padding): Rename to...
17377 (rs6000_function_arg_padding): ...this. Make static. Return
17378 pad_direction instead of direction.
17379 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17380 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17381 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17382 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17383 instead of direction. Use targetm.calls.function_arg_padding.
17384 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17385 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17386 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17387 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17388 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17389 (function_arg_padding): Rename to...
17390 (rs6000_function_arg_padding): ...this. Make static. Return
17391 pad_direction instead of direction.
17392 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17393 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17394 * config/s390/s390.c (s390_function_arg_padding): New function.
17395 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17396 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17397 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17398 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17399 (function_arg_padding): Rename to...
17400 (sparc_function_arg_padding): ...this. Make static. Return
17401 pad_direction instead of direction.
17402 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17403 * config/spu/spu.c (spu_function_arg_padding): New function.
17404 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17405 * system.h (FUNCTION_ARG_PADDING): Poison.
17407 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17408 Alan Hayward <alan.hayward@arm.com>
17409 David Sherwood <david.sherwood@arm.com>
17411 * target.def (modes_tieable_p): New hook.
17412 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17413 (TARGET_MODES_TIEABLE_P): ...this.
17414 * doc/tm.texi.in: Regenerate.
17415 * hooks.h (hook_bool_mode_mode_true): Declare.
17416 * hooks.c (hook_bool_mode_mode_true): New function.
17417 * combine.c (subst): Use targetm.modes_tieable_p instead of
17419 * dse.c (find_shift_sequence): Likewise.
17420 * expmed.c (extract_low_bits): Likewise.
17421 * lower-subreg.c: Include target.h.
17422 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17424 * rtlanal.c (rtx_cost): Likewise.
17425 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17426 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17427 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17428 (TARGET_MODES_TIEABLE_P): Redefine.
17429 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17430 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17431 (TARGET_MODES_TIEABLE_P): Redefine.
17432 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17433 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17434 (arc_modes_tieable_p): New function.
17435 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17436 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17437 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17438 (arm_modes_tieable_p): Make static.
17439 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17440 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17441 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17442 (TARGET_MODES_TIEABLE_P): Redefine.
17443 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17444 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17445 (TARGET_MODES_TIEABLE_P): Redefine.
17446 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17447 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17448 (cr16_modes_tieable_p): New function.
17449 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17450 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17451 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17452 (TRULY_NOOP_TRUNCATION): Update comment.
17453 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17454 (TRULY_NOOP_TRUNCATION): Update comment.
17455 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17456 (frv_modes_tieable_p): New function.
17457 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17458 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17459 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17460 (TARGET_MODES_TIEABLE_P): Redefine.
17461 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17462 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17463 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17464 (TARGET_MODES_TIEABLE_P): Redefine.
17465 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17466 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17467 (ia64_modes_tieable_p): New function.
17468 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17469 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17470 (iq2000_modes_tieable_p): New function.
17471 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17472 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17473 (lm32_modes_tieable_p): New function.
17474 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17475 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17476 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17477 (TARGET_MODES_TIEABLE_P): Redefine.
17478 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17479 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17480 (m32r_modes_tieable_p): New function.
17481 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17482 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17483 (m68k_modes_tieable_p): New function.
17484 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17485 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17486 (mcore_modes_tieable_p): New function.
17487 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17488 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17490 (TARGET_MODES_TIEABLE_P): Redefine.
17491 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17492 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17493 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17494 (TARGET_MODES_TIEABLE_P): Redefine.
17495 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17496 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17497 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17498 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17499 (mn10300_modes_tieable_p): ...this and make static.
17500 (TARGET_MODES_TIEABLE_P): Redefine.
17501 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17502 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17503 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17504 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17505 (msp430_modes_tieable_p): Make static.
17506 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17507 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17508 (TARGET_MODES_TIEABLE_P): Redefine.
17509 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17510 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17511 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17512 (TARGET_MODES_TIEABLE_P): Redefine.
17513 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17514 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17515 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17516 (TARGET_MODES_TIEABLE_P): Redefine.
17517 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17518 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17519 (pdp11_modes_tieable_p): New function.
17520 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17521 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17522 (rs6000_modes_tieable_p): New function.
17523 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17524 * config/powerpcspe/powerpcspe.md: Update comment.
17525 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17526 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17527 (TARGET_MODES_TIEABLE_P): Redefine.
17528 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17529 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17530 (rl78_modes_tieable_p): New function.
17531 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17532 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17533 (rs6000_modes_tieable_p): New function.
17534 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17535 * config/rs6000/rs6000.md: Update comment.
17536 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17537 * config/rx/rx.c (rx_modes_tieable_p): New function.
17538 (TARGET_MODES_TIEABLE_P): Redefine.
17539 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17540 * config/s390/s390.c (s390_modes_tieable_p): New function.
17541 (TARGET_MODES_TIEABLE_P): Redefine.
17542 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17543 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17544 (sh_modes_tieable_p): New function.
17545 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17546 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17547 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17548 (sparc_modes_tieable_p): Make static.
17549 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17550 * config/spu/spu.c (spu_modes_tieable_p): New function.
17551 (TARGET_MODES_TIEABLE_P): Redefine.
17552 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17553 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17554 (TARGET_MODES_TIEABLE_P): Redefine.
17555 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17556 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17557 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17558 * config/v850/v850.c (v850_modes_tieable_p): New function.
17559 (TARGET_MODES_TIEABLE_P): Redefine.
17560 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17561 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17562 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17563 (visium_modes_tieable_p): New function.
17564 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17565 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17566 (xtensa_modes_tieable_p): New function.
17567 * system.h (MODES_TIEABLE_P): Poison.
17569 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17570 Alan Hayward <alan.hayward@arm.com>
17571 David Sherwood <david.sherwood@arm.com>
17573 * target.def (hard_regno_mode_ok): New hook.
17574 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17575 (TARGET_HARD_REGNO_MODE_OK): ...this.
17576 * doc/tm.texi.in: Regenerate.
17577 * hooks.h (hook_bool_uint_mode_true): Declare.
17578 * hooks.c (hook_bool_uint_mode_true): New function.
17579 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17580 HARD_REGNO_MODE_OK.
17581 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17582 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17583 instead of HARD_REGNO_MODE_OK.
17584 * caller-save.c: Include target.h.
17585 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17586 HARD_REGNO_MODE_OK.
17587 * combine.c (can_combine_p): Likewise.
17588 (combinable_i3pat): Likewise.
17589 (can_change_dest_mode): Likewise.
17590 * expr.c (init_expr_target): Likewise.
17591 (convert_move): Likewise.
17592 (convert_modes): Likewise.
17593 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17594 (setup_prohibited_mode_move_regs): Likewise.
17595 * ira.h (target_ira): Likewise.
17596 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17597 * lra-constraints.c (process_alt_operands): Likewise.
17598 (split_reg): Likewise.
17599 * recog.c (peep2_find_free_register): Likewise.
17600 * ree.c (combine_reaching_defs): Likewise.
17601 * regcprop.c (maybe_mode_change): Likewise.
17602 * reginfo.c (init_reg_sets_1): Likewise.
17603 (choose_hard_reg_mode): Likewise.
17604 (simplifiable_subregs): Likewise.
17605 * regrename.c (check_new_reg_p): Likewise.
17606 * reload.c (find_valid_class): Likewise.
17607 (find_valid_class_1): Likewise.
17608 (reload_inner_reg_of_subreg): Likewise.
17609 (push_reload): Likewise.
17610 (combine_reloads): Likewise.
17611 (find_dummy_reload): Likewise.
17612 (find_reloads): Likewise.
17613 * reload1.c (find_reg): Likewise.
17614 (set_reload_reg): Likewise.
17615 (allocate_reload_reg): Likewise.
17616 (choose_reload_regs): Likewise.
17617 (reload_adjust_reg_for_temp): Likewise.
17618 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17619 (simplify_subreg_regno): Likewise.
17620 * sel-sched.c (init_regs_for_mode): Likewise.
17621 * varasm.c (make_decl_rtl): Likewise.
17622 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17623 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17624 HARD_REGNO_MODE_OK.
17625 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17626 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17628 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17629 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17630 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17631 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17632 (arc_mode_class): Delete.
17633 (HARD_REGNO_MODE_OK): Delete.
17634 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17635 (arc_hard_regno_mode_ok): Rename old array to...
17636 (arc_hard_regno_mode_ok_modes): ...this.
17637 (arc_conditional_register_usage): Update accordingly.
17638 (arc_mode_class): Make static.
17639 (arc_hard_regno_mode_ok): New function.
17640 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17641 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17642 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17643 (arm_hard_regno_mode_ok): Make static.
17644 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17645 HARD_REGNO_MODE_OK.
17646 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17647 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17648 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17650 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17651 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17652 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17653 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17654 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17655 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17656 * config/bfin/predicates.md (valid_reg_operand): Use
17657 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17658 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17659 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17661 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17662 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17663 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17664 (cr16_hard_regno_mode_ok): Make static and return a bool.
17665 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17666 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17667 (cris_hard_regno_mode_ok): New function.
17668 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17669 (epiphany_mode_class): Delete.
17670 (HARD_REGNO_MODE_OK): Delete.
17671 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17672 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17673 (hard_regno_mode_ok): Rename to...
17674 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17675 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17676 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17677 HARD_REGNO_MODE_OK.
17678 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17679 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17680 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17681 (frv_hard_regno_mode_ok): Make static and return a bool.
17682 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17683 HARD_REGNO_MODE_OK.
17684 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17685 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17686 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17687 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17690 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17691 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17692 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17694 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17695 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17696 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17697 (ia64_hard_regno_mode_ok): New function.
17698 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17699 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17700 (iq2000_hard_regno_mode_ok): New function.
17701 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17702 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17703 (lm32_hard_regno_mode_ok): New function.
17704 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17705 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17706 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17707 instead of HARD_REGNO_MODE_OK.
17708 (m32c_hard_regno_ok): Rename to...
17709 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17710 (m32c_cannot_change_mode_class): Update accordingly.
17711 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17712 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17713 (m32r_mode_class): Delete.
17714 (HARD_REGNO_MODE_OK): Delete.
17715 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17716 (m32r_hard_regno_mode_ok): Rename to...
17717 (m32r_hard_regno_modes): ...this.
17718 (m32r_mode_class): Make static.
17719 (m32r_hard_regno_mode_ok): New function.
17720 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17721 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17722 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17723 (m68k_hard_regno_mode_ok): Make static.
17724 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17725 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17726 (mcore_hard_regno_mode_ok): New function.
17727 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17728 (HARD_REGNO_MODE_OK): Delete.
17729 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17731 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17732 (microblaze_hard_regno_mode_ok): New function.
17733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17734 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17735 (mips_hard_regno_mode_ok): Delete.
17736 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17737 (mips_hard_regno_mode_ok_p): ...this and make static.
17738 (mips_hard_regno_mode_ok_p): Rename to...
17739 (mips_hard_regno_mode_ok_uncached): ...this.
17740 (mips_hard_regno_mode_ok): New function.
17741 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17742 of HARD_REGNO_MODE_OK.
17743 (mips_option_override): Update after above name changes.
17744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17745 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17746 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17747 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17748 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17749 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17750 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17751 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17752 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17753 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17754 (msp430_hard_regno_mode_ok): Make static and return a bool.
17755 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17756 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17757 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17759 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17760 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17761 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17762 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17763 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17764 (PA_HARD_REGNO_MODE_OK): ...this
17765 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17766 (PA_HARD_REGNO_MODE_OK): ...this.
17767 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17768 (pa_hard_regno_mode_ok): New function.
17769 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17770 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17771 (pdp11_hard_regno_mode_ok): New function.
17772 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17773 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17775 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17777 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17778 (rs6000_hard_regno_mode_ok): Rename to...
17779 (rs6000_hard_regno_mode_ok_uncached): ...this.
17780 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17781 (rs6000_hard_regno_mode_ok): New function.
17782 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17783 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17784 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17785 (riscv_hard_regno_mode_ok): ...this and make static.
17786 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17787 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17788 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17789 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17790 (rl78_hard_regno_mode_ok): Make static and return bool.
17791 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17792 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17794 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17795 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17796 (rs6000_hard_regno_mode_ok): Rename to...
17797 (rs6000_hard_regno_mode_ok_uncached): ...this.
17798 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17799 (rs6000_hard_regno_mode_ok): New function.
17800 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17801 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17802 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17803 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17804 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17805 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17807 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17808 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17809 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17810 (sh_hard_regno_mode_ok): Make static.
17811 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17812 instead of HARD_REGNO_MODE_OK.
17813 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17814 (sparc_mode_class): Delete.
17815 (HARD_REGNO_MODE_OK): Delete.
17816 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17817 (hard_regno_mode_classes): Make static.
17818 (sparc_mode_class): Likewise.
17819 (sparc_hard_regno_mode_ok): New function.
17820 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17821 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17822 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17825 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17827 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17828 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17829 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17830 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17831 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17832 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17833 (visium_hard_regno_mode_ok): New function.
17834 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17835 instead of HARD_REGNO_MODE_OK.
17836 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17837 (HARD_REGNO_MODE_OK): Delete.
17838 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17839 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17840 (xtensa_option_override): Update accordingly.
17841 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17842 (xtensa_hard_regno_mode_ok): New function.
17843 * system.h (HARD_REGNO_MODE_OK): Poison.
17845 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17846 Alan Hayward <alan.hayward@arm.com>
17847 David Sherwood <david.sherwood@arm.com>
17849 * target.def (hard_regno_call_part_clobbered): New hook.
17850 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17851 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17852 * doc/tm.texi: Regenerate.
17853 * hooks.h (hook_bool_uint_mode_false): Declare.
17854 * hooks.c (hook_bool_uint_mode_false): New function.
17855 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17856 * cselib.c (cselib_process_insn): Use
17857 targetm.hard_regno_call_part_clobbered instead of
17858 HARD_REGNO_CALL_PART_CLOBBERED.
17859 * ira-conflicts.c (ira_build_conflicts): Likewise.
17860 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17861 * lra-constraints.c (need_for_call_save_p): Likewise.
17862 * lra-lives.c: Include target.h.
17863 (check_pseudos_live_through_calls): Use
17864 targetm.hard_regno_call_part_clobbered instead of
17865 HARD_REGNO_CALL_PART_CLOBBERED.
17866 * regcprop.c: Include target.h.
17867 (copyprop_hardreg_forward_1): Use
17868 targetm.hard_regno_call_part_clobbered instead of
17869 HARD_REGNO_CALL_PART_CLOBBERED.
17870 * reginfo.c (choose_hard_reg_mode): Likewise.
17871 * regrename.c (check_new_reg_p): Likewise.
17872 * reload.c (find_equiv_reg): Likewise.
17873 * reload1.c (emit_reload_insns): Likewise.
17874 * sched-deps.c (deps_analyze_insn): Likewise.
17875 * sel-sched.c (init_regs_for_mode): Likewise.
17876 (mark_unavailable_hard_regs): Likewise.
17877 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17878 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17879 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17881 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17882 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17883 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17885 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17887 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17888 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17889 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17891 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17892 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17893 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17895 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17896 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17898 * config/powerpcspe/powerpcspe.c
17899 (rs6000_hard_regno_call_part_clobbered): New function.
17900 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17901 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17902 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17904 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17905 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17906 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17908 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17909 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17910 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17912 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17913 Alan Hayward <alan.hayward@arm.com>
17914 David Sherwood <david.sherwood@arm.com>
17916 * rtl.h (subreg_memory_offset): Declare.
17917 * emit-rtl.c (subreg_memory_offset): New function.
17918 * expmed.c (store_bit_field_1): Use it.
17919 * expr.c (undefined_operand_subword_p): Likewise.
17920 * simplify-rtx.c (simplify_subreg): Likewise.
17922 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17924 PR rtl-optimization/57448
17927 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17928 using atomic_load pattern.
17929 (expand_atomic_store): Likewise.
17931 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17934 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17935 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17936 handling. Use replace_call_with_value with NULL instead of
17937 gsi_replace, unlink_stmt_vdef and release_defs.
17939 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17942 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17944 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17947 * rtl.h (emit_library_call_value_1): Declare.
17948 (emit_library_call): Replace declaration with a series of overloads.
17949 Remove the parameter count argument.
17950 (emit_library_call_value): Likewise.
17951 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17952 with an "rtx_mode_t *".
17953 (emit_library_call_value): Delete.
17954 (emit_library_call): Likewise.
17955 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17956 (asan_emit_allocas_unpoison): Likewise.
17957 * builtins.c (expand_builtin_powi): Likewise.
17958 (expand_asan_emit_allocas_unpoison): Likewise.
17959 * cfgexpand.c (expand_main_function): Likewise.
17960 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17961 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17962 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17963 * config/arm/arm.c (arm_trampoline_init): Likewise.
17964 (arm_call_tls_get_addr): Likewise.
17965 (arm_expand_divmod_libfunc): Likewise.
17966 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17967 (smulsi3_highpart): Likewise.
17968 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17969 (c6x_expand_compare): Likewise.
17970 (c6x_expand_movmem): Likewise.
17971 * config/frv/frv.c (frv_trampoline_init): Likewise.
17972 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17973 (ix86_expand_divmod_libfunc): Likewise.
17974 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17975 (ia64_expand_compare): Likewise.
17976 (ia64_profile_hook): Likewise.
17977 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17978 (nonlocal_goto): Likewise.
17979 (restore_stack_nonlocal): Likewise.
17980 * config/m32r/m32r.c (block_move_call): Likewise.
17981 (m32r_trampoline_init): Likewise.
17982 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17983 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17984 (m68k_call_m68k_read_tp): Likewise.
17985 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17986 (microblaze_expand_divide): Likewise.
17987 * config/mips/mips.h (mips_args): Likewise.
17988 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17989 (MIPS_ICACHE_SYNC): Likewise.
17990 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17991 (nios2_trampoline_init): Likewise.
17992 * config/pa/pa.c (hppa_tls_call): Likewise.
17993 (pa_trampoline_init): Likewise.
17994 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17995 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17996 (expand_strn_compare): Likewise.
17997 (rs6000_generate_compare): Likewise.
17998 (rs6000_expand_float128_convert): Likewise.
17999 (output_profile_hook): Likewise.
18000 (rs6000_trampoline_init): Likewise.
18001 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18002 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18003 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18004 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18005 (rs6000_generate_compare): Likewise.
18006 (rs6000_expand_float128_convert): Likewise.
18007 (output_profile_hook): Likewise.
18008 (rs6000_trampoline_init): Likewise.
18009 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18010 * config/sh/sh.c (sh_trampoline_init): Likewise.
18011 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18012 (sparc_emit_float_lib_cmp): Likewise.
18013 (sparc32_initialize_trampoline): Likewise.
18014 (sparc64_initialize_trampoline): Likewise.
18015 (sparc_profile_hook): Likewise.
18016 * config/spu/spu.c (ea_load_store): Likewise.
18017 * config/spu/spu.md (floatunssidf2): Likewise.
18018 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18019 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18020 * config/visium/visium.c (expand_block_move_4): Likewise.
18021 (expand_block_move_2): Likewise.
18022 (expand_block_move_1): Likewise.
18023 (expand_block_set_4): Likewise.
18024 (expand_block_set_2): Likewise.
18025 (expand_block_set_1): Likewise.
18026 (visium_trampoline_init): Likewise.
18027 (visium_profile_hook): Likewise.
18028 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18029 (xtensa_setup_frame_addresses): Likewise.
18030 (xtensa_trampoline_init): Likewise.
18031 * except.c (sjlj_emit_function_enter): Likewise.
18032 (sjlj_emit_function_exit): Likewise.
18033 * explow.c (allocate_dynamic_stack_space): Likewise.
18034 (probe_stack_range): Likewise.
18035 * expr.c (convert_mode_scalar): Likewise.
18036 * optabs.c (expand_binop): Likewise.
18037 (expand_twoval_binop_libfunc): Likewise.
18038 (expand_unop): Likewise.
18039 (prepare_cmp_insn): Likewise.
18040 (prepare_float_lib_cmp): Likewise.
18041 (expand_float): Likewise.
18042 (expand_fix): Likewise.
18043 (expand_fixed_convert): Likewise.
18044 (maybe_emit_sync_lock_test_and_set): Likewise.
18045 (expand_atomic_compare_and_swap): Likewise.
18046 (expand_mem_thread_fence): Likewise.
18047 (expand_atomic_fetch_op): Likewise.
18049 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18051 * doc/generic.texi (OpenACC): Adjust URL.
18052 * doc/invoke.texi (C Dialect Options): Ditto.
18054 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18056 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18057 predicate for operand 1. Add (m,<S>) constraint.
18058 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18059 Prevent memory operand 1 with register operand 2.
18061 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18063 PR rtl-optimization/82024
18064 * combine.c (try_combine): If the combination result is a PARALLEL,
18065 and we only need to retain the SET in there that would be placed
18066 at I2, check that we can place that at I3 instead, before doing so.
18068 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18071 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18073 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18074 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18077 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18078 Jeff Law <law@redhat.com>
18080 * varasm.c (bss_initializer_p): Do not put constants into .bss
18081 (categorize_decl_for_section): Handle bss_initializer_p returning
18082 false when DECL_INITIAL is NULL.
18084 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18087 * config/s390/s390.c (s390_can_inline_p): New function.
18089 2017-09-01 Jeff Law <law@redhat.com>
18091 PR tree-optimization/82052
18092 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18093 Always initialize the returned slot after a hash table miss
18094 when INSERT is true.
18096 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18098 * config/s390/s390.md (mem_signal_fence): Remove.
18099 * doc/md.texi (mem_signal_fence): Remove.
18100 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18102 * target-insns.def (mem_signal_fence): Remove.
18104 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18107 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18110 * asan.c (create_odr_indicator): Strip name encoding from assembler
18111 name before appending it after __odr_asan_.
18113 2017-09-01 Martin Liska <mliska@suse.cz>
18115 PR tree-optimization/82059
18116 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18117 frequency only when an edge is redirected.
18119 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18121 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18122 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18123 (arc_conditional_register_usage): Remove ARC600 lp_count
18125 (arc_file_start): Emit Tag_ARC_CPU_variation.
18126 (arc_can_use_doloop_p): New conditions to use ZOLs.
18127 (hwloop_fail): New function.
18128 (hwloop_optimize): Likewise.
18129 (hwloop_pattern_reg): Likewise.
18130 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18131 (arc_reorg_loops): New function, calls reorg_loops.
18132 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18133 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18135 (arc_loop_hazard): Remove function, functionality moved into
18137 (arc_hazard): Remove arc_loop_hazard call.
18138 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18139 into hwloop_optimize.
18140 (arc_label_align): Remove ZOL handling.
18141 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18142 * config/arc/arc.md (doloop_begin): Remove pattern.
18143 (doloop_begin_i): Likewise.
18144 (doloop_end_i): Likewise.
18145 (doloop_fallback): Likewise.
18146 (doloop_fallback_m): Likewise.
18147 (doloop_end): Reimplement expand.
18148 (arc_lp): New pattern for LP instruction.
18149 (loop_end): New pattern.
18150 (loop_fail): Likewise.
18151 (decrement_and_branch_until_zero): Likewise.
18152 * config/arc/arc.opt (mlpc-width): New option.
18153 * doc/invoke.texi (mlpc-width): Document option.
18155 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18157 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18158 (arc_ccfsm_advance): Fix checking for delay slots.
18159 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18161 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18163 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18165 (movhi_insn): Update store instruction constraint which are saving
18166 6-bit short immediates.
18167 (movsi_insn): Consider also short scaled load operations.
18168 (zero_extendhisi2_i): Use Usd constraint instead of T.
18169 (extendhisi2_i): Add q constraint.
18170 (arc_clzsi2): Add type and length attributes.
18171 (arc_ctzsi2): Likewise.
18172 * config/arc/constraints.md (Usc): Update constraint, the
18173 assembler can parse two relocations for a single instruction.
18175 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18177 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18178 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18180 2017-08-31 Olivier Hainque <hainque@adacore.com>
18182 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18183 match as powerpc-wrs-vxworks*.
18185 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18187 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18188 register constraint for by-element operand.
18189 (aarch64_mls_elt_merge<mode>): Likewise.
18191 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18193 * config/arc/arc.c (arc_can_follow_jump): Check for short
18196 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18198 * config.gcc: Use g.opt for arc.
18199 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18200 functionality moved to ...
18201 (legitimate_scaled_address_p): New function, ...here.
18202 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18203 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18204 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18206 (arc_override_options): Handle G option.
18207 (arc_output_pic_addr_const): Correct function definition.
18208 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18209 (arc_decl_anon_ns_mem_p): Delete.
18210 (arc_in_small_data_p): Overhaul this function to take into
18211 consideration the value given via G option.
18212 (arc_rewrite_small_data_1): Renamed and corrected old
18213 arc_rewrite_small_data function.
18214 (arc_rewrite_small_data): New function.
18215 (small_data_pattern): Don't use pic_offset_table_rtx.
18216 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18217 * config/arc/simdext.md (movmisalignv2hi): Use
18218 prepare_move_operands function.
18220 (movmisalign*): Likewise.
18221 * doc/invoke.texi (ARC options): Document -G option.
18223 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18225 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18227 * config/arc/arc.c (arc_print_operand): Output scalled address for
18228 sdata whenever is possible.
18229 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18230 load/stores are available.
18231 (compact_sda_memory_operand): Check for the alignment required by
18232 code density instructions.
18233 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18235 * config/arc/constraints.md (Usd): Update constraint.
18236 (Us0): New constraint.
18237 (Usc): Update constraint.
18239 2017-08-31 Richard Biener <rguenther@suse.de>
18241 PR middle-end/82054
18242 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18243 function only once.
18245 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18247 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18248 Resize type_signature.
18250 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18251 Alan Hayward <alan.hayward@arm.com>
18252 David Sherwood <david.sherwood@arm.com>
18254 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18255 subregs whose inner modes can be stored in GPRs.
18256 (aarch64_classify_index): Likewise.
18258 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18259 Alan Hayward <alan.hayward@arm.com>
18260 David Sherwood <david.sherwood@arm.com>
18262 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18263 (V_INT_EQUIV): ...this.
18264 (v_cmp_result): Rename to...
18265 (v_int_equiv): ...this.
18266 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18267 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18268 (copysign<mode>3): Likewise.
18269 (aarch64_simd_bsl<mode>_internal): Likewise.
18270 (aarch64_simd_bsl<mode>): Likewise.
18271 (vec_cmp<mode><mode>): Likewise.
18272 (vcond<mode><mode>): Likewise.
18273 (vcond<v_cmp_mixed><mode>): Likewise.
18274 (vcondu<mode><v_cmp_mixed>): Likewise.
18275 (aarch64_cm<optab><mode>): Likewise.
18276 (aarch64_cmtst<mode>): Likewise.
18277 (aarch64_fac<optab><mode>): Likewise.
18278 (vec_perm_const<mode>): Likewise.
18279 (vcond_mask_<mode><v_cmp_result>): Rename to...
18280 (vcond_mask_<mode><v_int_equiv>): ...this.
18281 (vec_cmp<mode><v_cmp_result>): Rename to...
18282 (vec_cmp<mode><v_int_equiv>): ...this.
18284 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18285 Alan Hayward <alan.hayward@arm.com>
18286 David Sherwood <david.sherwood@arm.com>
18288 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18290 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18291 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18292 (UNSPEC_LD4_DREG): New unspecs.
18293 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18294 (aarch64_ld2<mode>_dreg_be): Replace with...
18295 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18297 (aarch64_ld3<mode>_dreg_le)
18298 (aarch64_ld3<mode>_dreg_be): Replace with...
18299 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18301 (aarch64_ld4<mode>_dreg_le)
18302 (aarch64_ld4<mode>_dreg_be): Replace with...
18303 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18306 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18308 PR tree-optimization/81987
18309 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18310 insert an initializer in a location not dominated by the stride
18313 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18315 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18316 on the entire header of the finally block in the fallthru case.
18318 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18320 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18322 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18324 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18325 rs6000_emit_move_from_cr and call renamed function.
18326 (rs6000_emit_prologue): Call renamed functions.
18327 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18328 movesi_from_cr, remove volatile CRs.
18330 2017-08-30 Jon Beniston <jon@beniston.com>
18331 Richard Biener <rguenther@suse.de>
18333 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18334 of VECTOR_MODE_P check.
18335 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18336 element vector types.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18340 * df.h (df_read_modify_subreg_p): Remove in favor of...
18341 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18342 const_rtx instead of an rtx.
18343 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18344 * df-problems.c (df_word_lr_mark_ref): Likewise.
18345 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18346 (mark_pseudo_reg_dead): Likewise.
18347 (mark_ref_dead): Likewise.
18348 * reginfo.c (init_subregs_of_mode): Likewise.
18349 * sched-deps.c (sched_analyze_1): Likewise.
18350 * df-scan.c (df_def_record_1): Likewise.
18351 (df_uses_record): Likewise.
18352 (df_read_modify_subreg_p): Remove in favor of...
18353 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18354 const_rtx instead of an rtx.
18356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18357 Alan Hayward <alan.hayward@arm.com>
18358 David Sherwood <david.sherwood@arm.com>
18360 * rtl.h (partial_subreg_p): New function.
18361 * caller-save.c (save_call_clobbered_regs): Use it.
18362 * calls.c (expand_call): Likewise.
18363 * combine.c (combinable_i3pat): Likewise.
18364 (simplify_set): Likewise.
18365 (make_extraction): Likewise.
18366 (make_compound_operation_int): Likewise.
18367 (gen_lowpart_or_truncate): Likewise.
18368 (force_to_mode): Likewise.
18369 (make_field_assignment): Likewise.
18370 (reg_truncated_to_mode): Likewise.
18371 (record_truncated_value): Likewise.
18372 (move_deaths): Likewise.
18373 * cse.c (record_jump_cond): Likewise.
18374 (cse_insn): Likewise.
18375 * cselib.c (cselib_lookup_1): Likewise.
18376 * expmed.c (extract_bit_field_using_extv): Likewise.
18377 * function.c (assign_parm_setup_reg): Likewise.
18378 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18379 * ira-build.c (create_insn_allocnos): Likewise.
18380 * lra-coalesce.c (merge_pseudos): Likewise.
18381 * lra-constraints.c (match_reload): Likewise.
18382 (simplify_operand_subreg): Likewise.
18383 (curr_insn_transform): Likewise.
18384 * lra-lives.c (process_bb_lives): Likewise.
18385 * lra.c (new_insn_reg): Likewise.
18386 (lra_substitute_pseudo): Likewise.
18387 * regcprop.c (mode_change_ok): Likewise.
18388 (maybe_mode_change): Likewise.
18389 (copyprop_hardreg_forward_1): Likewise.
18390 * reload.c (push_reload): Likewise.
18391 (find_reloads): Likewise.
18392 (find_reloads_subreg_address): Likewise.
18393 * reload1.c (alter_reg): Likewise.
18394 (eliminate_regs_1): Likewise.
18395 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18397 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18399 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18400 back to if statements, including unpack.
18402 2017-08-30 Martin Liska <mliska@suse.cz>
18404 PR inline-asm/82001
18405 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18407 (func_checker::compare_asm_inputs_outputs): ... this function.
18408 (func_checker::compare_gimple_asm): Use the function to compare
18409 also ASM constrains.
18410 * ipa-icf-gimple.h: Rename the function.
18412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18413 Alan Hayward <alan.hayward@arm.com>
18414 David Sherwood <david.sherwood@arm.com>
18416 * coretypes.h (complex_mode): New type.
18417 * gdbhooks.py (build_pretty_printer): Handle it.
18418 * machmode.h (complex_mode): New class.
18419 (complex_mode::includes_p): New function.
18420 (is_complex_int_mode): Likewise.
18421 (is_complex_float_mode): Likewise.
18422 * genmodes.c (get_mode_class): Handle complex mode classes.
18423 * function.c (expand_function_end): Use is_complex_int_mode.
18425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18426 Alan Hayward <alan.hayward@arm.com>
18427 David Sherwood <david.sherwood@arm.com>
18429 * coretypes.h (scalar_mode_pod): New typedef.
18430 * gdbhooks.py (build_pretty_printer): Handle it.
18431 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18432 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18433 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18434 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18435 as_a <scalar_mode>.
18437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18438 Alan Hayward <alan.hayward@arm.com>
18439 David Sherwood <david.sherwood@arm.com>
18441 * machmode.h (mode_for_vector): Take a scalar_mode instead
18443 * stor-layout.c (mode_for_vector): Likewise.
18444 * explow.c (promote_mode): Use as_a <scalar_mode>.
18445 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18448 Alan Hayward <alan.hayward@arm.com>
18449 David Sherwood <david.sherwood@arm.com>
18451 * target.def (preferred_simd_mode): Take a scalar_mode
18452 instead of a machine_mode.
18453 * targhooks.h (default_preferred_simd_mode): Likewise.
18454 * targhooks.c (default_preferred_simd_mode): Likewise.
18455 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18456 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18457 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18458 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18459 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18460 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18461 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18462 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18464 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18465 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18466 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18467 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18468 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18469 * doc/tm.texi: Regenerate.
18470 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18474 Alan Hayward <alan.hayward@arm.com>
18475 David Sherwood <david.sherwood@arm.com>
18477 * target.def (scalar_mode_supported_p): Take a scalar_mode
18478 instead of a machine_mode.
18479 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18480 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18481 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18482 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18483 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18484 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18485 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18486 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18487 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18488 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18489 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18490 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18491 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18492 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18494 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18495 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18496 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18497 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18498 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18500 * doc/tm.texi: Regenerate.
18502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18503 Alan Hayward <alan.hayward@arm.com>
18504 David Sherwood <david.sherwood@arm.com>
18506 * coretypes.h (opt_scalar_mode): New typedef.
18507 * gdbhooks.py (build_pretty_printers): Handle it.
18508 * machmode.h (mode_iterator::get_2xwider): Add overload for
18510 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18512 * expr.c (convert_mode_scalar): Likewise.
18513 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18514 * optabs.c (expand_float): Likewise.
18515 (expand_fix): Likewise.
18516 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18519 Alan Hayward <alan.hayward@arm.com>
18520 David Sherwood <david.sherwood@arm.com>
18522 * optabs.c (expand_float): Explicitly check for scalars before
18523 using a branching expansion.
18524 (expand_fix): Likewise.
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * expr.c (convert_mode): Split scalar handling out into...
18531 (convert_mode_scalar): ...this new function. Treat the modes
18534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18535 Alan Hayward <alan.hayward@arm.com>
18536 David Sherwood <david.sherwood@arm.com>
18538 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18540 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18543 Alan Hayward <alan.hayward@arm.com>
18544 David Sherwood <david.sherwood@arm.com>
18546 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18547 rather than a machine_mode.
18548 (fixed_from_string): Likewise.
18549 (fixed_convert): Likewise.
18550 (fixed_convert_from_int): Likewise.
18551 (fixed_convert_from_real): Likewise.
18552 (real_convert_from_fixed): Likewise.
18553 * fixed-value.c (fixed_from_double_int): Likewise.
18554 (fixed_from_string): Likewise.
18555 (fixed_convert): Likewise.
18556 (fixed_convert_from_int): Likewise.
18557 (fixed_convert_from_real): Likewise.
18558 (real_convert_from_fixed): Likewise.
18559 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18566 of separate mode class checks. Do not allow vector modes here.
18567 (immed_wide_int_const): Use as_a <scalar_mode>.
18568 * explow.c (trunc_int_for_mode): Likewise.
18569 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18570 (wi::shwi): Likewise.
18571 (wi::min_value): Likewise.
18572 (wi::max_value): Likewise.
18573 * dwarf2out.c (loc_descriptor): Likewise.
18574 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18575 for CONST_WIDE_INT.
18577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18578 Alan Hayward <alan.hayward@arm.com>
18579 David Sherwood <david.sherwood@arm.com>
18581 * tree.h (SCALAR_TYPE_MODE): New macro.
18582 * expr.c (expand_expr_addr_expr_1): Use it.
18583 (expand_expr_real_2): Likewise.
18584 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18585 (fold_convert_const_fixed_from_int): Likewise.
18586 (fold_convert_const_fixed_from_real): Likewise.
18587 (native_encode_fixed): Likewise
18588 (native_encode_complex): Likewise
18589 (native_encode_vector): Likewise.
18590 (native_interpret_fixed): Likewise.
18591 (native_interpret_real): Likewise.
18592 (native_interpret_complex): Likewise.
18593 (native_interpret_vector): Likewise.
18594 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18595 (simd_clone_adjust_argument_types): Likewise.
18596 (simd_clone_init_simd_arrays): Likewise.
18597 (simd_clone_adjust): Likewise.
18598 * stor-layout.c (layout_type): Likewise.
18599 * tree.c (build_minus_one_cst): Likewise.
18600 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18601 * tree-inline.c (estimate_move_cost): Likewise.
18602 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18603 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18604 (vectorizable_reduction): Likewise.
18605 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18606 (vect_recog_mixed_size_cond_pattern): Likewise.
18607 (check_bool_pattern): Likewise.
18608 (adjust_bool_pattern): Likewise.
18609 (search_type_for_mask_1): Likewise.
18610 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18611 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18612 (vectorizable_load): Likewise.
18613 (vectorizable_store): Likewise.
18614 * ubsan.c (ubsan_encode_value): Likewise.
18615 * varasm.c (output_constant): Likewise.
18617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18618 Alan Hayward <alan.hayward@arm.com>
18619 David Sherwood <david.sherwood@arm.com>
18621 * coretypes.h (scalar_mode): New class.
18622 * machmode.h (scalar_mode): Likewise.
18623 (scalar_mode::includes_p): New function.
18624 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18625 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18626 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18627 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18628 * expmed.c (store_bit_field_1): Likewise.
18629 (extract_bit_field_1): Likewise.
18630 * expr.c (write_complex_part): Likewise.
18631 (read_complex_part): Likewise.
18632 (emit_move_complex_push): Likewise.
18633 (expand_expr_real_2): Likewise.
18634 * function.c (assign_parm_setup_reg): Likewise.
18635 (assign_parms_unsplit_complex): Likewise.
18636 * optabs.c (expand_binop): Likewise.
18637 * rtlanal.c (subreg_get_info): Likewise.
18638 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18639 * varasm.c (output_constant_pool_2): Likewise.
18641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18642 Alan Hayward <alan.hayward@arm.com>
18643 David Sherwood <david.sherwood@arm.com>
18645 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18647 (expmed_mult_highpart_optab): Likewise.
18648 (expmed_mult_highpart): Likewise.
18650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18651 Alan Hayward <alan.hayward@arm.com>
18652 David Sherwood <david.sherwood@arm.com>
18654 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18655 instead of a machine_mode.
18656 (builtin_memset_read_str): Likewise.
18657 * builtins.c (c_readstr): Likewise.
18658 (builtin_memcpy_read_str): Likewise.
18659 (builtin_strncpy_read_str): Likewise.
18660 (builtin_memset_read_str): Likewise.
18661 (builtin_memset_gen_str): Likewise.
18662 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18663 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18664 instead of a machine_mode.
18665 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18667 (make_extraction): Likewise.
18668 (try_widen_shift_mode): Take and return scalar_int_modes instead
18670 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18671 a scalar_int_mode instead of a machine_mode.
18672 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18673 (avr_addr_space_pointer_mode): Likewise.
18674 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18675 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18676 (msp430_unwind_word_mode): Likewise.
18677 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18678 (spu_addr_space_pointer_mode): Likewise.
18679 (spu_addr_space_address_mode): Likewise.
18680 (spu_libgcc_cmp_return_mode): Likewise.
18681 (spu_libgcc_shift_count_mode): Likewise.
18682 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18683 (rl78_addr_space_pointer_mode): Likewise.
18684 (fl78_unwind_word_mode): Likewise.
18685 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18687 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18688 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18689 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18690 (mips_valid_pointer_mode): Likewise.
18691 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18692 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18693 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18695 (ft32_addr_space_address_mode): Likewise.
18696 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18697 scalar_int_mode instead of a machine_mode.
18698 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18700 (m32c_addr_space_address_mode): Likewise.
18701 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18702 (rs6000_eh_return_filter_mode): Likewise.
18703 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18704 (rs6000_eh_return_filter_mode): Likewise.
18705 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18706 (s390_libgcc_shift_count_mode): Likewise.
18707 (s390_unwind_word_mode): Likewise.
18708 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18710 * target.def (mode_rep_extended): Likewise.
18711 (valid_pointer_mode): Likewise.
18712 (addr_space.valid_pointer_mode): Likewise.
18713 (eh_return_filter_mode): Return a scalar_int_mode rather than
18715 (libgcc_cmp_return_mode): Likewise.
18716 (libgcc_shift_count_mode): Likewise.
18717 (unwind_word_mode): Likewise.
18718 (addr_space.pointer_mode): Likewise.
18719 (addr_space.address_mode): Likewise.
18720 * doc/tm.texi: Regenerate.
18721 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18723 (do_jump): Use scalar_int_mode for local variables.
18724 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18725 rather than a machine_mode.
18726 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18727 (scompare_loc_descriptor_wide): Likewise.
18728 (scompare_loc_descriptor_narrow): Likewise.
18729 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18731 * except.c (sjlj_emit_dispatch_table): Likewise.
18732 (expand_builtin_eh_copy_values): Likewise.
18733 * explow.c (convert_memory_address_addr_space_1): Likewise.
18734 Take a scalar_int_mode rather than a machine_mode.
18735 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18736 than a machine_mode.
18737 (memory_address_addr_space): Use scalar_int_mode for local variables.
18738 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18739 rather than a machine_mode.
18740 * expmed.c (mask_rtx): Likewise.
18741 (init_expmed_one_conv): Likewise.
18742 (expand_mult_highpart_adjust): Likewise.
18743 (extract_high_half): Likewise.
18744 (expmed_mult_highpart_optab): Likewise.
18745 (expmed_mult_highpart): Likewise.
18746 (expand_smod_pow2): Likewise.
18747 (expand_sdiv_pow2): Likewise.
18748 (emit_store_flag_int): Likewise.
18749 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18751 (extract_low_bits): Likewise.
18752 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18754 * expr.c (pieces_addr::adjust): Likewise.
18755 (can_store_by_pieces): Likewise.
18756 (store_by_pieces): Likewise.
18757 (clear_by_pieces_1): Likewise.
18758 (expand_expr_addr_expr_1): Likewise.
18759 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18760 (expand_expr_real_1): Likewise.
18761 (try_casesi): Likewise.
18762 * final.c (shorten_branches): Likewise.
18763 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18764 type of "mode" to machine_mode.
18765 * internal-fn.c (expand_arith_overflow_result_store): Take a
18766 scalar_int_mode rather than a machine_mode.
18767 (expand_mul_overflow): Use scalar_int_mode for local variables.
18768 * loop-doloop.c (doloop_modify): Likewise.
18769 (doloop_optimize): Likewise.
18770 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18771 than a machine_mode.
18772 (expand_doubleword_shift_condmove): Likewise.
18773 (expand_doubleword_shift): Likewise.
18774 (expand_doubleword_clz): Likewise.
18775 (expand_doubleword_popcount): Likewise.
18776 (expand_doubleword_parity): Likewise.
18777 (expand_absneg_bit): Use scalar_int_mode for local variables.
18778 (prepare_float_lib_cmp): Likewise.
18779 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18780 rather than a machine_mode.
18781 (convert_memory_address_addr_space): Likewise.
18782 (get_mode_bounds): Likewise.
18783 (get_address_mode): Return a scalar_int_mode rather than a
18785 * rtlanal.c (get_address_mode): Likewise.
18786 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18787 than a machine_mode.
18788 * targhooks.c (default_mode_rep_extended): Likewise.
18789 (default_valid_pointer_mode): Likewise.
18790 (default_addr_space_valid_pointer_mode): Likewise.
18791 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18792 than a machine_mode.
18793 (default_libgcc_cmp_return_mode): Likewise.
18794 (default_libgcc_shift_count_mode): Likewise.
18795 (default_unwind_word_mode): Likewise.
18796 (default_addr_space_pointer_mode): Likewise.
18797 (default_addr_space_address_mode): Likewise.
18798 * targhooks.h (default_eh_return_filter_mode): Likewise.
18799 (default_libgcc_cmp_return_mode): Likewise.
18800 (default_libgcc_shift_count_mode): Likewise.
18801 (default_unwind_word_mode): Likewise.
18802 (default_addr_space_pointer_mode): Likewise.
18803 (default_addr_space_address_mode): Likewise.
18804 (default_mode_rep_extended): Take a scalar_int_mode rather than
18806 (default_valid_pointer_mode): Likewise.
18807 (default_addr_space_valid_pointer_mode): Likewise.
18808 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18810 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18811 rather than a machine_mode.
18812 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18813 for local variables.
18814 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18815 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18816 than a machine_mode.
18818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18819 Alan Hayward <alan.hayward@arm.com>
18820 David Sherwood <david.sherwood@arm.com>
18822 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18823 the mode argument to scalar_int_mode.
18824 (do_jump_by_parts_zero_rtx): Likewise.
18825 (do_jump_by_parts_equality_rtx): Likewise.
18826 (do_jump_by_parts_greater): Take a mode argument.
18827 (do_jump_by_parts_equality): Likewise.
18828 (do_jump_1): Update calls accordingly.
18830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18831 Alan Hayward <alan.hayward@arm.com>
18832 David Sherwood <david.sherwood@arm.com>
18834 * is-a.h (safe_dyn_cast): New function.
18835 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18836 (jump_table_for_label): Likewise.
18837 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18838 instead of an rtx_insn *.
18839 (shorten_branches): Use dyn_cast instead of LABEL_P and
18840 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18841 rtx_jump_table_data::get_data_mode.
18842 (final_scan_insn): Likewise.
18844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18845 Alan Hayward <alan.hayward@arm.com>
18846 David Sherwood <david.sherwood@arm.com>
18848 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18849 trying to combine a full-register integer set with a subreg
18852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18853 Alan Hayward <alan.hayward@arm.com>
18854 David Sherwood <david.sherwood@arm.com>
18856 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18857 that is always either address_mode or pointer_mode.
18859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18860 Alan Hayward <alan.hayward@arm.com>
18861 David Sherwood <david.sherwood@arm.com>
18863 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18864 when the two are known to be equal.
18866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18867 Alan Hayward <alan.hayward@arm.com>
18868 David Sherwood <david.sherwood@arm.com>
18870 * simplify-rtx.c (simplify_const_unary_operation): Use
18871 is_a <scalar_int_mode> instead of checking for a nonzero
18872 precision. Forcibly convert op_mode to a scalar_int_mode
18873 in that case. More clearly differentiate the operand and
18874 result modes and use the former when deciding what the value
18875 of a count-bits operation should be. Use is_int_mode instead
18876 of checking for a MODE_INT. Remove redundant check for whether
18877 this mode has a zero precision.
18879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18880 Alan Hayward <alan.hayward@arm.com>
18881 David Sherwood <david.sherwood@arm.com>
18883 * optabs.c (widen_leading): Change the type of the mode argument
18884 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18885 (widen_bswap): Likewise.
18886 (expand_parity): Likewise.
18887 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18888 (expand_ffs): Likewise.
18889 (epand_unop): Check for scalar integer modes before calling the
18892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18893 Alan Hayward <alan.hayward@arm.com>
18894 David Sherwood <david.sherwood@arm.com>
18896 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18898 (expand_expr_real_1): Update call accordingly.
18900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18901 Alan Hayward <alan.hayward@arm.com>
18902 David Sherwood <david.sherwood@arm.com>
18904 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18905 value_mode arguments. Use scalar_int_mode internally.
18906 (store_bit_field_1): Rename the new integer mode from imode
18907 to op0_mode and use it instead of GET_MODE (op0). Update calls
18908 to store_split_bit_field, store_bit_field_using_insv and
18909 store_fixed_bit_field.
18910 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18911 Use scalar_int_mode internally. Use a bit count rather than a mode
18912 when calculating the largest bit size for get_best_mode.
18913 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18914 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18915 Remove assertion that OP0 has a scalar integer mode.
18916 (store_split_bit_field): Add op0_mode and value_mode arguments.
18917 Update calls to extract_fixed_bit_field.
18918 (extract_bit_field_using_extv): Add an op0_mode argument.
18919 Use scalar_int_mode internally.
18920 (extract_bit_field_1): Rename the new integer mode from imode to
18921 op0_mode and use it instead of GET_MODE (op0). Update calls to
18922 extract_split_bit_field, extract_bit_field_using_extv and
18923 extract_fixed_bit_field.
18924 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18925 to extract_split_bit_field and extract_fixed_bit_field_1.
18926 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18927 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18928 on the target mode.
18929 (extract_split_bit_field): Add an op0_mode argument. Update call
18930 to extract_fixed_bit_field.
18932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18933 Alan Hayward <alan.hayward@arm.com>
18934 David Sherwood <david.sherwood@arm.com>
18936 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18937 * explow.c (hard_function_value): Likewise.
18938 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18939 convert_to_mode call outside the loop.
18940 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18941 for the mode iterator. Require the mode specified by max_pieces
18943 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18945 (copy_blkmode_to_reg): Likewise.
18946 (set_storage_via_setmem): Likewise.
18947 * optabs.c (prepare_cmp_insn): Likewise.
18948 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18949 * stor-layout.c (finish_bitfield_representative): Likewise.
18951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18952 Alan Hayward <alan.hayward@arm.com>
18953 David Sherwood <david.sherwood@arm.com>
18955 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18956 * expr.c (convert_move): Use them.
18957 (convert_modes): Likewise.
18958 (store_expr_with_bounds): Likewise.
18960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18961 Alan Hayward <alan.hayward@arm.com>
18962 David Sherwood <david.sherwood@arm.com>
18964 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18965 parameter for the mode of "x". Remove the "known_x", "known_mode"
18966 and "known_ret" arguments. Change the type of the mode argument
18967 to scalar_int_mode.
18968 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18969 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18970 (reg_num_sign_bit_copies_for_combine): Likewise.
18971 * rtlanal.c (nonzero_bits1): Likewise.
18972 (num_sign_bit_copies1): Likewise.
18973 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18974 (reg_num_sign_bit_copies_general): Likewise.
18975 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18976 (reg_nonzero_bits_general): Likewise.
18978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18979 Alan Hayward <alan.hayward@arm.com>
18980 David Sherwood <david.sherwood@arm.com>
18982 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18983 than in subroutines. Return 1 for non-integer modes.
18984 (cached_num_sign_bit_copies): Change the type of the mode parameter
18985 to scalar_int_mode.
18986 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18987 classes. Handle CONST_INT_P first and then check whether X also
18988 has a scalar integer mode. Check the same thing for inner registers
18989 of a SUBREG and for values that are being extended or truncated.
18991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18992 Alan Hayward <alan.hayward@arm.com>
18993 David Sherwood <david.sherwood@arm.com>
18995 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18996 in subroutines. Return the mode mask for non-integer modes.
18997 (cached_nonzero_bits): Change the type of the mode parameter
18998 to scalar_int_mode.
18999 (nonzero_bits1): Likewise. Remove early exit for other mode
19000 classes. Handle CONST_INT_P first and then check whether X
19001 also has a scalar integer mode.
19003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19004 Alan Hayward <alan.hayward@arm.com>
19005 David Sherwood <david.sherwood@arm.com>
19007 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19008 Return a scalar_int_mode and assert that the size is greater than
19010 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19011 (op_by_pieces_d::op_by_pieces_d): Likewise.
19012 (op_by_pieces_d::run): Likewise.
19013 (can_store_by_pieces): Likewise.
19015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19016 Alan Hayward <alan.hayward@arm.com>
19017 David Sherwood <david.sherwood@arm.com>
19019 * combine.c (extract_left_shift): Add a mode argument and update
19021 (make_compound_operation_int): Change the type of the mode parameter
19022 to scalar_int_mode and update the call to extract_left_shift.
19024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19025 Alan Hayward <alan.hayward@arm.com>
19026 David Sherwood <david.sherwood@arm.com>
19028 * combine.c (simplify_and_const_int): Change the type of the mode
19029 parameter to scalar_int_mode.
19030 (simplify_and_const_int_1): Likewise. Update recursive call.
19032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19033 Alan Hayward <alan.hayward@arm.com>
19034 David Sherwood <david.sherwood@arm.com>
19036 * combine.c (simplify_compare_const): Check that the mode is a
19037 scalar_int_mode (rather than VOIDmode) before testing its
19039 (simplify_comparison): Move COMPARISON_P handling out of the
19040 loop and restrict the latter part of the loop to scalar_int_modes.
19041 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19042 and when considering SUBREG_REGs. Use is_int_mode instead of
19043 checking GET_MODE_CLASS against MODE_INT.
19045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19046 Alan Hayward <alan.hayward@arm.com>
19047 David Sherwood <david.sherwood@arm.com>
19049 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19050 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19051 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
19058 (force_to_mode): ...here. Keep xmode up-to-date and use it
19059 instead of GET_MODE (x).
19061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19062 Alan Hayward <alan.hayward@arm.com>
19063 David Sherwood <david.sherwood@arm.com>
19065 * optabs-query.h (extraction_insn::struct_mode): Change type to
19066 opt_scalar_int_mode and update comment.
19067 (extraction_insn::field_mode): Change type to scalar_int_mode.
19068 (extraction_insn::pos_mode): Likewise.
19069 * combine.c (make_extraction): Update accordingly.
19070 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19071 (get_optab_extraction_insn): Likewise.
19072 * recog.c (simplify_while_replacing): Likewise.
19073 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19074 parameter to opt_scalar_int_mode.
19076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19077 Alan Hayward <alan.hayward@arm.com>
19078 David Sherwood <david.sherwood@arm.com>
19080 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19081 to a scalar_int_mode instead of a machine_mode.
19082 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19083 (get_best_mode): Return a boolean and use a pointer argument to store
19084 the selected mode. Replace the limit mode parameter with a bit limit.
19085 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19086 for the values returned by bit_field_mode_iterator::next_mode.
19087 (store_bit_field): Update call to get_best_mode.
19088 (store_fixed_bit_field): Likewise.
19089 (extract_fixed_bit_field): Likewise.
19090 * expr.c (optimize_bitfield_assignment_op): Likewise.
19091 * fold-const.c (optimize_bit_field_compare): Likewise.
19092 (fold_truth_andor_1): Likewise.
19093 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19094 Update for new type of m_mode.
19095 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19102 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19103 (store_bit_field): Check is_a <scalar_int_mode> before calling
19104 strict_volatile_bitfield_p.
19105 (extract_bit_field): Likewise.
19107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19108 Alan Hayward <alan.hayward@arm.com>
19109 David Sherwood <david.sherwood@arm.com>
19111 * target.def (cstore_mode): Return a scalar_int_mode.
19112 * doc/tm.texi: Regenerate.
19113 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19114 * targhooks.h (default_cstore_mode): Likewise.
19115 * targhooks.c (default_cstore_mode): Likewise, using a forced
19117 * expmed.c (emit_cstore): Expect the target of the cstore to be
19120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19121 Alan Hayward <alan.hayward@arm.com>
19122 David Sherwood <david.sherwood@arm.com>
19124 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19126 (niter_desc): Likewise mode.
19127 (iv_analyze): Add a mode parameter.
19129 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19130 and change its type to scalar_int_mode.
19131 * loop-iv.c: Update commentary at head of file.
19132 (iv_constant): Pass the mode paraeter before the rtx it describes
19133 and change its type to scalar_int_mode. Remove VOIDmode handling.
19134 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19135 (iv_extend): Likewise.
19136 (shorten_into_mode): Likewise.
19137 (iv_add): Use scalar_int_mode.
19138 (iv_mult): Likewise.
19139 (iv_shift): Likewise.
19140 (canonicalize_iv_subregs): Likewise.
19141 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19142 it describes and change its mode to scalar_int_mode. Also change
19143 the type of the returned inner_mode to scalar_int_mode.
19144 (get_biv_step): Likewise, turning outer_mode from a pointer
19145 into a direct parameter. Update call to get_biv_step_1.
19146 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19147 iv_constant and get_biv_step.
19148 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19149 and change its type to scalar_int_mode. Don't initialise iv->mode
19150 to VOIDmode and remove later checks for its still being VOIDmode.
19151 Update calls to iv_analyze_op and iv_analyze_expr. Check
19152 is_a <scalar_int_mode> when changing the mode under consideration.
19153 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19154 Update call to iv_analyze_expr.
19155 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19156 inner register is not also a scalar_int_mode. Update call to
19158 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19159 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19160 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19161 separate mode class checks. Update calls to iv_analyze. Remove
19162 fix-up of VOIDmodes after iv_analyze_biv.
19163 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19164 don't have a scalar_int_mode. Update call to biv_p.
19166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19167 Alan Hayward <alan.hayward@arm.com>
19168 David Sherwood <david.sherwood@arm.com>
19170 * cfgexpand.c (convert_debug_memory_address): Use
19171 as_a <scalar_int_mode>.
19172 * combine.c (expand_compound_operation): Likewise.
19173 (make_extraction): Likewise.
19174 (change_zero_ext): Likewise.
19175 (simplify_comparison): Likewise.
19176 * cse.c (cse_insn): Likewise.
19177 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19178 (mem_loc_descriptor): Likewise.
19179 (loc_descriptor): Likewise.
19180 * expmed.c (init_expmed_one_mode): Likewise.
19181 (synth_mult): Likewise.
19182 (emit_store_flag_1): Likewise.
19183 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19184 of a comparison with size.
19185 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19186 (reduce_to_bit_field_precision): Likewise.
19187 * function.c (expand_function_end): Likewise.
19188 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19189 * loop-doloop.c (doloop_modify): Likewise.
19190 * optabs.c (expand_binop): Likewise.
19191 (expand_unop): Likewise.
19192 (expand_copysign_absneg): Likewise.
19193 (prepare_cmp_insn): Likewise.
19194 (maybe_legitimize_operand): Likewise.
19195 * recog.c (const_scalar_int_operand): Likewise.
19196 * rtlanal.c (get_address_mode): Likewise.
19197 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19198 (simplify_cond_clz_ctz): Likewise.
19199 * tree-nested.c (get_nl_goto_field): Likewise.
19200 * tree.c (build_vector_type_for_mode): Likewise.
19201 * var-tracking.c (use_narrower_mode): Likewise.
19203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19204 Alan Hayward <alan.hayward@arm.com>
19205 David Sherwood <david.sherwood@arm.com>
19207 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19208 * builtins.c (expand_builtin_signbit): Use it.
19209 * cfgexpand.c (expand_debug_expr): Likewise.
19210 * dojump.c (do_jump): Likewise.
19211 (do_compare_and_jump): Likewise.
19212 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19213 * expmed.c (make_tree): Likewise.
19214 * expr.c (expand_expr_real_2): Likewise.
19215 (expand_expr_real_1): Likewise.
19216 (try_casesi): Likewise.
19217 * fold-const-call.c (fold_const_call_ss): Likewise.
19218 * fold-const.c (unextend): Likewise.
19219 (extract_muldiv_1): Likewise.
19220 (fold_single_bit_test): Likewise.
19221 (native_encode_int): Likewise.
19222 (native_encode_string): Likewise.
19223 (native_interpret_int): Likewise.
19224 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19225 * internal-fn.c (expand_addsub_overflow): Likewise.
19226 (expand_neg_overflow): Likewise.
19227 (expand_mul_overflow): Likewise.
19228 (expand_arith_overflow): Likewise.
19229 * match.pd: Likewise.
19230 * stor-layout.c (layout_type): Likewise.
19231 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19232 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19233 * tree-ssanames.c (get_range_info): Likewise.
19234 * tree-switch-conversion.c (array_value_type) Likewise.
19235 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19236 (vect_recog_divmod_pattern): Likewise.
19237 (vect_recog_mixed_size_cond_pattern): Likewise.
19238 * tree-vrp.c (extract_range_basic): Likewise.
19239 (simplify_float_conversion_using_ranges): Likewise.
19240 * tree.c (int_fits_type_p): Likewise.
19241 * ubsan.c (instrument_bool_enum_load): Likewise.
19242 * varasm.c (mergeable_string_section): Likewise.
19243 (narrowing_initializer_constant_valid_p): Likewise.
19244 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
19251 * expr.c (alignment_for_piecewise_move): Use it instead of
19252 GET_CLASS_NARROWEST_MODE (MODE_INT).
19253 (push_block): Likewise.
19254 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19256 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19259 Alan Hayward <alan.hayward@arm.com>
19260 David Sherwood <david.sherwood@arm.com>
19262 * postreload.c (move2add_valid_value_p): Change the type of the
19263 mode parameter to scalar_int_mode.
19264 (move2add_use_add2_insn): Add a mode parameter and use it instead
19266 (move2add_use_add3_insn): Likewise.
19267 (reload_cse_move2add): Update accordingly.
19269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19270 Alan Hayward <alan.hayward@arm.com>
19271 David Sherwood <david.sherwood@arm.com>
19273 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19275 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19276 * optabs.c (expand_unop): Likewise.
19278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19279 Alan Hayward <alan.hayward@arm.com>
19280 David Sherwood <david.sherwood@arm.com>
19282 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19283 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19284 (popcount_loc_descriptor): Likewise.
19285 (bswap_loc_descriptor): Likewise.
19286 (rotate_loc_descriptor): Likewise.
19287 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19288 calling the functions above.
19290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19291 Alan Hayward <alan.hayward@arm.com>
19292 David Sherwood <david.sherwood@arm.com>
19294 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19296 (try_combine): Likewise.
19297 (simplify_if_then_else): Likewise.
19298 * cse.c (cse_insn): Likewise.
19299 * dwarf2out.c (mem_loc_descriptor): Likewise.
19300 * emit-rtl.c (gen_lowpart_common): Likewise.
19301 * simplify-rtx.c (simplify_truncation): Likewise.
19302 (simplify_binary_operation_1): Likewise.
19303 (simplify_const_relational_operation): Likewise.
19304 (simplify_ternary_operation): Likewise.
19305 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19308 Alan Hayward <alan.hayward@arm.com>
19309 David Sherwood <david.sherwood@arm.com>
19311 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19312 * reload.c (push_reload): Likewise.
19313 (find_reloads): Likewise.
19315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19316 Alan Hayward <alan.hayward@arm.com>
19317 David Sherwood <david.sherwood@arm.com>
19319 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19320 (make_compound_operation_int): Likewise.
19321 (change_zero_ext): Likewise.
19322 * expr.c (convert_move): Likewise.
19323 (convert_modes): Likewise.
19324 * fwprop.c (forward_propagate_subreg): Likewise.
19325 * loop-iv.c (get_biv_step_1): Likewise.
19326 * optabs.c (widen_operand): Likewise.
19327 * postreload.c (move2add_valid_value_p): Likewise.
19328 * recog.c (simplify_while_replacing): Likewise.
19329 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19330 (simplify_binary_operation_1): Likewise. Remove redundant
19331 mode equality check.
19333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19334 Alan Hayward <alan.hayward@arm.com>
19335 David Sherwood <david.sherwood@arm.com>
19337 * combine.c (combine_simplify_rtx): Add checks for
19338 is_a <scalar_int_mode>.
19339 (simplify_if_then_else): Likewise.
19340 (make_field_assignment): Likewise.
19341 (simplify_comparison): Likewise.
19342 * ifcvt.c (noce_try_bitop): Likewise.
19343 * loop-invariant.c (canonicalize_address_mult): Likewise.
19344 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19347 Alan Hayward <alan.hayward@arm.com>
19348 David Sherwood <david.sherwood@arm.com>
19350 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19351 is_a <scalar_int_mode> instead of != BLKmode.
19353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19354 Alan Hayward <alan.hayward@arm.com>
19355 David Sherwood <david.sherwood@arm.com>
19357 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19358 instead of != VOIDmode.
19359 * combine.c (if_then_else_cond): Likewise.
19360 (change_zero_ext): Likewise.
19361 * dwarf2out.c (mem_loc_descriptor): Likewise.
19362 (loc_descriptor): Likewise.
19363 * rtlanal.c (canonicalize_condition): Likewise.
19364 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19367 Alan Hayward <alan.hayward@arm.com>
19368 David Sherwood <david.sherwood@arm.com>
19370 * simplify-rtx.c (simplify_binary_operation_1): Use
19371 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19374 Alan Hayward <alan.hayward@arm.com>
19375 David Sherwood <david.sherwood@arm.com>
19377 * wide-int.h (int_traits<unsigned char>) New class.
19378 (int_traits<unsigned short>) Likewise.
19379 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19380 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19382 * combine.c (set_nonzero_bits_and_sign_copies): Use
19383 is_a <scalar_int_mode>.
19384 (find_split_point): Likewise.
19385 (combine_simplify_rtx): Likewise.
19386 (simplify_logical): Likewise.
19387 (expand_compound_operation): Likewise.
19388 (expand_field_assignment): Likewise.
19389 (make_compound_operation): Likewise.
19390 (extended_count): Likewise.
19391 (change_zero_ext): Likewise.
19392 (simplify_comparison): Likewise.
19393 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19394 (ucompare_loc_descriptor): Likewise.
19395 (minmax_loc_descriptor): Likewise.
19396 (mem_loc_descriptor): Likewise.
19397 (loc_descriptor): Likewise.
19398 * expmed.c (init_expmed_one_mode): Likewise.
19399 * lra-constraints.c (lra_constraint_offset): Likewise.
19400 * optabs.c (prepare_libcall_arg): Likewise.
19401 * postreload.c (move2add_note_store): Likewise.
19402 * reload.c (operands_match_p): Likewise.
19403 * rtl.h (load_extend_op): Likewise.
19404 * rtlhooks.c (gen_lowpart_general): Likewise.
19405 * simplify-rtx.c (simplify_truncation): Likewise.
19406 (simplify_unary_operation_1): Likewise.
19407 (simplify_binary_operation_1): Likewise.
19408 (simplify_const_binary_operation): Likewise.
19409 (simplify_const_relational_operation): Likewise.
19410 (simplify_subreg): Likewise.
19411 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19412 * var-tracking.c (adjust_mems): Likewise.
19413 (prepare_call_arguments): Likewise.
19415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19416 Alan Hayward <alan.hayward@arm.com>
19417 David Sherwood <david.sherwood@arm.com>
19419 * machmode.h (is_int_mode): New fuction.
19420 * combine.c (find_split_point): Use it.
19421 (combine_simplify_rtx): Likewise.
19422 (simplify_if_then_else): Likewise.
19423 (simplify_set): Likewise.
19424 (simplify_shift_const_1): Likewise.
19425 (simplify_comparison): Likewise.
19426 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19427 * cse.c (notreg_cost): Likewise.
19428 (cse_insn): Likewise.
19429 * cselib.c (cselib_lookup_1): Likewise.
19430 * dojump.c (do_jump_1): Likewise.
19431 (do_compare_rtx_and_jump): Likewise.
19432 * dse.c (get_call_args): Likewise.
19433 * dwarf2out.c (rtl_for_decl_init): Likewise.
19434 (native_encode_initializer): Likewise.
19435 * expmed.c (emit_store_flag_1): Likewise.
19436 (emit_store_flag): Likewise.
19437 * expr.c (convert_modes): Likewise.
19438 (store_field): Likewise.
19439 (expand_expr_real_1): Likewise.
19440 * fold-const.c (fold_read_from_constant_string): Likewise.
19441 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19442 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19443 * optabs.c (expand_binop): Likewise.
19444 (expand_unop): Likewise.
19445 (expand_abs_nojump): Likewise.
19446 (expand_one_cmpl_abs_nojump): Likewise.
19447 * simplify-rtx.c (mode_signbit_p): Likewise.
19448 (val_signbit_p): Likewise.
19449 (val_signbit_known_set_p): Likewise.
19450 (val_signbit_known_clear_p): Likewise.
19451 (simplify_relational_operation_1): Likewise.
19452 * tree.c (vector_type_mode): Likewise.
19454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19455 Alan Hayward <alan.hayward@arm.com>
19456 David Sherwood <david.sherwood@arm.com>
19458 * machmode.h (smallest_mode_for_size): Fix formatting.
19459 (smallest_int_mode_for_size): New function.
19460 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19461 instead of smallest_mode_for_size.
19462 * combine.c (make_extraction): Likewise.
19463 * config/arc/arc.c (arc_expand_movmem): Likewise.
19464 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19465 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19466 * config/s390/s390.c (s390_expand_insv): Likewise.
19467 * config/sparc/sparc.c (assign_int_registers): Likewise.
19468 * config/spu/spu.c (spu_function_value): Likewise.
19469 (spu_function_arg): Likewise.
19470 * coverage.c (get_gcov_type): Likewise.
19471 (get_gcov_unsigned_t): Likewise.
19472 * dse.c (find_shift_sequence): Likewise.
19473 * expmed.c (store_bit_field_1): Likewise.
19474 * expr.c (convert_move): Likewise.
19475 (store_field): Likewise.
19476 * internal-fn.c (expand_arith_overflow): Likewise.
19477 * optabs-query.c (get_best_extraction_insn): Likewise.
19478 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19479 * stor-layout.c (layout_type): Likewise.
19480 (initialize_sizetypes): Likewise.
19481 * targhooks.c (default_get_mask_mode): Likewise.
19482 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19485 Alan Hayward <alan.hayward@arm.com>
19486 David Sherwood <david.sherwood@arm.com>
19488 * machmode.h (opt_mode::else_blk): New function.
19489 (int_mode_for_mode): Declare.
19490 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19491 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19493 * cfgexpand.c (expand_debug_expr): Likewise.
19494 * combine.c (gen_lowpart_or_truncate): Likewise.
19495 (gen_lowpart_for_combine): Likewise.
19496 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19497 * config/avr/avr.c (avr_to_int_mode): Likewise.
19498 (avr_out_plus_1): Likewise.
19499 (avr_out_plus): Likewise.
19500 (avr_out_round): Likewise.
19501 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19502 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19503 (s390_expand_vcond): Likewise.
19504 * config/spu/spu.c (spu_split_immediate): Likewise.
19505 (spu_expand_mov): Likewise.
19506 * dse.c (get_stored_val): Likewise.
19507 * expmed.c (store_bit_field_1): Likewise.
19508 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19510 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19511 (extract_low_bits): Likewise.
19512 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19513 handling rather than repeating the check.
19514 (emit_group_store): Likewise.
19515 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19516 * optabs.c (expand_absneg_bit): Likewise.
19517 (expand_copysign_absneg): Likewise.
19518 (expand_copysign_bit): Likewise.
19519 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19520 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19521 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19522 * var-tracking.c (prepare_call_arguments): Likewise.
19523 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19524 int_mode_for_mode instead of mode_for_size.
19525 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19528 Alan Hayward <alan.hayward@arm.com>
19529 David Sherwood <david.sherwood@arm.com>
19531 * machmode.h (int_mode_for_size): New function.
19532 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19533 instead of mode_for_size.
19534 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19536 * combine.c (expand_field_assignment): Use int_mode_for_size
19537 instead of mode_for_size.
19538 (make_extraction): Likewise.
19539 (simplify_shift_const_1): Likewise.
19540 (simplify_comparison): Likewise.
19541 * dojump.c (do_jump): Likewise.
19542 * dwarf2out.c (mem_loc_descriptor): Likewise.
19543 * emit-rtl.c (init_derived_machine_modes): Likewise.
19544 * expmed.c (flip_storage_order): Likewise.
19545 (convert_extracted_bit_field): Likewise.
19546 * expr.c (copy_blkmode_from_reg): Likewise.
19547 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19548 * internal-fn.c (expand_mul_overflow): Likewise.
19549 * lower-subreg.c (simple_move): Likewise.
19550 * optabs-libfuncs.c (init_optabs): Likewise.
19551 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19552 * tree.c (vector_type_mode): Likewise.
19553 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19554 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19555 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19556 * tree-vect-stmts.c (vectorizable_load): Likewise.
19557 (vectorizable_store): Likewise.
19559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19560 Alan Hayward <alan.hayward@arm.com>
19561 David Sherwood <david.sherwood@arm.com>
19563 * coretypes.h (pod_mode): New type.
19564 (scalar_int_mode_pod): New typedef.
19565 * machmode.h (pod_mode): New class.
19566 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19567 * genmodes.c (emit_mode_int_n): Update accordingly.
19568 * lower-subreg.h (target_lower_subreg): Change type to
19569 scalar_int_mode_pod.
19570 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19571 scalar_int_mode_pod.
19573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19574 Alan Hayward <alan.hayward@arm.com>
19575 David Sherwood <david.sherwood@arm.com>
19577 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19578 machine_mode to scalar_int_mode.
19579 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19580 (rs6000_option_override_internal): Remove cast to int.
19581 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19582 machine_mode to scalar_int_mode.
19583 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19584 (rs6000_option_override_internal): Remove cast to int.
19585 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19586 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19588 * config/s390/s390.c (s390_expand_builtin): Likewise.
19589 * coretypes.h (scalar_int_mode): New type.
19590 (opt_scalar_int_mode): New typedef.
19591 * machmode.h (scalar_int_mode): New class.
19592 (scalar_int_mode::includes_p): New function.
19593 (byte_mode): Change type to scalar_int_mode.
19594 (word_mode): Likewise.
19595 (ptr_mode): Likewise.
19596 * emit-rtl.c (byte_mode): Likewise.
19597 (word_mode): Likewise.
19598 (ptr_mode): Likewise.
19599 (init_derived_machine_modes): Update accordingly.
19600 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19601 and MODE_PARTIAL_INT.
19602 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19603 opt_scalar_int_mode.
19605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19606 Alan Hayward <alan.hayward@arm.com>
19607 David Sherwood <david.sherwood@arm.com>
19609 * target.def (libgcc_floating_mode_supported_p): Take a
19611 * doc/tm.texi: Regenerate.
19612 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19614 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19615 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19619 Alan Hayward <alan.hayward@arm.com>
19620 David Sherwood <david.sherwood@arm.com>
19622 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19623 * doc/tm.texi: Regenerate.
19624 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19625 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19626 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19627 * targhooks.h (default_floatn_mode): Likewise.
19628 * targhooks.c (default_floatn_mode): Likewise.
19629 * tree.c (build_common_tree_nodes): Update accordingly.
19631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19632 Alan Hayward <alan.hayward@arm.com>
19633 David Sherwood <david.sherwood@arm.com>
19635 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19636 (mode_iterator::iterate_p): Likewise.
19637 (mode_iterator::get_wider): Likewise.
19638 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19641 Alan Hayward <alan.hayward@arm.com>
19642 David Sherwood <david.sherwood@arm.com>
19644 * coretypes.h (opt_scalar_float_mode): New typedef.
19645 * machmode.h (float_mode_for_size): New function.
19646 * emit-rtl.c (double_mode): Delete.
19647 (init_emit_once): Use float_mode_for_size.
19648 * stor-layout.c (layout_type): Likewise.
19649 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19652 Alan Hayward <alan.hayward@arm.com>
19653 David Sherwood <david.sherwood@arm.com>
19655 * output.h (assemble_real): Take a scalar_float_mode.
19656 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19657 * config/arm/arm.md (consttable_4): Likewise.
19658 (consttable_8): Likewise.
19659 (consttable_16): Likewise.
19660 * config/mips/mips.md (consttable_float): Likewise.
19661 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19662 * varasm.c (assemble_real): Take a scalar_float_mode.
19663 (output_constant_pool_2): Update accordingly.
19664 (output_constant): Likewise.
19666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19667 Alan Hayward <alan.hayward@arm.com>
19668 David Sherwood <david.sherwood@arm.com>
19670 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19671 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19672 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19673 (native_encode_real): Likewise.
19674 (native_interpret_real): Likewise.
19675 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19676 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19679 Alan Hayward <alan.hayward@arm.com>
19680 David Sherwood <david.sherwood@arm.com>
19682 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19683 <scalar_float_mode>. Simplify.
19684 (gen_extend_conv_libfunc): Likewise.
19686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19687 Alan Hayward <alan.hayward@arm.com>
19688 David Sherwood <david.sherwood@arm.com>
19690 * coretypes.h (scalar_float_mode): New type.
19691 * machmode.h (mode_traits::from_int): Use machine_mode if
19692 USE_ENUM_MODES is defined.
19693 (is_a): New function.
19695 (dyn_cast): Likewise.
19696 (scalar_float_mode): New class.
19697 (scalar_float_mode::includes_p): New function.
19698 (is_float_mode): Likewise.
19699 * gdbhooks.py (MachineModePrinter): New class.
19700 (build_pretty_printer): Use it for scalar_float_mode.
19701 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19702 (format_helper::format_helper): Turn into a template.
19703 * genmodes.c (get_mode_class): New function.
19704 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19705 or machine_mode if none.
19706 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19707 as_a <scalar_float_mode>.
19708 * dwarf2out.c (mem_loc_descriptor): Likewise.
19709 (insert_float): Likewise.
19710 (add_const_value_attribute): Likewise.
19711 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19712 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19713 (expand_unop): Update accordingly.
19714 (expand_abs_nojump): Likewise.
19715 (expand_copysign_absneg): Take a scalar_float_mode.
19716 (expand_copysign_bit): Likewise.
19717 (expand_copysign): Update accordingly.
19719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19720 Alan Hayward <alan.hayward@arm.com>
19721 David Sherwood <david.sherwood@arm.com>
19723 * coretypes.h (opt_mode): New class.
19724 * machmode.h (opt_mode): Likewise.
19725 (opt_mode::else_void): New function.
19726 (opt_mode::require): Likewise.
19727 (opt_mode::exists): Likewise.
19728 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19729 (GET_MODE_2XWIDER_MODE): Likewise.
19730 (mode_iterator::get_wider): Update accordingly.
19731 (mode_iterator::get_2xwider): Likewise.
19732 (mode_iterator::get_known_wider): Likewise, turning into a template.
19733 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19734 forcing a wider mode to exist.
19735 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19736 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19737 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19738 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19739 * lower-subreg.c (init_lower_subreg): Likewise.
19740 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19741 on the final iteration.
19742 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19743 a wider mode exists before asking for a move pattern.
19744 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19745 forcing a wider mode to exist.
19746 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19747 returning false if no such mode exists.
19748 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19749 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19750 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19751 Avoid checking for a MODE_INT if we already know the mode is not a
19753 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19754 forcing a wider mode to exist.
19755 (expmed_mult_highpart_optab): Likewise.
19756 (expmed_mult_highpart): Likewise.
19757 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19759 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19760 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19761 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19762 * internal-fn.c (expand_mul_overflow): Update use of
19763 GET_MODE_2XWIDER_MODE.
19764 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19765 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19766 GET_MODE_WIDER_MODE.
19767 (convert_plusminus_to_widen): Likewise.
19768 * tree-switch-conversion.c (array_value_type): Likewise.
19769 * var-tracking.c (emit_note_insn_var_location): Likewise.
19770 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19771 Return false inside rather than outside the loop if no wider mode
19773 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19774 and GET_MODE_2XWIDER_MODE
19775 (can_compare_p): Use else_void.
19776 * gdbhooks.py (OptMachineModePrinter): New class.
19777 (build_pretty_printer): Use it for opt_mode.
19779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19780 Alan Hayward <alan.hayward@arm.com>
19781 David Sherwood <david.sherwood@arm.com>
19783 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19784 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19787 Alan Hayward <alan.hayward@arm.com>
19788 David Sherwood <david.sherwood@arm.com>
19790 * machmode.h (mode_traits): New structure.
19791 (get_narrowest_mode): New function.
19792 (mode_iterator::start): Likewise.
19793 (mode_iterator::iterate_p): Likewise.
19794 (mode_iterator::get_wider): Likewise.
19795 (mode_iterator::get_known_wider): Likewise.
19796 (mode_iterator::get_2xwider): Likewise.
19797 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19798 (FOR_EACH_MODE): Likewise.
19799 (FOR_EACH_MODE_FROM): Likewise.
19800 (FOR_EACH_MODE_UNTIL): Likewise.
19801 (FOR_EACH_WIDER_MODE): Likewise.
19802 (FOR_EACH_2XWIDER_MODE): Likewise.
19803 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19804 * combine.c (simplify_comparison): Likewise
19805 * config/i386/i386.c (type_natural_mode): Likewise.
19806 * cse.c (cse_insn): Likewise.
19807 * dse.c (find_shift_sequence): Likewise.
19808 * emit-rtl.c (init_derived_machine_modes): Likewise.
19809 (init_emit_once): Likewise.
19810 * explow.c (hard_function_value): Likewise.
19811 * expmed.c (extract_fixed_bit_field_1): Likewise.
19812 (extract_bit_field_1): Likewise.
19813 (expand_divmod): Likewise.
19814 (emit_store_flag_1): Likewise.
19815 * expr.c (init_expr_target): Likewise.
19816 (convert_move): Likewise.
19817 (alignment_for_piecewise_move): Likewise.
19818 (widest_int_mode_for_size): Likewise.
19819 (emit_block_move_via_movmem): Likewise.
19820 (copy_blkmode_to_reg): Likewise.
19821 (set_storage_via_setmem): Likewise.
19822 (compress_float_constant): Likewise.
19823 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19824 * optabs-query.c (get_best_extraction_insn): Likewise.
19825 * optabs.c (expand_binop): Likewise.
19826 (expand_twoval_unop): Likewise.
19827 (expand_twoval_binop): Likewise.
19828 (widen_leading): Likewise.
19829 (widen_bswap): Likewise.
19830 (expand_parity): Likewise.
19831 (expand_unop): Likewise.
19832 (prepare_cmp_insn): Likewise.
19833 (prepare_float_lib_cmp): Likewise.
19834 (expand_float): Likewise.
19835 (expand_fix): Likewise.
19836 (expand_sfix_optab): Likewise.
19837 * postreload.c (move2add_use_add2_insn): Likewise.
19838 * reg-stack.c (reg_to_stack): Likewise.
19839 * reginfo.c (choose_hard_reg_mode): Likewise.
19840 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19841 * stor-layout.c (mode_for_size): Likewise.
19842 (smallest_mode_for_size): Likewise.
19843 (mode_for_vector): Likewise.
19844 (finish_bitfield_representative): Likewise.
19845 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19846 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19847 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19848 * var-tracking.c (prepare_call_arguments): Likewise.
19850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19851 Alan Hayward <alan.hayward@arm.com>
19852 David Sherwood <david.sherwood@arm.com>
19854 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19855 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19856 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19857 * machmode.h (mode_size): Move earlier in file.
19858 (mode_precision): Likewise.
19859 (mode_inner): Likewise.
19860 (mode_nunits): Likewise.
19861 (mode_unit_size): Likewise.
19862 (unit_unit_precision): Likewise.
19863 (mode_wider): Likewise.
19864 (mode_2xwider): Likewise.
19865 (machine_mode): New class.
19866 (mode_to_bytes): New function.
19867 (mode_to_bits): Likewise.
19868 (mode_to_precision): Likewise.
19869 (mode_to_inner): Likewise.
19870 (mode_to_unit_size): Likewise.
19871 (mode_to_unit_precision): Likewise.
19872 (mode_to_nunits): Likewise.
19873 (GET_MODE_SIZE): Use mode_to_bytes.
19874 (GET_MODE_BITSIZE): Use mode_to_bits.
19875 (GET_MODE_PRECISION): Use mode_to_precision.
19876 (GET_MODE_INNER): Use mode_to_inner.
19877 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19878 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19879 (GET_MODE_NUNITS): Use mode_to_nunits.
19880 * system.h (ALWAYS_INLINE): New macro.
19881 * config/powerpcspe/powerpcspe-c.c
19882 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19883 int for arg1_mode and arg2_mode.
19885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19886 Alan Hayward <alan.hayward@arm.com>
19887 David Sherwood <david.sherwood@arm.com>
19889 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19890 Prefix mode names with E_ in case statements.
19891 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19892 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19893 (aarch64_split_simd_move): Likewise.
19894 (aarch64_gen_storewb_pair): Likewise.
19895 (aarch64_gen_loadwb_pair): Likewise.
19896 (aarch64_gen_store_pair): Likewise.
19897 (aarch64_gen_load_pair): Likewise.
19898 (aarch64_get_condition_code_1): Likewise.
19899 (aarch64_constant_pool_reload_icode): Likewise.
19900 (get_rsqrte_type): Likewise.
19901 (get_rsqrts_type): Likewise.
19902 (get_recpe_type): Likewise.
19903 (get_recps_type): Likewise.
19904 (aarch64_gimplify_va_arg_expr): Likewise.
19905 (aarch64_simd_container_mode): Likewise.
19906 (aarch64_emit_load_exclusive): Likewise.
19907 (aarch64_emit_store_exclusive): Likewise.
19908 (aarch64_expand_compare_and_swap): Likewise.
19909 (aarch64_gen_atomic_cas): Likewise.
19910 (aarch64_emit_bic): Likewise.
19911 (aarch64_emit_atomic_swap): Likewise.
19912 (aarch64_emit_atomic_load_op): Likewise.
19913 (aarch64_evpc_trn): Likewise.
19914 (aarch64_evpc_uzp): Likewise.
19915 (aarch64_evpc_zip): Likewise.
19916 (aarch64_evpc_ext): Likewise.
19917 (aarch64_evpc_rev): Likewise.
19918 (aarch64_evpc_dup): Likewise.
19919 (aarch64_gen_ccmp_first): Likewise.
19920 (aarch64_gen_ccmp_next): Likewise.
19921 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19922 (alpha_emit_xfloating_libcall): Likewise.
19923 (emit_insxl): Likewise.
19924 (alpha_arg_type): Likewise.
19925 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19926 (arc_preferred_simd_mode): Likewise.
19927 (arc_secondary_reload): Likewise.
19928 (get_arc_condition_code): Likewise.
19929 (arc_print_operand): Likewise.
19930 (arc_legitimate_constant_p): Likewise.
19931 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19932 * config/arc/arc.md (casesi_load): Likewise.
19933 (casesi_compact_jump): Likewise.
19934 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19935 (cc_use_register): Likewise.
19936 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19937 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19938 (arm_init_iwmmxt_builtins): Likewise.
19939 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19940 (neon_expand_vector_init): Likewise.
19941 (arm_attr_length_move_neon): Likewise.
19942 (maybe_get_arm_condition_code): Likewise.
19943 (arm_emit_vector_const): Likewise.
19944 (arm_preferred_simd_mode): Likewise.
19945 (arm_output_iwmmxt_tinsr): Likewise.
19946 (thumb1_output_casesi): Likewise.
19947 (thumb2_output_casesi): Likewise.
19948 (arm_emit_load_exclusive): Likewise.
19949 (arm_emit_store_exclusive): Likewise.
19950 (arm_expand_compare_and_swap): Likewise.
19951 (arm_evpc_neon_vuzp): Likewise.
19952 (arm_evpc_neon_vzip): Likewise.
19953 (arm_evpc_neon_vrev): Likewise.
19954 (arm_evpc_neon_vtrn): Likewise.
19955 (arm_evpc_neon_vext): Likewise.
19956 (arm_validize_comparison): Likewise.
19957 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19958 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19959 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19960 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19961 (c6x_preferred_simd_mode): Likewise.
19962 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19963 (epiphany_rtx_costs): Likewise.
19964 * config/epiphany/predicates.md (proper_comparison_operator):
19966 * config/frv/frv.c (condexec_memory_operand): Likewise.
19967 (frv_emit_move): Likewise.
19968 (output_move_single): Likewise.
19969 (output_condmove_single): Likewise.
19970 (frv_hard_regno_mode_ok): Likewise.
19971 (frv_matching_accg_mode): Likewise.
19972 * config/h8300/h8300.c (split_adds_subs): Likewise.
19973 (h8300_rtx_costs): Likewise.
19974 (h8300_print_operand): Likewise.
19975 (compute_mov_length): Likewise.
19976 (output_logical_op): Likewise.
19977 (compute_logical_op_length): Likewise.
19978 (compute_logical_op_cc): Likewise.
19979 (h8300_shift_needs_scratch_p): Likewise.
19980 (output_a_shift): Likewise.
19981 (compute_a_shift_length): Likewise.
19982 (compute_a_shift_cc): Likewise.
19983 (expand_a_rotate): Likewise.
19984 (output_a_rotate): Likewise.
19985 * config/i386/i386.c (classify_argument): Likewise.
19986 (function_arg_advance_32): Likewise.
19987 (function_arg_32): Likewise.
19988 (function_arg_64): Likewise.
19989 (function_value_64): Likewise.
19990 (ix86_gimplify_va_arg): Likewise.
19991 (ix86_legitimate_constant_p): Likewise.
19992 (put_condition_code): Likewise.
19993 (split_double_mode): Likewise.
19994 (ix86_avx256_split_vector_move_misalign): Likewise.
19995 (ix86_expand_vector_logical_operator): Likewise.
19996 (ix86_split_idivmod): Likewise.
19997 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19998 (ix86_build_const_vector): Likewise.
19999 (ix86_build_signbit_mask): Likewise.
20000 (ix86_match_ccmode): Likewise.
20001 (ix86_cc_modes_compatible): Likewise.
20002 (ix86_expand_branch): Likewise.
20003 (ix86_expand_sse_cmp): Likewise.
20004 (ix86_expand_sse_movcc): Likewise.
20005 (ix86_expand_int_sse_cmp): Likewise.
20006 (ix86_expand_vec_perm_vpermi2): Likewise.
20007 (ix86_expand_vec_perm): Likewise.
20008 (ix86_expand_sse_unpack): Likewise.
20009 (ix86_expand_int_addcc): Likewise.
20010 (ix86_split_to_parts): Likewise.
20011 (ix86_vectorize_builtin_gather): Likewise.
20012 (ix86_vectorize_builtin_scatter): Likewise.
20013 (avx_vpermilp_parallel): Likewise.
20014 (inline_memory_move_cost): Likewise.
20015 (ix86_tieable_integer_mode_p): Likewise.
20016 (x86_maybe_negate_const_int): Likewise.
20017 (ix86_expand_vector_init_duplicate): Likewise.
20018 (ix86_expand_vector_init_one_nonzero): Likewise.
20019 (ix86_expand_vector_init_one_var): Likewise.
20020 (ix86_expand_vector_init_concat): Likewise.
20021 (ix86_expand_vector_init_interleave): Likewise.
20022 (ix86_expand_vector_init_general): Likewise.
20023 (ix86_expand_vector_set): Likewise.
20024 (ix86_expand_vector_extract): Likewise.
20025 (emit_reduc_half): Likewise.
20026 (ix86_emit_i387_round): Likewise.
20027 (ix86_mangle_type): Likewise.
20028 (ix86_expand_round_sse4): Likewise.
20029 (expand_vec_perm_blend): Likewise.
20030 (canonicalize_vector_int_perm): Likewise.
20031 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20032 (expand_vec_perm_1): Likewise.
20033 (expand_vec_perm_interleave3): Likewise.
20034 (expand_vec_perm_even_odd_pack): Likewise.
20035 (expand_vec_perm_even_odd_1): Likewise.
20036 (expand_vec_perm_broadcast_1): Likewise.
20037 (ix86_vectorize_vec_perm_const_ok): Likewise.
20038 (ix86_expand_vecop_qihi): Likewise.
20039 (ix86_expand_mul_widen_hilo): Likewise.
20040 (ix86_expand_sse2_abs): Likewise.
20041 (ix86_expand_pextr): Likewise.
20042 (ix86_expand_pinsr): Likewise.
20043 (ix86_preferred_simd_mode): Likewise.
20044 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20045 * config/i386/sse.md (*andnot<mode>3): Likewise.
20046 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20047 (*<code><mode>3): Likewise.
20048 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20049 (ia64_expand_atomic_op): Likewise.
20050 (ia64_arg_type): Likewise.
20051 (ia64_mode_to_int): Likewise.
20052 (ia64_scalar_mode_supported_p): Likewise.
20053 (ia64_vector_mode_supported_p): Likewise.
20054 (expand_vec_perm_broadcast): Likewise.
20055 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20056 (iq2000_function_arg_advance): Likewise.
20057 (iq2000_function_arg): Likewise.
20058 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20059 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20060 (m68k_libcall_value): Likewise.
20061 (m68k_function_value): Likewise.
20062 (sched_attr_op_type): Likewise.
20063 * config/mcore/mcore.c (mcore_output_move): Likewise.
20064 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20066 (microblaze_function_arg): Likewise.
20067 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20068 (mips_print_operand): Likewise.
20069 (mips_mode_ok_for_mov_fmt_p): Likewise.
20070 (mips_vector_mode_supported_p): Likewise.
20071 (mips_preferred_simd_mode): Likewise.
20072 (mips_expand_vpc_loongson_even_odd): Likewise.
20073 (mips_expand_vec_unpack): Likewise.
20074 (mips_expand_vi_broadcast): Likewise.
20075 (mips_expand_vector_init): Likewise.
20076 (mips_expand_vec_reduc): Likewise.
20077 (mips_expand_msa_cmp): Likewise.
20078 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20079 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20080 (cc_flags_for_mode): Likewise.
20081 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20082 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20083 (nds32_output_casesi_pc_relative): Likewise.
20084 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20085 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20086 (nvptx_gen_unpack): Likewise.
20087 (nvptx_gen_pack): Likewise.
20088 (nvptx_gen_shuffle): Likewise.
20089 (nvptx_gen_wcast): Likewise.
20090 (nvptx_preferred_simd_mode): Likewise.
20091 * config/pa/pa.c (pa_secondary_reload): Likewise.
20092 * config/pa/predicates.md (base14_operand): Likewise.
20093 * config/powerpcspe/powerpcspe-c.c
20094 (altivec_resolve_overloaded_builtin): Likewise.
20095 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20097 (rs6000_preferred_simd_mode): Likewise.
20098 (output_vec_const_move): Likewise.
20099 (rs6000_expand_vector_extract): Likewise.
20100 (rs6000_split_vec_extract_var): Likewise.
20101 (reg_offset_addressing_ok_p): Likewise.
20102 (rs6000_legitimate_offset_address_p): Likewise.
20103 (rs6000_legitimize_address): Likewise.
20104 (rs6000_emit_set_const): Likewise.
20105 (rs6000_const_vec): Likewise.
20106 (rs6000_emit_move): Likewise.
20107 (spe_build_register_parallel): Likewise.
20108 (rs6000_darwin64_record_arg_recurse): Likewise.
20109 (swap_selector_for_mode): Likewise.
20110 (spe_init_builtins): Likewise.
20111 (paired_init_builtins): Likewise.
20112 (altivec_init_builtins): Likewise.
20113 (do_load_for_compare): Likewise.
20114 (rs6000_generate_compare): Likewise.
20115 (rs6000_expand_float128_convert): Likewise.
20116 (emit_load_locked): Likewise.
20117 (emit_store_conditional): Likewise.
20118 (rs6000_output_function_epilogue): Likewise.
20119 (rs6000_handle_altivec_attribute): Likewise.
20120 (rs6000_function_value): Likewise.
20121 (emit_fusion_gpr_load): Likewise.
20122 (emit_fusion_p9_load): Likewise.
20123 (emit_fusion_p9_store): Likewise.
20124 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20125 (fusion_gpr_mem_load): Likewise.
20126 (fusion_addis_mem_combo_load): Likewise.
20127 (fusion_addis_mem_combo_store): Likewise.
20128 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20129 (fusion_gpr_mem_load): Likewise.
20130 (fusion_addis_mem_combo_load): Likewise.
20131 (fusion_addis_mem_combo_store): Likewise.
20132 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20134 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20135 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20136 (rs6000_preferred_simd_mode): Likewise.
20137 (output_vec_const_move): Likewise.
20138 (rs6000_expand_vector_extract): Likewise.
20139 (rs6000_split_vec_extract_var): Likewise.
20140 (reg_offset_addressing_ok_p): Likewise.
20141 (rs6000_legitimate_offset_address_p): Likewise.
20142 (rs6000_legitimize_address): Likewise.
20143 (rs6000_emit_set_const): Likewise.
20144 (rs6000_const_vec): Likewise.
20145 (rs6000_emit_move): Likewise.
20146 (rs6000_darwin64_record_arg_recurse): Likewise.
20147 (swap_selector_for_mode): Likewise.
20148 (paired_init_builtins): Likewise.
20149 (altivec_init_builtins): Likewise.
20150 (rs6000_expand_float128_convert): Likewise.
20151 (emit_load_locked): Likewise.
20152 (emit_store_conditional): Likewise.
20153 (rs6000_output_function_epilogue): Likewise.
20154 (rs6000_handle_altivec_attribute): Likewise.
20155 (rs6000_function_value): Likewise.
20156 (emit_fusion_gpr_load): Likewise.
20157 (emit_fusion_p9_load): Likewise.
20158 (emit_fusion_p9_store): Likewise.
20159 * config/rx/rx.c (rx_gen_move_template): Likewise.
20160 (flags_from_mode): Likewise.
20161 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20162 (s390_slb_comparison): Likewise.
20163 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20164 (s390_vector_mode_supported_p): Likewise.
20165 (s390_cc_modes_compatible): Likewise.
20166 (s390_match_ccmode_set): Likewise.
20167 (s390_canonicalize_comparison): Likewise.
20168 (s390_emit_compare_and_swap): Likewise.
20169 (s390_branch_condition_mask): Likewise.
20170 (s390_rtx_costs): Likewise.
20171 (s390_secondary_reload): Likewise.
20172 (__SECONDARY_RELOAD_CASE): Likewise.
20173 (s390_expand_cs): Likewise.
20174 (s390_preferred_simd_mode): Likewise.
20175 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20176 * config/sh/sh.c (sh_print_operand): Likewise.
20177 (dump_table): Likewise.
20178 (sh_secondary_reload): Likewise.
20179 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20180 * config/sh/sh.md (casesi_worker_1): Likewise.
20181 (casesi_worker_2): Likewise.
20182 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20183 (fcc_comparison_operator): Likewise.
20184 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20185 (emit_soft_tfmode_cvt): Likewise.
20186 (sparc_preferred_simd_mode): Likewise.
20187 (output_cbranch): Likewise.
20188 (sparc_print_operand): Likewise.
20189 (sparc_expand_vec_perm_bmask): Likewise.
20190 (vector_init_bshuffle): Likewise.
20191 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20192 (spu_vector_mode_supported_p): Likewise.
20193 (spu_expand_insv): Likewise.
20194 (spu_emit_branch_or_set): Likewise.
20195 (spu_handle_vector_attribute): Likewise.
20196 (spu_builtin_splats): Likewise.
20197 (spu_builtin_extract): Likewise.
20198 (spu_builtin_promote): Likewise.
20199 (spu_expand_sign_extend): Likewise.
20200 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20201 (tilegx_simd_int): Likewise.
20202 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20203 (tilepro_simd_int): Likewise.
20204 * config/v850/v850.c (const_double_split): Likewise.
20205 (v850_print_operand): Likewise.
20206 (ep_memory_offset): Likewise.
20207 * config/vax/vax.c (vax_rtx_costs): Likewise.
20208 (vax_output_int_move): Likewise.
20209 (vax_output_int_add): Likewise.
20210 (vax_output_int_subtract): Likewise.
20211 * config/visium/predicates.md (visium_branch_operator): Likewise.
20212 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20213 (visium_print_operand_address): Likewise.
20214 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20215 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20216 (xtensa_expand_conditional_branch): Likewise.
20217 (xtensa_copy_incoming_a7): Likewise.
20218 (xtensa_output_literal): Likewise.
20219 * dfp.c (decimal_real_maxval): Likewise.
20220 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20223 Alan Hayward <alan.hayward@arm.com>
20224 David Sherwood <david.sherwood@arm.com>
20226 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20227 (mode_nunits_inline): Likewise.
20228 (mode_inner_inline): Likewise.
20229 (mode_unit_size_inline): Likewise.
20230 (mode_unit_precision_inline): Likewise.
20231 (emit_insn_modes_h): Likewise. Also emit a #define of the
20233 (emit_mode_wider): Add an E_ prefix to mode names.
20234 (emit_mode_complex): Likewise.
20235 (emit_mode_inner): Likewise.
20236 (emit_mode_adjustments): Likewise.
20237 (emit_mode_int_n): Likewise.
20238 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20239 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20240 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20241 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20242 (CRC32_BUILTIN, ENTRY): Likewise.
20243 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20244 (aarch64_pop_regs): Likewise.
20245 (aarch64_process_components): Likewise.
20246 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20247 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20248 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20249 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20250 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20251 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20253 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20254 (pcmpistr): Likewise.
20255 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20256 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20257 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20259 * config/rl78/rl78.c (mduc_regs): Likewise.
20260 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20261 (htm_expand_builtin): Likewise.
20262 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20263 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20264 * config/xtensa/xtensa.c (print_operand): Likewise.
20265 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20266 (NUM_MODE_VECTOR_INT): Likewise.
20267 * genoutput.c (null_operand): Likewise.
20268 (output_operand_data): Likewise.
20269 * genrecog.c (print_parameter_value): Likewise.
20270 * lra.c (debug_operand_data): Likewise.
20272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20274 * dwarf2out.c (scompare_loc_descriptor_wide)
20275 (scompare_loc_descriptor_narrow): New functions, split out from...
20276 (scompare_loc_descriptor): ...here.
20277 * expmed.c (emit_store_flag_int): New function, split out from...
20278 (emit_store_flag): ...here.
20280 2017-08-30 Richard Biener <rguenther@suse.de>
20282 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20283 (dwarf2out_early_finish): Move setting of AT_pubnames from
20284 early debug output to early finish.
20286 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20288 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20289 and -mdata-region to the assembler.
20291 2017-08-30 Richard Biener <rguenther@suse.de>
20293 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20295 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20296 (dwarf2out_early_global_decl): Only generate a DIE for the
20297 abstract origin if it doesn't already exist or is a declaration DIE.
20298 (resolve_addr): Do not add the linkage name twice when
20299 generating a stub DIE for the DW_TAG_GNU_call_site target.
20301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20303 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20304 Use machine_mode rather than int for arg1_mode.
20306 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20309 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20310 that the second argument of the built-in functions to unpack
20311 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20312 switch statement instead a lot of if statements.
20313 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20314 Allow 64-bit values to be in Altivec registers as well as
20315 traditional floating point registers.
20316 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20318 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20320 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20321 MAX_REGS_PER_ADDRESS == 1.
20323 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20325 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20326 * config/i386/i386.c (ix86_option_override_internal): Simplify
20327 setting of opts->x_flag_entry.
20329 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20330 Jakub Jelinek <jakub@redhat.com>
20331 Richard Biener <rguenther@suse.de>
20333 PR tree-optimization/81503
20334 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20335 folded constant fits in the target type; reorder tests for clarity.
20337 2017-08-29 Martin Liska <mliska@suse.cz>
20339 * passes.def: Include pass_lower_switch.
20340 * stmt.c (dump_case_nodes): Remove and move to
20341 tree-switch-conversion.
20342 (case_values_threshold): Likewise.
20343 (expand_switch_as_decision_tree_p): Likewise.
20344 (emit_case_decision_tree): Likewise.
20345 (expand_case): Likewise.
20346 (balance_case_nodes): Likewise.
20347 (node_has_low_bound): Likewise.
20348 (node_has_high_bound): Likewise.
20349 (node_is_bounded): Likewise.
20350 (emit_case_nodes): Likewise.
20351 (struct simple_case_node): New struct.
20352 (add_case_node): Remove.
20353 (emit_case_dispatch_table): Use vector instead of case_list.
20354 (reset_out_edges_aux): Remove.
20355 (compute_cases_per_edge): Likewise.
20356 (expand_case): Build list of simple_case_node.
20357 (expand_sjlj_dispatch_table): Use it.
20358 * tree-switch-conversion.c (struct case_node): Moved from
20359 stmt.c and adjusted.
20360 (emit_case_nodes): Likewise.
20361 (node_has_low_bound): Likewise.
20362 (node_has_high_bound): Likewise.
20363 (node_is_bounded): Likewise.
20364 (case_values_threshold): Likewise.
20365 (reset_out_edges_aux): Likewise.
20366 (compute_cases_per_edge): Likewise.
20367 (add_case_node): Likewise.
20368 (dump_case_nodes): Likewise.
20369 (balance_case_nodes): Likewise.
20370 (expand_switch_as_decision_tree_p): Likewise.
20371 (emit_jump): Likewise.
20372 (emit_case_decision_tree): Likewise.
20373 (try_switch_expansion): Likewise.
20374 (do_jump_if_equal): Likewise.
20375 (emit_cmp_and_jump_insns): Likewise.
20376 (fix_phi_operands_for_edge): New function.
20377 (record_phi_operand_mapping): Likewise.
20378 (class pass_lower_switch): New pass.
20379 (pass_lower_switch::execute): New function.
20380 (make_pass_lower_switch): Likewise.
20381 (conditional_probability):
20382 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20383 * tree-pass.h: Add make_pass_lower_switch.
20385 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20388 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20391 2017-08-29 Richard Biener <rguenther@suse.de>
20393 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20394 we do not add a DW_AT_inline attribute twice.
20395 (gen_subprogram_die): Remove code setting DW_AT_inline on
20396 DECL_ABSTRACT_P nodes.
20398 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20400 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20401 calls to internal functions.
20402 (gimplify_modify_expr): Likewise.
20403 * tree-call-cdce.c (use_internal_fn): Likewise.
20404 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20405 (convert_to_divmod): Set the nothrow flag.
20406 * tree-if-conv.c (predicate_mem_writes): Likewise.
20407 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20408 (vectorizable_call): Likewise.
20409 (vectorizable_store): Likewise.
20410 (vectorizable_load): Likewise.
20411 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20412 (vect_recog_mask_conversion_pattern): Likewise.
20414 2017-08-29 Martin Liska <mliska@suse.cz>
20417 * gcc.c (driver_handle_option): Add new argument.
20418 * opts-common.c (handle_option): Pass
20419 target_option_override_hook.
20420 * opts-global.c (lang_handle_option): Add new option.
20421 (set_default_handlers): Add new argument.
20422 (decode_options): Likewise.
20423 * opts.c (target_handle_option): Likewise.
20424 (common_handle_option): Call target_option_override_hook.
20425 * opts.h (struct cl_option_handler_func): Add hook for
20426 target option override.
20427 (struct cl_option_handlers): Likewise.
20428 (set_default_handlers): Add new argument.
20429 (decode_options): Likewise.
20430 (common_handle_option): Likewise.
20431 (target_handle_option): Likewise.
20432 * toplev.c (toplev::main): Pass targetm.target_option.override
20435 2017-08-29 Richard Biener <rguenther@suse.de>
20436 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20438 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20439 life to the active subtree.
20441 2017-08-28 Jeff Law <law@redhat.com>
20443 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20444 derive_equivalences.
20445 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20446 Code moved into....
20447 (edge_info::derive_equivalences): New private member function
20449 * tree-ssa-dom.c (class edge_info): Changed from a struct
20450 to a class. Add ctor/dtor, methods and data members.
20451 (edge_info::edge_info): Renamed from allocate_edge_info.
20452 Initialize additional members.
20453 (edge_info::~edge_info): New.
20454 (free_dom_edge_info): Delete the edge info.
20455 (record_edge_info): Use new class & associated member functions.
20456 Tighten forms for testing for edge equivalences.
20457 (record_temporary_equivalences): Iterate over the simple
20458 equivalences rather than assuming there's only one per edge.
20459 (cprop_into_successor_phis): Iterate over the simple
20460 equivalences rather than assuming there's only one per edge.
20461 (optimize_stmt): Use operand_equal_p rather than pointer
20462 equality for mini-DSE code.
20464 2017-08-28 Nathan Sidwell <nathan@acm.org>
20466 * gcc.c (execute): Fold SIGPIPE handling into switch
20467 statement. Adjust internal error message.
20469 2017-08-28 Richard Biener <rguenther@suse.de>
20472 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20473 Do nothing for removed DIEs.
20475 2017-08-28 Richard Biener <rguenther@suse.de>
20477 PR tree-optimization/81977
20478 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20481 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20484 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20485 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20486 using targetm.gen_mem_thread_fence.
20488 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20491 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20492 predicate to register_operand. Reorder operands.
20493 (*btr<mode>): Ditto.
20494 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20495 (*btr<mode>_mask): Ditto.
20497 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20499 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20500 * config/rs6000/xmmintrin.h: New file.
20501 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20503 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20506 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20507 parameter and_insn and return it.
20508 (recombine_lvx_pattern): Insert a copy to ensure availability of
20509 the base register of the copied masking operation at the point of
20510 the instruction replacement.
20511 (recombine_stvx_pattern): Likewise.
20513 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20515 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20516 undocumented switches.
20517 (-mpower9-dform-vector): Likewise.
20518 (-mpower9-dform): Likewise.
20519 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20520 comments to delete references to -mpower9-dform* switches.
20521 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20522 Delete reference to -mpower9-dform* switches, test for
20523 -mpower9-vector instead.
20524 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20525 (OTHER_P9_VECTOR_MASKS): Likewise.
20526 (POWERPC_MASKS): Likewise.
20527 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20528 tests against -mpower9-dform* to -mpower9-vector. Delete code
20529 that checked for -mpower9-dform* consistancy with other options.
20530 Add test for -mpower9-misc to enable other power9 switches.
20531 (rs6000_init_hard_regno_mode_ok): Likewise.
20532 (rs6000_option_override_internal): Likewise.
20533 (rs6000_emit_prologue): Likewise.
20534 (rs6000_emit_epilogue): Likewise.
20535 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20536 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20537 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20539 (emit_fusion_p9_store): Likewise.
20540 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20541 resetting these macros if the assembler does not support ISA 3.0
20543 (TARGET_P9_DFORM_VECTOR): Likewise.
20544 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20545 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20547 2017-08-25 Alan Modra <amodra@gmail.com>
20549 PR rtl-optimization/81747
20550 * cse.c (cse_extended_basic_block): Don't attempt to record
20551 equivalences for degenerate conditional branches that branch
20552 to their fall-through.
20554 2017-08-24 Martin Sebor <msebor@redhat.com>
20556 PR middle-end/81908
20557 * gimple-fold.c (size_must_be_zero_p): New function.
20558 (gimple_fold_builtin_memory_op): Call it.
20560 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20562 * config/rs6000/mm_malloc.h: New file.
20564 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20566 PR tree-optimization/81913
20567 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20568 analysis when either IVs in condition can wrap.
20570 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20572 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20573 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20575 2017-08-24 Richard Biener <rguenther@suse.de>
20578 * targhooks.c (default_target_can_inline_p): Properly
20579 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20580 is present and always compare.
20581 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20582 infer -mfpmath=sse from TARGET_SSE_P.
20583 (ix86_can_inline_p): Properly use target_option_default_node when
20584 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20586 2017-08-24 Richard Biener <rguenther@suse.de>
20589 * dwarf2out.c (output_die): Handle flag_generate_offload like
20591 (output_comp_unit): Likewise.
20592 (gen_array_type_die): Likewise.
20593 (dwarf2out_early_finish): Likewise.
20594 (note_variable_value_in_expr): Likewise.
20595 (dwarf2out_finish): Likewise. Adjust assert.
20596 * cgraphunit.c (symbol_table::compile): Move setting of
20597 flag_generate_offload earlier ...
20598 (symbol_table::finalize_compilation_unit): ... here, before
20599 early debug finalization.
20601 2017-08-24 Richard Biener <rguenther@suse.de>
20603 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20604 and ipa-fnsummary.h.
20605 (ix86_can_inline_p): When ix86_fpmath flags do not match
20606 check whether the callee uses FP math at all.
20608 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20610 PR middle-end/81931
20611 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20612 instead of TYPE_PRECISION.
20614 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20615 Alan Hayward <alan.hayward@arm.com>
20616 David Sherwood <david.sherwood@arm.com>
20618 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20620 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20622 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20624 (ix86_frame::stack_realign_allocate): New field.
20625 (struct machine_frame_state): Modify comments.
20626 (machine_frame_state::sp_realigned_fp_end): New field.
20627 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20628 layout calculation.
20629 (sp_valid_at): Add assertion to assure no attempt to access invalid
20630 offset of a realigned stack.
20631 (fp_valid_at): Likewise.
20632 (choose_baseaddr): Modify comments.
20633 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20634 ix86_expand_prologue.
20635 (ix86_expand_prologue): Modify stack realignment and allocation.
20636 (ix86_expand_epilogue): Modify comments.
20637 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20638 avx2_runtime, avx512f, and avx512f_runtime.
20640 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20642 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20643 (mstackrealign): Do not init to -1.
20644 * config/i386/i386.c (ix86_option_override_internal):
20645 Check opts_set, not opts when setting default value of
20646 opts->x_ix86_force_align_arg_pointer.
20648 2017-08-23 Richard Biener <rguenther@suse.de>
20650 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20651 lang_hooks.decl_printable_name.
20652 * print-rtl-function.c (print_rtx_function): Likewise.
20653 * tree-pretty-print.c (dump_function_header): Likewise.
20655 2017-08-23 Richard Biener <rguenther@suse.de>
20658 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20659 -g0 at compile-time.
20661 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20663 PR middle-end/19706
20664 * doc/sourcebuild.texi (Other hardware attributes):
20667 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20669 PR middle-end/19706
20670 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20671 Fix single-use check.
20673 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20675 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20677 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20679 * doc/install.texi: Modify to add more details on running selected
20682 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20684 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20685 is combined with -mabi=ms.
20686 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20689 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20691 PR tree-optimization/81488
20692 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20693 and cached_basis fields.
20694 (MAX_SPREAD): New constant.
20695 (alloc_cand_and_find_basis): Initialize new fields.
20696 (clear_visited): New function.
20697 (create_phi_basis_1): Rename from create_phi_basis, set visited
20698 and cached_basis fields.
20699 (create_phi_basis): New wrapper function.
20700 (phi_add_costs_1): Rename from phi_add_costs, add spread
20701 parameter, set visited field, short-circuit when limits reached.
20702 (phi_add_costs): New wrapper function.
20703 (record_phi_increments_1): Rename from record_phi_increments, set
20705 (record_phi_increments): New wrapper function.
20706 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20707 (phi_incr_cost): New wrapper function.
20708 (all_phi_incrs_profitable_1): Rename from
20709 all_phi_incrs_profitable, set visited field.
20710 (all_phi_incrs_profitable): New wrapper function.
20712 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20713 Alan Hayward <alan.hayward@arm.com>
20714 David Sherwood <david.sherwood@arm.com>
20716 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20717 that takes the outer and inner modes.
20718 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20719 comparison as the canonical test for a paradoxical subreg.
20720 * combine.c (simplify_set): Use paradoxical_subreg_p.
20721 (make_extraction): Likewise.
20722 (force_to_mode): Likewise.
20723 (rtx_equal_for_field_assignment_p): Likewise.
20724 (gen_lowpart_for_combine): Likewise.
20725 (simplify_comparison): Likewise.
20726 * cse.c (equiv_constant): Likewise.
20727 * expmed.c (store_bit_field_1): Likewise.
20728 * final.c (alter_subreg): Likewise.
20729 * fwprop.c (propagate_rtx): Likewise.
20730 (forward_propagate_subreg): Likewise.
20731 * ira-conflicts.c (ira_build_conflicts): Likewise.
20732 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20733 * lra-constraints.c (curr_insn_transform): Likewise.
20734 (split_reg): Likewise.
20735 * lra-eliminations.c (move_plus_up): Likewise.
20736 (lra_eliminate_regs_1): Likewise.
20737 * recog.c (general_operand): Likewise.
20738 * ree.c (combine_reaching_defs): Likewise.
20739 * reload.c (push_reload): Likewise.
20740 (find_reloads): Likewise.
20741 * reload1.c (elimination_effects): Likewise.
20742 (compute_reload_subreg_offset): Likewise.
20743 (choose_reload_regs): Likewise.
20744 * rtlanal.c (subreg_lsb_1): Likewise.
20745 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20746 (simplify_subreg): Likewise.
20747 * var-tracking.c (track_loc_p): Likewise.
20748 * emit-rtl.c (byte_lowpart_offset): Likewise.
20749 (paradoxical_subreg_p): Delete out-of-line definition.
20751 2017-08-22 Jeff Law <law@redhat.com>
20753 PR tree-optimization/81741
20754 PR tree-optimization/71947
20755 * tree-ssa-dom.c: Include tree-inline.h.
20756 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20757 equivalences if one is more expensive to compute than the other.
20758 * tree-ssa-scopedtables.h (class const_or_copies): Make
20759 record_const_or_copy_raw method private.
20760 (class avail_exprs_stack): New method simplify_binary_operation.
20761 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20762 avail_exprs_stack::simplify_binary_operation as needed.
20763 (avail_exprs_stack::simplify_binary_operation): New function.
20765 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20767 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20768 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20769 (DOT_SYMBOLS): Likewise.
20770 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20771 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20772 (RS6000_ABI_NAME): Likewise.
20773 (TARGET_CMODEL): Likewise.
20774 (TOC_SECTION_ASM_OP): Likewise.
20775 (SET_CMODEL): New macro.
20776 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20778 2017-08-22 Richard Biener <rguenther@suse.de>
20780 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20781 to allow for free-lang-data replacements similar to verify_type_variant.
20783 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20784 Alan Hayward <alan.hayward@arm.com>
20785 David Sherwood <david.sherwood@arm.com>
20787 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20788 VOIDmode for the LABEL_REF.
20790 2017-08-22 Richard Biener <rguenther@suse.de>
20792 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20794 2017-08-22 Richard Biener <rguenther@suse.de>
20796 * common.opt (feliminate-dwarf2-dups): Ignore.
20797 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20798 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20799 same_die_p_wrap, compute_section_prefix,
20800 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20801 (comdat_symbol_id, comdat_symbol_number): Likewise.
20802 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20804 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20805 (output_die): Mark unreachable path unreachable.
20806 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20807 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20808 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20809 (dwarf2out_early_finish): Likewise.
20811 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20813 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20815 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20818 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20819 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20820 (avr_attribute_table) <io, io_low, address>: Initialize
20821 .decl_required with true.
20823 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20825 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20826 undocumented debugging options.
20827 (-mvsx-scalar-double): Likewise.
20828 (-mallow-df-permute): Likewise.
20829 (-mvectorize-builtins): Likewise.
20830 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20831 (rs6000_builtin_vectorized_function): Likewise.
20832 (rs6000_builtin_md_vectorized_function): Likewise.
20833 (rs6000_opt_vars): Likewise.
20835 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20838 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20839 (*btrq_imm): Rename from *btrq.
20840 (*btcq_imm): Rename from *btcq.
20841 (btsc): New code attribute.
20842 (*<btsc><mode>): New insn pattern.
20843 (*btr<mode>): Ditto.
20844 (*<btsc><mode>_mask): New insn_and_split pattern.
20845 (*btr<mode>_mask): Ditto.
20847 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20848 Alan Hayward <alan.hayward@arm.com>
20849 David Sherwood <david.sherwood@arm.com>
20851 * function.c (pad_below): Simplify padding calculation.
20853 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20854 Alan Hayward <alan.hayward@arm.com>
20855 David Sherwood <david.sherwood@arm.com>
20857 * target.def (function_prologue): Remove frame size argument.
20858 (function_epilogue): Likewise.
20859 * doc/tm.texi: Regenerate.
20860 * final.c (final_start_function): Update call to function_prologue.
20861 (final_end_function): Update call to function_epilogue.
20862 (default_function_pro_epilogue): Remove frame size argument.
20863 * output.h (default_function_pro_epilogue): Likewise.
20864 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20865 (arm_output_function_prologue): Likewise.
20866 * config/frv/frv.c (frv_function_prologue): Likewise.
20867 (frv_function_epilogue): Likewise.
20868 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20869 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20870 (ia64_output_function_epilogue): Likewise.
20871 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20872 (m32r_output_function_epilogue): Likewise.
20873 * config/microblaze/microblaze.c (microblaze_function_prologue)
20874 (microblaze_function_epilogue): Likewise.
20875 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20876 (mips_output_function_epilogue): Likewise.
20877 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20878 (mmix_target_asm_function_epilogue): Likewise.
20879 * config/msp430/msp430.c (msp430_start_function): Likewise.
20880 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20881 (nds32_asm_function_epilogue): Likewise.
20882 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20883 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20884 (pa_output_function_epilogue): Likewise.
20885 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20886 (rs6000_output_function_epilogue): Likewise.
20887 * config/rl78/rl78.c (rl78_start_function): Likewise.
20888 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20889 (rs6000_output_function_epilogue): Likewise.
20890 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20891 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20892 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20893 (sparc_asm_function_epilogue): Likewise.
20895 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20897 * tree.h (type_has_mode_precision_p): New function.
20898 * convert.c (convert_to_integer_1): Use it.
20899 * expr.c (expand_expr_real_2): Likewise.
20900 (expand_expr_real_1): Likewise.
20901 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20902 * match.pd: Likewise.
20903 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20904 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20905 * tree-tailcall.c (process_assignment): Likewise.
20906 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20907 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20908 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20909 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20910 (vectorizable_assignment): Likewise.
20911 (vectorizable_shift): Likewise.
20912 (vectorizable_operation): Likewise.
20913 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20915 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20917 * match.pd: Add pow (C, x) simplification.
20919 2017-08-21 Richard Biener <rguenther@suse.de>
20921 PR tree-optimization/81900
20922 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20923 for blocks with abnormal predecessors.
20924 (compute_antic): Do not set visited flag prematurely.
20926 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20929 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20931 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20933 * stor-layout.h (vector_type_mode): Move to...
20934 * tree.h (vector_type_mode): ...here.
20935 * stor-layout.c (vector_type_mode): Move to...
20936 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20938 2017-08-21 Richard Biener <rguenther@suse.de>
20940 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20941 register_external_die hooks.
20942 (debug_false_tree_charstarstar_uhwistar): Declare.
20943 (debug_nothing_tree_charstar_uhwi): Likewise.
20944 * debug.c (do_nothing_debug_hooks): Adjust.
20945 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20946 (debug_nothing_tree_charstar_uhwi): Likewise.
20947 * dbxout.c (dbx_debug_hooks): Adjust.
20948 (xcoff_debug_hooks): Likewise.
20949 * sdbout.c (sdb_debug_hooks): Likewise.
20950 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20951 * dwarf2out.c (macinfo_label_base): New global.
20952 (dwarf2out_register_external_die): New function for the
20953 register_external_die hook.
20954 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20955 (dwarf2_debug_hooks): Use them.
20956 (dwarf2_lineno_debug_hooks): Adjust.
20957 (struct die_struct): Add with_offset flag.
20958 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20959 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20960 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20961 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20962 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20963 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20964 defining section names for the early LTO debug variants.
20965 (reset_indirect_string): New helper.
20966 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20967 (print_dw_val): Add support for offsetted symbol references.
20968 (get_ultimate_context): Split out from is_cxx.
20969 (is_cxx): Use get_ultimate_context.
20970 (is_fortran): Add decl overload.
20971 (compute_comp_unit_symbol): Split out worker from
20972 compute_section_prefix.
20973 (compute_section_prefix): Call compute_comp_unit_symbol and
20974 set comdat_type_p here.
20975 (output_die): Skip DIE symbol output for the LTO added one.
20976 Handle DIE symbol references with offset.
20977 (output_comp_unit): Guard section name mangling properly.
20978 For LTO debug sections emit a symbol at the section beginning
20979 which we use to refer to its DIEs.
20980 (add_abstract_origin_attribute): For DIEs registered via
20981 dwarf2out_register_external_die directly refer to the early
20982 DIE rather than indirectly through the shadow one we created.
20983 Remove obsolete call to dwarf2out_abstract_function for
20984 non-function/block origins.
20985 (gen_array_type_die): When generating early LTO debug do
20986 not emit DW_AT_string_length.
20987 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20988 late when in LTO. As suggested place a gcc_unreachable for
20989 the DECL_ABSTRACT_P case.
20990 (gen_subprogram_die): Avoid another specification DIE
20991 for early built declarations/definitions for the late LTO case.
20992 (gen_variable_die): Add type references for late duplicated VLA dies
20994 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20995 we have the abstract instance already.
20996 (process_scope_var): Adjust decl DIE contexts in LTO which
20997 first puts them in limbo.
20998 (gen_decl_die): Do not generate type DIEs late apart from
20999 types for VLAs or for decls we do not yet have a DIE. Do not
21000 call dwarf2out_abstract_function late.
21001 (dwarf2out_early_global_decl): Make sure to create DIEs
21002 for abstract instances of a decl first.
21003 (dwarf2out_late_global_decl): Adjust comment.
21004 (output_macinfo_op): With multiple macro sections use
21005 macinfo_label_base to distinguish labels.
21006 (output_macinfo): Likewise. Update macinfo_label_base.
21007 Pass in the line info label.
21008 (note_variable_value_in_expr): When generating LTO resolve
21009 all variable values here by generating DIEs as needed.
21010 (init_sections_and_labels): Add early LTO debug flag parameter
21011 and generate different sections and names if set. Add generation
21012 counter for the labels so we can have multiple of them.
21013 (reset_dies): Helper to allow DIEs to be output multiple times.
21014 (dwarf2out_finish): When outputting DIEs to the fat part of an
21015 LTO object first reset DIEs.
21016 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21017 (modified_type_die): Check for decl_ultimate_origin being self
21019 (gen_type_die_with_usage): Likewise.
21020 (gen_typedef_die): Allow decl_ultimate_origin being self.
21021 (set_decl_abstract_flags): Remove.
21022 (set_block_abstract_flags): Likewise.
21023 (dwarf2out_abstract_function): Treat the early generated DIEs
21024 as the abstract copy and only add DW_AT_inline and
21025 DW_AT_artificial here and call set_decl_origin_self.
21026 If the DIE has an abstract origin don't do anything.
21027 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21028 if we have none yet (Go fails to build one, PR78628).
21029 (variably_modified_type_p): Prevent endless recursion for Ada
21030 cyclic pointer types.
21031 * lto-streamer-in.c: Include debug.h.
21032 (dref_queue): New global.
21033 (lto_read_tree_1): Stream in DIE references.
21034 (lto_input_tree): Register DIE references.
21035 (input_function): Stream DECL_DEBUG_ARGS.
21036 * lto-streamer-out.c: Include debug.h.
21037 (lto_write_tree_1): Output DIE references.
21038 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21039 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21040 (output_function): Stream DECL_DEBUG_ARGS.
21041 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21042 Stream DECL_ABSTRACT_ORIGIN.
21043 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21044 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21045 DECL_CONTEXT for file-scope decls.
21046 * lto-streamer.h (struct dref_entry): Declare.
21047 (dref_queue): Likewise.
21048 * cfgexpand.c (pass_expand::execute): Do not call the
21049 outlining_inline_function hook here.
21050 * lto-wrapper.c (debug_obj): New global.
21051 (tool_cleanup): Unlink it if required.
21052 (debug_objcopy): New function.
21053 (run_gcc): Handle early debug sections in the IL files by
21054 extracting them to separate files, partially linkin them and
21055 feeding the result back as result to the linker.
21056 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21057 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21058 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21059 sections into a separate segment.
21060 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21062 (darwin_asm_dwarf_section): Likewise.
21063 (darwin_asm_output_dwarf_offset): Likewise.
21064 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
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 * read-md.h (md_reader::record_potential_iterator_use): Replace
21071 pointer argument with an rtx and an index.
21072 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21073 (apply_mode_iterator): Likewise.
21074 (apply_code_iterator): Likewise.
21075 (apply_int_iterator): Likewise.
21076 (apply_subst_iterator): Likewise.
21077 (record_iterator_use): Likewise.
21078 (record_attribute_use): Likewise.
21079 (md_reader::record_potential_iterator_use): Likewise. Update calls
21080 to record_iterator_use and apply_iterator.
21081 (iterator_use): Replace ptr with x and index.
21082 (attribute_use): Likewise.
21083 (apply_attribute_uses): Update calls to apply_iterator.
21084 (apply_iterators): Likewise. Update initialization of iterator_use.
21085 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21086 and record_potential_iterator_use.
21087 (rtx_reader::read_rtx_operand): Likewise.
21089 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21090 Alan Hayward <alan.hayward@arm.com>
21091 David Sherwood <david.sherwood@arm.com>
21093 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21096 2017-08-21 Richard Biener <rguenther@suse.de>
21098 PR middle-end/81884
21099 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21100 at struct end conservatively when comparing common bases.
21102 2017-08-21 Richard Biener <rguenther@suse.de>
21104 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21105 (mem_ref_in_stmt): Remove.
21106 (determine_max_movement): Use ref index to get at the reference.
21107 (invariantness_dom_walker::before_dom_children): Deal with
21108 lim data already initialized.
21109 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21111 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21113 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21114 (TARGET_ISA_ROUND): Ditto.
21115 (TARGET_ROUND): Ditto.
21116 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21117 * config/i386/i386.md: Ditto.
21118 * config/i386/sse.md: Ditto.
21119 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21120 with OPTION_MASK_ISA_SSE4_1.
21122 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21125 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21126 __builtin_ia32_lzcnt_u16.
21128 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21131 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21132 (rs6000_set_current_function): Rewrite function to use it.
21134 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21137 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21138 and TYPE_WARN_IF_NOT_ALIGN.
21139 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21140 (handle_warn_if_not_align): New.
21141 (place_union_field): Call handle_warn_if_not_align.
21142 (place_field): Call handle_warn_if_not_align.
21143 Copy TYPE_WARN_IF_NOT_ALIGN.
21144 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21145 (layout_type): Likewise.
21146 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21148 (tree_decl_common): Add warn_if_not_align.
21149 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21150 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21151 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21152 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21153 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21154 * doc/extend.texi: Document warn_if_not_aligned attribute.
21155 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21157 2017-08-17 Martin Liska <mliska@suse.cz>
21160 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21161 (get_data_dependence): Use it as pointer type.
21162 (distribute_loop): Likewise.
21164 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21166 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21167 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21168 (p8_vmrgow_<mode>_direct): New define_insn.
21169 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21170 handle endianness for vmrgew and vmrgow permute patterns.
21172 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21174 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21175 * config/rs6000/rs6000-cpus.def: Remove comment.
21176 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21177 (POWERPC_MASKS): Likewise.
21178 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21179 use of TARGET_VSX_TIMODE.
21180 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21181 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21182 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21183 (rs6000_option_override_internal): Remove dead code.
21184 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21185 (rs6000_legitimize_reload_address): Likewise.
21186 (rs6000_legitimate_address_p): Likewise.
21187 (rs6000_opt_masks): Delete "vsx-timode".
21188 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21189 from function comment.
21190 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21191 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21192 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21194 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21195 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21196 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21197 (VSX_TI): Likewise.
21199 (define_peephole2): Likewise.
21201 2017-08-17 Martin Sebor <msebor@redhat.com>
21204 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21205 past the end of an array.
21206 (test_pp_format): Add test cases.
21208 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21210 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21211 missing ECF_NOTHROW flags.
21213 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21216 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21217 operands residing in integer registers.
21218 (*vsx_le_perm_load_<mode>): Likewise.
21219 (*vsx_le_perm_store_<mode>): Likewise.
21220 (define_peephole2): Add peepholes to optimize the above.
21222 2017-08-17 Marek Polacek <polacek@redhat.com>
21224 PR middle-end/81814
21225 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21226 to mimic what shorten_compare did. Change the return type to bool.
21227 (fold_cond_expr_with_comparison): Update call to
21228 operand_equal_for_comparison_p.
21229 (fold_ternary_loc): Likewise.
21231 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21233 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21235 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21237 2017-08-17 Richard Biener <rguenther@suse.de>
21239 * tree-ssa-structalias.c (solve_graph): When propagating
21240 to successors update the graphs succ edges and avoid duplicate work.
21242 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21245 * config/i386/i386.c (ix86_option_override_internal): Save target
21246 specific options after ix86_stack_protector_guard_reg was changed.
21248 2017-08-17 Richard Biener <rguenther@suse.de>
21250 PR tree-optimization/81827
21251 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21252 (new_var_info): Initialize it conservatively.
21253 (get_call_vi): Mark register vars.
21254 (new_scalar_tmp_constraint_exp): Likewise.
21255 (handle_rhs_call): Likewise.
21256 (handle_const_call): Likewise.
21257 (create_function_info_for): Likewise.
21258 (solve_constraints): Sort varinfos to separate register from
21259 non-register vars to pack points-to solution bitmaps during
21262 2017-08-17 Marek Polacek <polacek@redhat.com>
21264 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21266 2017-08-17 Richard Biener <rguenther@suse.de>
21268 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21269 to true when overflow is undefined and we saturated the result.
21271 2017-08-17 Alan Modra <amodra@gmail.com>
21274 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21275 Don't use store multiple if only one reg needs saving.
21276 (interesting_frame_related_regno): New function.
21277 (rs6000_frame_related): Don't emit frame info for regs that
21279 (rs6000_emit_epilogue): Likewise.
21281 2017-08-16 Nathan Sidwell <nathan@acm.org>
21283 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21284 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21285 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21286 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21287 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21288 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21289 (verify_type): Adjust for TYPE_BINFO move.
21290 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21291 process TYPE_BINFO directly.
21292 (hash_tree): Likewise.
21293 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21295 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21298 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21300 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21302 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21305 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21306 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21307 Add ix86_binary_operator_ok to insn constraint.
21308 (*iordi_1_bts): Ditto.
21309 (*xordi_1_btc): Ditto.
21310 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21311 Update corresponding peephole2 pattern.
21315 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21317 PR tree-optimization/81832
21318 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21319 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21321 2017-08-16 Marek Polacek <polacek@redhat.com>
21324 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21325 regarding size_zero_node.
21327 2017-08-16 Martin Liska <mliska@suse.cz>
21330 * config.gcc: Respect previously set extra_objs in case
21333 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21335 PR tree-optimization/81835
21336 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21337 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21338 not depend on the phi.
21340 2017-08-16 Alan Modra <amodra@gmail.com>
21342 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21345 2017-08-16 Alan Modra <amodra@gmail.com>
21347 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21348 (save_reg_p): ..into this. Update all callers.
21349 (first_reg_to_save): Simplify.
21351 2017-08-16 Alan Modra <amodra@gmail.com>
21353 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21356 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21360 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21361 constant count if that count is less than 32.
21363 2017-08-15 Nathan Sidwell <nathan@acm.org>
21365 * gcc.c (execute): Emit friendlier message if inferior is killed
21366 by an external cause.
21368 2017-08-15 Richard Biener <rguenther@suse.de>
21370 PR tree-optimization/81790
21371 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21372 CONSTRUCTORs from simplifying and VN.
21374 2017-08-14 Martin Sebor <msebor@redhat.com>
21376 * builtin-attrs.def: Add comments.
21378 2017-08-14 Martin Sebor <msebor@redhat.com>
21381 * doc/extend.texi (attribute nonstring): Document new attribute.
21383 2017-08-14 Martin Sebor <msebor@redhat.com>
21386 * tree-diagnostic.c (default_tree_printer): Handle %G.
21387 * gimple-pretty-print.h (percent_G_format): Declare new function.
21388 * gimple-pretty-print.c (percent_G_format): Define.
21389 * tree-pretty-print.c (percent_K_format): Add argument.
21391 2017-08-14 Martin Sebor <msebor@redhat.com>
21393 PR translation/79998
21394 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21395 Remove a stray space.
21397 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21400 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21401 (*iordi_1_bts): Ditto.
21402 (*xordi_1_btc): Ditto.
21404 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21407 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21410 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21411 quoted strings, and make more translator-friendly.
21412 (darwin_rs6000_override_options): Likewise.
21413 (rs6000_option_override_internal): Likewise.
21414 (rs6000_return_in_memory): Fix overlong line.
21415 (init_cmulative_args): Use quoted strings, and make more
21416 translator-friendly.
21417 (rs6000_pass_by_reference): Fix overlong line.
21418 (def_builtin): Use quoted strings.
21419 (altivec_expand_predicate_builtin): Use quoted strings, and make
21420 more translator-friendly.
21421 (htm_expand_builtin): Use quoted strings.
21422 (cpu_expand_builtin): Use quoted strings, and make more
21423 translator-friendly.
21424 (altivec_expand_builtin): Likewise.
21425 (paired_expand_predicate_builtin): Likewise.
21426 (rs6000_invalid_builtin): Likewise.
21427 (builtin_function_type): Use quoted strings.
21428 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21429 more translator-friendly.
21430 (rs6000_trampoline_init): Likewise.
21431 (rs6000_handle_altivec_attribute): Likewise.
21432 (rs6000_inner_target_options): Use quoted strings.
21433 (rs6000_disable_incompatible_switches): Likewise.
21434 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21435 strings, and make more translator-friendly.
21436 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21438 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21440 PR tree-optimization/81799
21441 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21442 cond_expr to simple gimple operand.
21444 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21446 PR middle-end/46932
21447 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21449 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21453 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21454 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21455 TARGET_GASISR_PROLOGUES.
21456 * config/avr/avr.c (avr_option_override): Same.
21457 (avr_pass_pre_proep::execute): Same.
21459 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21462 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21463 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21465 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21467 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21468 define_insn_and_split. Split to a memory load from 0 in
21469 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21470 using PTR mode iterator.
21471 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21472 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21473 (*add_tp_<mode>): Redefine as define_insn_and_split.
21474 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21475 address space. Merge with *add_tp_x32 using PTR mode iterator.
21476 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21477 Split to an add with a memory load from 0 in
21478 DEFAULT_TLS_SEG_REG address space.
21480 2017-08-12 Andrew Pinski <apinski@cavium.com>
21482 * config/aarch64/aarch64-option-extensions.def (rdma):
21483 Fix feature string to what Linux prints out in /proc/cpuinfo.
21485 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21488 * dwarf2out.c (modified_type_die): For C typedef types that have
21489 an ultimate origin, process the ultimate origin instead of the
21491 (gen_typedef_die): Assert that input DECLs have no ultimate
21493 (gen_type_die_with_usage): For typedef variants that have an
21494 ultimate origin, just call gen_decl_die on the original DECL.
21495 (process_scope_var): Avoid creating DIEs for local typedefs and
21496 concrete static variables.
21498 2017-08-12 Alan Modra <amodra@gmail.com>
21502 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21503 match gnu-user.h startfile.
21504 (ENDFILE_LINUX_SPEC): Similarly.
21506 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21509 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21511 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21513 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21514 * config/aarch64/aarch64.md (mov<mode>): Change.
21515 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21516 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21517 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21519 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21521 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21522 for storage order barriers.
21524 2017-08-11 Martin Liska <mliska@suse.cz>
21527 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21528 variables of void type.
21530 2017-08-11 Martin Liska <mliska@suse.cz>
21532 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21533 TARGET_SUPPORTS_ALIASES.
21534 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21535 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21536 (optimize_weakref): Likewise.
21537 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21538 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21539 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21541 2017-08-11 Martin Liska <mliska@suse.cz>
21544 * config/i386/i386.c (make_resolver_func): Do complete
21545 refactoring of the function.
21547 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21550 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21551 * config/i386/i386.c (ix86_stack_protect_guard): Use
21552 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21553 * doc/invoke.texi (x86 Options): Document
21554 -mstack-protector-guard-symbol= option.
21556 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21558 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21559 * config/i386/i386.c (ix86_split_stack_guard): New function.
21560 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21561 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21562 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21563 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21564 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21565 (split_stack_space_check): Call ix86_split_stack_guard.
21567 2017-08-10 Martin Sebor <msebor@redhat.com>
21569 * print-tree.c (print_node): Print location using the established
21571 Replace spaces with colons.
21572 (debug_raw, debug): Ditto.
21574 2017-08-10 Martin Sebor <msebor@redhat.com>
21577 * pretty-print.c (pp_format): Correct the handling of %s precision.
21579 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21582 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21584 (ix86_finalize_stack_frame_flags): This. Also clear
21585 frame_pointer_needed if -fno-omit-frame-pointer is used without
21587 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21588 with ix86_finalize_stack_frame_flags.
21589 (ix86_expand_epilogue): Likewise.
21590 (ix86_expand_split_stack_prologue): Likewise.
21591 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21593 2017-08-10 Martin Liska <mliska@suse.cz>
21596 * c-attribs.c (handle_target_attribute):
21597 Report warning for an empty string argument of target attribute.
21599 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21602 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21604 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21605 or DECL_NONLOCAL labels.
21606 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21607 or DECL_NONLOCAL labels here.
21609 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21611 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21612 to indicate when early gimple folding has been disabled.
21613 (rs6000_gimple_fold_builtin): Add debug content.
21614 (rs6000_invalid_builtin): Fix whitespace.
21615 (rs6000_expand_builtin): Fix whitespace.
21616 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21618 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21621 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21622 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21624 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21626 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21627 (qdf24xx): Likewise.
21628 * config/aarch64/aarch64.md: Include falkor.md.
21629 * config/aarch64/falkor.md: New.
21631 2017-08-09 Marek Polacek <polacek@redhat.com>
21634 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21635 * diagnostic.c (emit_diagnostic): Add a comment.
21636 (emit_diagnostic_valist): New function.
21638 2017-08-09 Marek Polacek <polacek@redhat.com>
21641 * input.c (make_location): New overload.
21642 * input.h (make_location): Declare.
21644 2017-08-08 Alan Modra <amodra@gmail.com>
21645 H.J. Lu <hongjiu.lu@intel.com>
21648 * gcc.c (NO_PIE_SPEC): Delete.
21649 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21651 (LINK_PIE_SPEC): ..to here.
21652 (LINK_COMMAND_SPEC): Support -no-pie.
21653 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21654 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21656 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21657 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21658 (ENDFILE_CRTEND_SPEC): Similarly.
21660 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21663 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21664 (mstack-protector-guard-offset=): Ditto.
21665 * config/i386/i386.c (ix86_option_override): Handle
21666 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21668 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21669 ix86_stack_protect_guard_offset variables.
21670 (TARGET_STACK_PROTECT_GUARD): Always define.
21671 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21672 and -mstack-protector-guard-offset= options.
21674 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21676 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21677 boundary case for the last candidate.
21679 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21681 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21682 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21684 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21686 PR middle-end/19706
21687 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21688 * config/aarch64/aarch64-builtins.c
21689 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21690 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21691 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21693 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21694 Andrew Pinski <pinskia@gmail.com>
21696 PR middle-end/19706
21697 * internal-fn.def (XORSIGN): New.
21698 * optabs.def (xorsign_optab): New.
21699 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21700 (convert_expand_mult_copysign): New.
21701 (pass_optimize_widening_mul::execute): Call
21702 convert_expand_mult_copysign.
21704 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21706 PR tree-optimization/81354
21707 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21708 Insert on edges rather than explicitly creating landing pads.
21709 (analyze_candidates_and_replace): Commit edge inserts.
21711 2017-08-08 Richard Biener <rguenther@suse.de>
21713 PR middle-end/81719
21714 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21715 (expand_simple_operations): Also look through ADDR_EXPRs with
21716 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21718 2017-08-08 Richard Biener <rguenther@suse.de>
21720 PR tree-optimization/81723
21721 * tree-vect-slp.c (struct bst_traits): New hash traits.
21722 (bst_fail): New global.
21723 (vect_build_slp_tree_2): New worker, split out from ...
21724 (vect_build_slp_tree): ... this now wrapping it with using
21725 bst_fail set to cache SLP tree build fails. Properly handle
21727 (vect_analyze_slp_instance): Allocate and free bst_fail.
21729 2017-08-08 Martin Liska <mliska@suse.cz>
21732 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21733 LABEL_DECLs that can be from a different function.
21735 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21737 PR tree-optimization/81744
21738 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21739 loop's number of iterations.
21741 2017-08-08 Martin Liska <mliska@suse.cz>
21743 * asan.c: Include header files.
21744 * attribs.c (build_decl_attribute_variant): New function moved
21746 (build_type_attribute_qual_variant): Likewise.
21747 (cmp_attrib_identifiers): Likewise.
21748 (simple_cst_list_equal): Likewise.
21749 (omp_declare_simd_clauses_equal): Likewise.
21750 (attribute_value_equal): Likewise.
21751 (comp_type_attributes): Likewise.
21752 (build_type_attribute_variant): Likewise.
21753 (lookup_ident_attribute): Likewise.
21754 (remove_attribute): Likewise.
21755 (merge_attributes): Likewise.
21756 (merge_type_attributes): Likewise.
21757 (merge_decl_attributes): Likewise.
21758 (merge_dllimport_decl_attributes): Likewise.
21759 (handle_dll_attribute): Likewise.
21760 (attribute_list_equal): Likewise.
21761 (attribute_list_contained): Likewise.
21762 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21763 (lookup_attribute_by_prefix): Likewise.
21764 * bb-reorder.c: Include header files.
21765 * builtins.c: Likewise.
21766 * calls.c: Likewise.
21767 * cfgexpand.c: Likewise.
21768 * cgraph.c: Likewise.
21769 * cgraphunit.c: Likewise.
21770 * convert.c: Likewise.
21771 * dwarf2out.c: Likewise.
21772 * final.c: Likewise.
21773 * fold-const.c: Likewise.
21774 * function.c: Likewise.
21775 * gimple-expr.c: Likewise.
21776 * gimple-fold.c: Likewise.
21777 * gimple-pretty-print.c: Likewise.
21778 * gimple.c: Likewise.
21779 * gimplify.c: Likewise.
21780 * hsa-common.c: Likewise.
21781 * hsa-gen.c: Likewise.
21782 * internal-fn.c: Likewise.
21783 * ipa-chkp.c: Likewise.
21784 * ipa-cp.c: Likewise.
21785 * ipa-devirt.c: Likewise.
21786 * ipa-fnsummary.c: Likewise.
21787 * ipa-inline.c: Likewise.
21788 * ipa-visibility.c: Likewise.
21790 * lto-cgraph.c: Likewise.
21791 * omp-expand.c: Likewise.
21792 * omp-general.c: Likewise.
21793 * omp-low.c: Likewise.
21794 * omp-offload.c: Likewise.
21795 * omp-simd-clone.c: Likewise.
21796 * opts-global.c: Likewise.
21797 * passes.c: Likewise.
21798 * predict.c: Likewise.
21799 * sancov.c: Likewise.
21800 * sanopt.c: Likewise.
21801 * symtab.c: Likewise.
21802 * toplev.c: Likewise.
21803 * trans-mem.c: Likewise.
21804 * tree-chkp.c: Likewise.
21805 * tree-eh.c: Likewise.
21806 * tree-into-ssa.c: Likewise.
21807 * tree-object-size.c: Likewise.
21808 * tree-parloops.c: Likewise.
21809 * tree-profile.c: Likewise.
21810 * tree-ssa-ccp.c: Likewise.
21811 * tree-ssa-live.c: Likewise.
21812 * tree-ssa-loop.c: Likewise.
21813 * tree-ssa-sccvn.c: Likewise.
21814 * tree-ssa-structalias.c: Likewise.
21815 * tree-ssa.c: Likewise.
21816 * tree-streamer-in.c: Likewise.
21817 * tree-vectorizer.c: Likewise.
21818 * tree-vrp.c: Likewise.
21819 * tsan.c: Likewise.
21820 * ubsan.c: Likewise.
21821 * varasm.c: Likewise.
21822 * varpool.c: Likewise.
21823 * tree.c: Remove functions moved to attribs.[ch].
21824 * tree.h: Likewise.
21825 * config/aarch64/aarch64.c: Add attrs.h header file.
21826 * config/alpha/alpha.c: Likewise.
21827 * config/arc/arc.c: Likewise.
21828 * config/arm/arm.c: Likewise.
21829 * config/avr/avr.c: Likewise.
21830 * config/bfin/bfin.c: Likewise.
21831 * config/c6x/c6x.c: Likewise.
21832 * config/cr16/cr16.c: Likewise.
21833 * config/cris/cris.c: Likewise.
21834 * config/darwin.c: Likewise.
21835 * config/epiphany/epiphany.c: Likewise.
21836 * config/fr30/fr30.c: Likewise.
21837 * config/frv/frv.c: Likewise.
21838 * config/ft32/ft32.c: Likewise.
21839 * config/h8300/h8300.c: Likewise.
21840 * config/i386/winnt.c: Likewise.
21841 * config/ia64/ia64.c: Likewise.
21842 * config/iq2000/iq2000.c: Likewise.
21843 * config/lm32/lm32.c: Likewise.
21844 * config/m32c/m32c.c: Likewise.
21845 * config/m32r/m32r.c: Likewise.
21846 * config/m68k/m68k.c: Likewise.
21847 * config/mcore/mcore.c: Likewise.
21848 * config/microblaze/microblaze.c: Likewise.
21849 * config/mips/mips.c: Likewise.
21850 * config/mmix/mmix.c: Likewise.
21851 * config/mn10300/mn10300.c: Likewise.
21852 * config/moxie/moxie.c: Likewise.
21853 * config/msp430/msp430.c: Likewise.
21854 * config/nds32/nds32-isr.c: Likewise.
21855 * config/nds32/nds32.c: Likewise.
21856 * config/nios2/nios2.c: Likewise.
21857 * config/nvptx/nvptx.c: Likewise.
21858 * config/pa/pa.c: Likewise.
21859 * config/pdp11/pdp11.c: Likewise.
21860 * config/powerpcspe/powerpcspe.c: Likewise.
21861 * config/riscv/riscv.c: Likewise.
21862 * config/rl78/rl78.c: Likewise.
21863 * config/rx/rx.c: Likewise.
21864 * config/s390/s390.c: Likewise.
21865 * config/sh/sh.c: Likewise.
21866 * config/sol2.c: Likewise.
21867 * config/sparc/sparc.c: Likewise.
21868 * config/spu/spu.c: Likewise.
21869 * config/stormy16/stormy16.c: Likewise.
21870 * config/tilegx/tilegx.c: Likewise.
21871 * config/tilepro/tilepro.c: Likewise.
21872 * config/v850/v850.c: Likewise.
21873 * config/vax/vax.c: Likewise.
21874 * config/visium/visium.c: Likewise.
21875 * config/xtensa/xtensa.c: Likewise.
21877 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21880 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21881 constraints since the -mupper-regs-* switches have been
21883 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21884 into a vector from a double word element that was extracted from
21885 another vector, and eliminate extra XXPERMDI instructions.
21886 (vsx_concat_<mode>_2): Likewise.
21887 (vsx_concat_<mode>_3): Likewise.
21888 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21889 concat to allow optimizing inserts from previous extracts.
21891 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21893 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21894 memory reference to a SSP offset in TLS address space.
21895 (ix86_print_operand) <case '@'>: Remove.
21896 (ix86_print_operand_punct_valid_p): Remove '@' code.
21897 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21898 UNSPEC_SP_TLS_TEST.
21899 (stack_tls_protect_set_<mode>): Remove.
21900 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21901 (stack_tls_protect_test_<mode>): Remove.
21902 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21904 2017-08-07 Olivier Hainque <hainque@adacore.com>
21907 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21909 2017-08-07 Douglas Rupp <rupp@adacore.com>
21911 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21912 variable was referenced as multidir in command.
21914 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21917 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21920 2017-08-07 Martin Liska <mliska@suse.cz>
21922 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21923 * config/rl78/rl78.c: Add include of attribs.h.
21924 * config/sh/sh.c: Likewise.
21925 * config/v850/v850.c: Likewise.
21927 2017-08-07 Tom de Vries <tom@codesourcery.com>
21929 PR middle-end/78266
21930 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21932 2017-08-07 Martin Liska <mliska@suse.cz>
21934 * config/mips/mips.c: Include attribs.h.
21936 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21939 * doc/invoke.texi: Document change in behvaior for -Ofast for
21942 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21944 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21946 (aarch64_pop_regs): Likewise.
21947 (aarch64_gen_load_pair): Likewise.
21948 (aarch64_save_callee_saves): Likewise.
21949 (aarch64_restore_callee_saves): Likewise.
21951 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21953 * config/i386/i386.c: Revert the last change.
21955 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21958 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21960 (ix86_finalize_stack_frame_flags): This. Also clear
21961 frame_pointer_needed if -fno-omit-frame-pointer is used without
21963 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21964 with ix86_finalize_stack_frame_flags.
21965 (ix86_expand_epilogue): Likewise.
21966 (ix86_expand_split_stack_prologue): Likewise.
21968 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21971 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21972 to P_AES for Westmere.
21974 2017-08-07 Jonathan Yong <10walls@gmail.com>
21976 * config/i386/mingw.opt (fset-stack-executable): Removed.
21977 * config/i386/cygming.opt (fset-stack-executable): Moved
21979 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21981 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21983 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21985 2017-08-07 Marek Polacek <polacek@redhat.com>
21987 PR middle-end/81737
21988 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21990 2017-08-07 Martin Liska <mliska@suse.cz>
21992 * attribs.h (canonicalize_attr_name): New function.
21993 (cmp_attribs): Move from c-format.c and adjusted.
21994 (is_attribute_p): Moved from tree.h.
21995 * tree-inline.c: Add new includes.
21996 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21997 (private_is_attribute_p): Remove.
21998 (private_lookup_attribute): Likewise.
21999 (private_lookup_attribute_by_prefix): Simplify.
22000 (remove_attribute): Use is_attribute_p.
22001 * tree.h: Remove removed declarations.
22003 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22005 PR middle-end/81698
22006 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22007 instead of computing it in the function. Formatting fix.
22008 (expand_case): Don't rely on default_edge being the first edge,
22009 clear it if removing it, pass default_edge to
22010 emit_case_dispatch_table.
22011 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22014 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22016 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22017 insn in the function, emit NOP after the insn.
22019 2017-08-06 Tom de Vries <tom@codesourcery.com>
22021 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22024 2017-08-06 Tom de Vries <tom@codesourcery.com>
22026 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22029 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22031 PR tree-optimization/57371
22032 * match.pd: New pattern.
22034 2017-08-04 Marek Polacek <polacek@redhat.com>
22036 PR middle-end/81695
22037 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22038 perform the computation in offset_int.
22040 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22042 PR tree-optimization/81136
22043 * tree-vectorizer.h: Include tree-hash-traits.h.
22044 (vec_base_alignments): New typedef.
22045 (vec_info): Add a base_alignments field.
22046 (vect_record_base_alignments): Declare.
22047 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22049 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22050 (create_data_ref): Add an is_conditional_in_stmt argument.
22051 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22052 the is_conditional_in_stmt field.
22053 (data_ref_loc): Add an is_conditional_in_stmt field.
22054 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22055 (find_data_references_in_stmt): Update call to create_data_ref.
22056 (graphite_find_data_references_in_stmt): Likewise.
22057 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22058 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22059 (vect_record_base_alignment): New function.
22060 (vect_record_base_alignments): Likewise.
22061 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22062 for nested statements even if we fail to compute a misalignment.
22063 Use pooled base alignments for unconditional references.
22064 (vect_find_same_alignment_drs): Compare base addresses instead
22066 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22067 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22069 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22071 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22072 Add an explicit name for the enum. Use auto_vec for slp_instances
22073 and grouped_stores.
22074 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22076 (_bb_vec_info): Add a constructor and destructor.
22077 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22078 (destroy_loop_vec_info): Delete.
22079 (vect_destroy_datarefs): Likewise.
22080 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22081 (vec_info::vec_info): New function.
22082 (vec_info::~vec_info): Likewise.
22083 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22084 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22085 destroy_loop_vec_info.
22086 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22087 (_loop_vec_info::_loop_vec_info): ...this.
22088 (destroy_loop_vec_info): Replace with...
22089 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22090 the stmt_vec_infos. Leave handling of vec_info information to its
22091 destructor. Remove explicit vector releases.
22092 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22093 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22094 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22095 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22096 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22097 (destroy_bb_vec_info): Replace with...
22098 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22099 information to its destructor.
22100 (vect_slp_analyze_bb_1): Use new and delete instead of
22101 new_bb_vec_info and destroy_bb_vec_info.
22102 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22105 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22107 * tree-data-ref.h (subscript): Add access_fn field.
22108 (data_dependence_relation): Add could_be_independent_p.
22109 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22110 (same_access_functions): Move to tree-data-ref.c.
22111 * tree-data-ref.c (ref_contains_union_access_p): New function.
22112 (access_fn_component_p): Likewise.
22113 (access_fn_components_comparable_p): Likewise.
22114 (dr_analyze_indices): Add a reference to access_fn_component_p.
22115 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22117 (constant_access_functions): Likewise.
22118 (add_other_self_distances): Likewise.
22119 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22120 (initialize_data_dependence_relation): Use XCNEW and remove
22121 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22122 of access functions that have the same type. Allow the
22123 subsequence to end with different bases in some circumstances.
22124 Record the chosen access functions in SUB_ACCESS_FN.
22125 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22126 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22127 (subscript_dependence_tester_1): Likewise dra and drb.
22128 (build_classic_dist_vector): Update calls accordingly.
22129 (subscript_dependence_tester): Likewise.
22130 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22131 DDR_COULD_BE_INDEPENDENT_P.
22132 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22133 comp_alias_ddrs instead of may_alias_ddrs.
22134 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22136 (vect_analyze_data_ref_dependence): Use it if
22137 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22138 distance vectors if that fails.
22139 (dependence_distance_ge_vf): New function.
22140 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22141 LOOP_VINFO_MAY_ALIAS_DDRS.
22143 2017-08-04 Richard Biener <rguenther@suse.de>
22145 PR middle-end/81705
22146 * fold-const.c (fold_binary_loc): Properly restrict
22147 minus_var0 && minus_var1 case when associating undefined overflow
22150 2017-08-04 Tom de Vries <tom@codesourcery.com>
22152 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22154 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22157 Don't start diagnostic messages with a capital letter.
22158 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22160 (rs6000_invalid_builtin): Likewise.
22161 (rs6000_trampoline_init): Likewise.
22163 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22166 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22167 after setting changeable df flags.
22169 2017-08-03 Richard Biener <rguenther@suse.de>
22171 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22172 up if the use is in USE - X.
22174 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22176 * toplev.c (dumpfile.h): New include.
22177 (internal_error_reentered): New static function. Use it...
22178 (internal_error_function): ...here to handle reentered internal_error.
22180 2017-08-03 Richard Biener <rguenther@suse.de>
22182 PR middle-end/81148
22183 * fold-const.c (split_tree): Add minus_var and minus_con
22184 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22185 here but always use minus_*.
22186 (associate_trees): Assert we never associate with MINUS_EXPR
22187 and NULL first operand. Do not recurse for PLUS_EXPR operands
22188 when associating as MINUS_EXPR either.
22189 (fold_binary_loc): Track minus_var and minus_con.
22191 2017-08-03 Tom de Vries <tom@codesourcery.com>
22194 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22195 ACCEL_COMPILER, apply finish_options on
22196 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22198 2017-08-03 Tom de Vries <tom@codesourcery.com>
22201 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22202 function_entry_patch_area_size > 0.
22204 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22207 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22208 instead of 10??LU, perform unit multiplication in wide_int,
22209 don't change alloc_object_size_limit if the limit is larger
22212 PR tree-optimization/81655
22213 PR tree-optimization/81588
22214 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22215 the case when ranges[i].low and high are 1 for unsigned type with
22218 PR middle-end/81052
22219 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22220 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22222 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22224 * tree-vrp.h: Add include guard.
22226 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22229 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22230 (ud2): New insn pattern.
22231 * config/i386/i386.c (ix86_expand_epilogue):
22232 For naked functions, generate ud2 instead of trap insn.
22234 2017-08-02 Marek Polacek <polacek@redhat.com>
22237 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22239 2017-08-02 Tom de Vries <tom@codesourcery.com>
22240 Cesar Philippidis <cesar@codesourcery.com>
22242 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22243 Add missing edge probabilities.
22245 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22247 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22248 Correct endianness.
22250 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22252 PR middle-end/79499
22253 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22254 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22255 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22256 prologue_seq sequences - if any.
22258 2017-08-02 Richard Biener <rguenther@suse.de>
22260 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22261 via vectors if supported, integer extracts via punning if supported
22262 or otherwise vector extracts.
22264 2017-08-02 Richard Biener <rguenther@suse.de>
22266 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22268 (bitmap_insert_into_set): ... this.
22270 2017-08-02 Richard Biener <rguenther@suse.de>
22272 PR tree-optimization/81633
22274 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22276 PR tree-optimization/71752
22277 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22279 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22281 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22282 (machine_function::call_ms2sysv_pad_out): Remove field.
22283 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22284 (ix86_compute_frame_layout): Likewise.
22286 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22289 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22290 attribute with interrupt attribute.
22292 2017-08-01 Andrew Pinski <apinski@cavium.com>
22294 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22295 BIT_INSERT_EXPR's operand 1
22296 to see if the types precision matches.
22298 2017-08-01 Martin Liska <mliska@suse.cz>
22300 PR middle-end/70140
22301 * builtins.c (expand_builtin_memcpy_args): Remove.
22302 (expand_builtin_memcpy): Call newly added function
22303 expand_builtin_memory_copy_args.
22304 (expand_builtin_memcpy_with_bounds): Likewise.
22305 (expand_builtin_mempcpy): Remove last argument.
22306 (expand_builtin_mempcpy_with_bounds): Likewise.
22307 (expand_builtin_memory_copy_args): New function created from
22308 expand_builtin_mempcpy_args with small modifications.
22309 (expand_builtin_mempcpy_args): Remove.
22310 (expand_builtin_stpcpy): Remove unused argument.
22311 (expand_builtin): Likewise.
22312 (expand_builtin_with_bounds): Likewise.
22314 2017-08-01 Martin Liska <mliska@suse.cz>
22317 Make mempcpy more optimal (PR middle-end/70140).
22319 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22322 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22323 __builtin_vec_cmpne verify both arguments are compatible vectors
22324 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22325 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22326 move computation of aligned to after checking the argument types.
22330 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22331 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22334 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22335 Doug Rupp <rupp@adacore.com>
22336 Olivier Hainque <hainque@adacore.com>
22338 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22339 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22341 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22342 for TARGET_OS_CPP_BUILTIN.
22343 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22344 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22346 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22347 passing required abi options to the assembler for EABI configurations.
22348 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22349 of .text.hot and .text.unlikely sections for kernel modules when
22350 using ARM style exceptions.
22351 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22352 options. Add EXTRA_CC1_SPEC.
22353 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22355 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22357 (ARM_TARGET2_DWARF_FORMAT): Define.
22358 * config/arm/t-vxworks: Adjust multilib control to removal of the
22359 Diab command line options.
22361 2017-08-01 Martin Liska <mliska@suse.cz>
22363 PR gcov-profile/81561
22364 * gcov.c (unblock): Make unblocking safe as we need to preserve
22365 index correspondence of blocks and block_lists.
22367 2017-08-01 Richard Biener <rguenther@suse.de>
22369 PR tree-optimization/81181
22370 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22371 (compute_antic): ... end of iteration here.
22373 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22375 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22376 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22377 (ftree-slp-vectorize): Likewise.
22378 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22379 can no longer be set independent of flag_tree_loop_vectorize.
22380 * omp-general.c (emp_max_vf): Likewise.
22381 * opts.c (enable_fdo_optimizations): Remove references to
22382 flag_tree_vectorize, these are now implicit.
22383 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22384 and leave it for the options machinery.
22386 2017-08-01 Martin Liska <mliska@suse.cz>
22388 PR middle-end/70140
22389 * builtins.c (expand_builtin_memcpy_args): Remove.
22390 (expand_builtin_memcpy): Call newly added function
22391 expand_builtin_memory_copy_args.
22392 (expand_builtin_memcpy_with_bounds): Likewise.
22393 (expand_builtin_mempcpy): Remove last argument.
22394 (expand_builtin_mempcpy_with_bounds): Likewise.
22395 (expand_builtin_memory_copy_args): New function created from
22396 expand_builtin_mempcpy_args with small modifications.
22397 (expand_builtin_mempcpy_args): Remove.
22398 (expand_builtin_stpcpy): Remove unused argument.
22399 (expand_builtin): Likewise.
22400 (expand_builtin_with_bounds): Likewise.
22402 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22405 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22406 print "ds:" only for immediates in generic address space.
22408 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22411 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22412 (ix86_function_ok_for_sibcall): Return false for naked functions.
22414 2017-08-01 Richard Biener <rguenther@suse.de>
22416 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22417 (compute_antic): Seed worklist with exit block predecessors.
22418 * cfganal.c (dfs_find_deadend): For a cycle return the source
22419 of the edge closing it.
22421 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22423 * config/aarch64/aarch64.c
22424 (aarch64_can_const_movi_rtx_p): Move 0 check.
22426 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22428 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22429 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22431 * match.pd: Ditto in address comparison pattern.
22433 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22435 PR tree-optimization/81627
22436 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22437 closed ssa form for store-store chain.
22439 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22441 PR tree-optimization/81620
22442 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22443 for store-store chain.
22445 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22447 PR tree-optimization/81588
22448 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22449 ranges[i].in_p, invert comparison code ccode. For >/>=,
22450 swap rhs1 and rhs2 and comparison code unconditionally,
22451 for </<= don't do that. Don't swap rhs1/rhs2 again if
22452 ranges[i].in_p, instead invert comparison code ccode if
22453 opcode or oe->rank is BIT_IOR_EXPR.
22456 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22457 a direct optab to conversion optab.
22458 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22459 with GET_MODE_INNER as last argument instead of optab_handler.
22460 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22461 vector extraction if possible and optab is available.
22462 * expr.c (store_constructor): Use convert_optab_handler instead
22463 of optab_handler. Use vector initialization from smaller
22464 vectors if possible and optab is available.
22465 * tree-vect-stmts.c (vectorizable_load): Likewise.
22466 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22467 now have two modes.
22468 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22469 of vec_init from half-sized vectors with the same element mode.
22470 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22471 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22472 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22473 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22474 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22475 after mode in gen_vec_extract* calls.
22476 (vec_extract<mode>): Renamed to ...
22477 (vec_extract<mode><ssescalarmodelower>): ... this.
22478 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22479 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22480 element mode after mode in gen_vec_init* calls.
22481 (VEC_INIT_HALF_MODE): New mode iterator.
22482 (vec_init<mode>): Renamed to ...
22483 (vec_init<mode><ssescalarmodelower>): ... this.
22484 (vec_init<mode><ssehalfvecmodelower>): New expander.
22485 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22486 (vec_extractv2sfsf): ... this.
22487 (vec_initv2sf): Renamed to ...
22488 (vec_initv2sfsf): ... this.
22489 (vec_extractv2si): Renamed to ...
22490 (vec_extractv2sisi): ... this.
22491 (vec_initv2si): Renamed to ...
22492 (vec_initv2sisi): ... this.
22493 (vec_extractv4hi): Renamed to ...
22494 (vec_extractv4hihi): ... this.
22495 (vec_initv4hi): Renamed to ...
22496 (vec_initv4hihi): ... this.
22497 (vec_extractv8qi): Renamed to ...
22498 (vec_extractv8qiqi): ... this.
22499 (vec_initv8qi): Renamed to ...
22500 (vec_initv8qiqi): ... this.
22501 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22502 (vec_init<mode>): Renamed to ...
22503 (vec_init<mode><VEC_base_l>): ... this.
22504 (vec_extract<mode>): Renamed to ...
22505 (vec_extract<mode><VEC_base_l>): ... this.
22506 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22507 (vec_initv2sfsf): ... this.
22508 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22509 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22510 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22511 element mode after mode in gen_vec_init* calls.
22512 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22513 (vec_init<mode><Vel>): ... this.
22514 (vec_extract<mode>): Renamed to ...
22515 (vec_extract<mode><Vel>): ... this.
22516 * config/aarch64/iterators.md (Vel): New mode attribute.
22517 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22518 Add element mode after mode in gen_vec_extract* calls.
22519 * config/s390/vector.md (non_vec_l): New mode attribute.
22520 (vec_extract<mode>): Renamed to ...
22521 (vec_extract<mode><non_vec_l>): ... this.
22522 (vec_init<mode>): Renamed to ...
22523 (vec_init<mode><non_vec_l>): ... this.
22524 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22525 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22527 * config/arm/iterators.md (V_elem_l): New mode attribute.
22528 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22529 (vec_extract<mode><V_elem_l>): ... this.
22530 (vec_extractv2di): Renamed to ...
22531 (vec_extractv2didi): ... this.
22532 (vec_init<mode>): Renamed to ...
22533 (vec_init<mode><V_elem_l>): ... this.
22534 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22535 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22536 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22537 Add element mode after gen_vec_extract* calls.
22538 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22539 (vec_init<mode><unitmode>): ... this.
22540 (vec_extract<mode>): Renamed to ...
22541 (vec_extract<mode><unitmode>): ... this.
22542 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22543 (vec_init<mode><unitmode>): ... this.
22544 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22545 (vec_initv2sfsf): ... this.
22546 (vec_extractv2sf): Renamed to ...
22547 (vec_extractv2sfsf): ... this.
22548 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22549 Add element mode after gen_vec_extract* calls.
22550 * config/mips/mips.md (unitmode): New mode iterator.
22551 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22552 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22553 * config/spu/spu.md (inner_l): New mode attribute.
22554 (vec_init<mode>): Renamed to ...
22555 (vec_init<mode><inner_l>): ... this.
22556 (vec_extract<mode>): Renamed to ...
22557 (vec_extract<mode><inner_l>): ... this.
22558 * config/sparc/sparc.md (veltmode): New mode iterator.
22559 (vec_init<VMALL:mode>): Renamed to ...
22560 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22561 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22562 (vec_initv2sisi): ... this.
22563 (vec_initv2sf): Renamed to ...
22564 (vec_initv2sfsf): ... this.
22565 (vec_extractv2sf): Renamed to ...
22566 (vec_extractv2sfsf): ... this.
22567 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22568 (vec_init<mode>): Renamed to ...
22569 (vec_init<mode><VEC_base_l>): ... this.
22570 (vec_extract<mode>): Renamed to ...
22571 (vec_extract<mode><VEC_base_l>): ... this.
22572 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22573 (vec_initv2sfsf): ... this.
22574 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22575 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22576 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22577 gen_vec_init* calls.
22579 2017-08-01 Richard Biener <rguenther@suse.de>
22581 PR tree-optimization/81297
22582 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22583 TREE_OVERFLOW from INTEGER_CSTs.
22585 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22587 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22589 2017-07-31 Carl Love <cel@us.ibm.com>
22591 * config/rs6000/rs6000-c: Add support for built-in functions
22592 vector signed char vec_xl_be (signed long long, signed char *);
22593 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22594 vector signed int vec_xl_be (signed long long, signed int *);
22595 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22596 vector signed long long vec_xl_be (signed long long, signed long long *);
22597 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22598 vector signed short vec_xl_be (signed long long, signed short *);
22599 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22600 vector double vec_xl_be (signed long long, double *);
22601 vector float vec_xl_be (signed long long, float *);
22602 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22603 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22604 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22606 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22607 (altivec_expand_builtin): Add switch statement to call
22608 altivec_expand_xl_be for each builtin.
22609 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22610 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22611 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22612 __builtin_vsx_le_be_v16qi.
22613 * doc/extend.texi: Update the built-in documentation file for the
22614 new built-in functions.
22616 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22619 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22621 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22623 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22625 * config.gcc: Add z14.
22626 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22627 CPU model numbers for z13s and z14.
22628 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22630 * config/s390/s390-opts.h (enum processor_type): Rename
22631 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22632 * config/s390/s390.c (processor_table): Add field for CPU name to
22633 be passed to Binutils.
22634 (s390_asm_output_machine_for_arch): Use the new field in
22635 processor_table for Binutils.
22636 (s390_expand_builtin): Replace arch12 with z14.
22637 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22638 (s390_get_sched_attrmask): Likewise.
22639 (s390_get_unit_mask): Likewise.
22640 * config/s390/s390.opt: Add z14 to processor_type enum.
22642 2017-07-31 Martin Jambor <mjambor@suse.cz>
22645 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22646 regardless of optimization level.
22648 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22649 Martin Liska <mliska@suse.cz>
22651 * predict.def: Remove old comment and adjust probability.
22652 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22653 PREDICT statements.
22655 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22658 * config/i386/i386.c (ix86_function_naked): New function.
22659 (ix86_can_use_return_insn_p): Return false for naked functions.
22660 (ix86_expand_prologue): Skip prologue for naked functions.
22661 (ix86_expand_epilogue): Skip epilogue for naked functions
22662 and emit trap instruction.
22663 (ix86_warn_func_return): New function.
22664 (ix86_attribute_table): Add "naked" attribute specification.
22665 (TARGET_WARN_FUNC_RETURN): Define.
22666 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22668 2017-07-31 Martin Liska <mliska@suse.cz>
22670 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22671 (dump_gimple_bb_header): Always dump BB info.
22672 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22674 2017-07-31 Martin Liska <mliska@suse.cz>
22677 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22678 also with current_function_decl non-null equality.
22680 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22683 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22684 change type to the element type, instead add eltype variable and
22685 use it where we are interested in the element type.
22687 PR tree-optimization/81603
22688 * ipa-polymorphic-call.c
22689 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22690 offset arithmetic in offset_int, bail out if the resulting bit offset
22691 doesn't fit into shwi.
22693 2017-07-31 Martin Liska <mliska@suse.cz>
22695 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22696 (gimplify_save_expr): Fix comment.
22698 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22701 * config/i386/i386.c (ix86_function_arg): Update arguments for
22703 (ix86_compute_frame_layout): Set the initial stack offset to
22704 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22705 INCOMING_FRAME_SP_OFFSET.
22706 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22707 stack before exception handler returns.
22708 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22709 the 'ERROR_CODE' for exception handler.
22711 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22713 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22714 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22715 (ASM_OUTPUT_REG_POP): Ditto.
22716 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22717 instead of asm_fprintf to output pure string.
22719 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22721 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22722 to imported_module_or_decl hook.
22723 (debug_nothing_tree_tree_tree_bool): Remove.
22724 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22725 * debug.c (do_nothing_debug_hooks): Use
22726 debug_nothing_tree_tree_tree_bool_bool instead of
22727 debug_nothing_tree_tree_tree_bool.
22728 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22729 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22730 * sdbout.c (sdb_debug_hooks): Likewise.
22731 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22732 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22734 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22735 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22736 attribute, don't add anything.
22738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22740 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22741 (fold_build2_stat_loc): Likewise.
22742 (fold_build3_stat_loc): Likewise.
22743 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22744 (fold_build1_loc): Remove macro.
22745 (fold_build2_loc): Likewise.
22746 (fold_build3_loc): Likewise.
22748 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22750 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22751 (gimple_build_debug_bind_source_stat): Likewise.
22752 * gimple.h (gimple_build_debug_bind): Remove macro.
22753 (gimple_build_debug_bind_source): Likewise.
22755 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22757 * bitmap.c (bitmap_alloc): Adjust.
22758 (bitmap_gc_alloc): Likewise.
22759 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22761 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22763 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22764 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22765 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22766 (bitmap_gc_alloc_stat): Likewise.
22767 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22769 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22771 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22772 * rtl.h (shallow_copy_rtx): Remove macro.
22774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22776 * emit-rtl.c (gen_raw_REG): Adjust.
22777 * gengenrtl.c (gendef): Likewise.
22778 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22779 * rtl.h (rtx_alloc): Remove macro.
22781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22783 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22784 (build_tree_list_stat): Likewise.
22785 * tree.h (build_tree_list): Remove macro.
22786 (build_tree_list_vec): Likewise.
22788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22790 * tree.c (make_vector_stat): Remove _stat from name.
22791 (build_vector_stat): Likewise.
22792 * tree.h (make_vector_stat): Remove macro.
22793 (build_vector_stat): Likewise.
22795 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22797 * tree.h (build_var_debug_value): Remove prototype.
22799 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22801 * tree.c (tree_cons_stat): Remove _stat from name.
22802 * tree.h (tree_cons): Remove macro.
22804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22806 * tree.c (build_vl_exp_stat): Remove _stat from name.
22807 * tree.h (build_vl_exp): Remove macro.
22809 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22811 * tree.c (build_decl_stat): Remove _stat from name.
22812 * tree.h (build_decl): Remove macro.
22814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22816 * gimple.c (gimple_build_with_ops_stat): Adjust.
22817 (gimple_alloc_stat): Remove _stat from name.
22818 * gimple.h (gimple_alloc): Remove macro.
22820 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22822 * tree.c (make_tree_vec_stat): Remove _stat from name.
22823 (grow_tree_vec_stat): Likewise.
22824 * tree.h (make_tree_vec_stat): Adjust prototype.
22825 (grow_tree_vec_stat): Likewise.
22826 (make_tree_vec): Remove macro.
22827 (grow_tree_vec): Likewise.
22829 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22831 * fold-const.c (fold_build1_stat_loc): Adjust.
22832 (fold_build2_stat_loc): Likewise.
22833 (fold_build3_stat_loc): Likewise.
22834 * tree.c (build0_stat): Remove _stat from name.
22835 (build1_stat): Likewise.
22836 (build2_stat): Likewise.
22837 (build3_stat): Likewise.
22838 (build4_stat): Likewise.
22839 (build5_stat): Likewise.
22840 * tree.h (build1_loc): Remove macro, and rename _stat function
22842 (build2_loc): Likewise.
22843 (build3_loc): Likewise.
22844 (build4_loc): Likewise.
22845 (build5_loc): Likewise.
22847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22849 * tree.c (make_int_cst_stat): Remove _stat from name.
22850 * tree.h (make_int_cst_stat): Adjust prototype.
22851 (make_int_cst): Remove macro.
22853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22855 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22856 * tree.h (make_tree_binfo_stat): Adjust prototype.
22857 (make_tree_binfo): Remove.
22859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22861 * tree.c (copy_node_stat): Rename to copy_node.
22862 (build_distinct_type_copy): Adjust.
22863 * tree.h (copy_node_stat): Adjust prototype.
22864 (copy_node): Remove macro.
22866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22868 * tree.c (make_node_stat): rename to make_node.
22869 (build_tree_list_stat): Adjust.
22870 (build0_stat): Likewise.
22871 (build2_stat): Likewise.
22872 (build3_stat): Likewise.
22873 (build4_stat): Likewise.
22874 (build5_stat): Likewise.
22875 (build_decl_stat): Likewise.
22876 * tree.h (make_node_stat): Adjust prototype.
22877 (make_node): remove macro.
22879 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22881 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22882 (PPC_FEATURE2_SCV): Likewise.
22883 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22885 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22887 * config/aarch64/aarch64.c
22888 (aarch64_internal_mov_immediate): Add new special pattern.
22889 * config/aarch64/aarch64.md (*movdi_aarch64):
22890 Add reg/32bit const mov case.
22892 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22893 Richard Sandiford <richard.sandiford@linaro.org>
22895 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22896 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22897 Add integer and movi cases.
22898 (movi-split-hf-df-sf split, fp16): New.
22899 (enabled): Added TARGET_FP_F16INST.
22900 * config/aarch64/iterators.md (GPF_HF): New.
22901 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22903 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22905 * config/aarch64/aarch64.c
22906 (aarch64_simd_container_mode): Add prototype.
22907 (aarch64_expand_mov_immediate): Add HI support.
22908 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22909 (aarch64_can_const_movi_rtx_p): New.
22910 (aarch64_preferred_reload_class):
22911 Remove restrictions of using FP registers for certain SIMD operations.
22912 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22913 (aarch64_valid_floating_const): Add integer move validation.
22914 (aarch64_simd_imm_scalar_p): Remove.
22915 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22916 (aarch64_legitimate_constant_p): Expand list of supported cases.
22917 * config/aarch64/aarch64-protos.h
22918 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22919 (aarch64_reinterpret_float_as_int): New.
22920 (aarch64_simd_imm_scalar_p): Remove.
22921 * config/aarch64/constraints.md (Uvi): New.
22922 (Dd): Split into Ds and new Dd.
22923 * config/aarch64/aarch64.md (*movsi_aarch64):
22925 (*movdi_aarch64): Add SIMD mov case.
22927 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22929 * tree-predcom.c: (struct chain): Handle store-store chain in which
22930 stores for elimination only store loop invariant values.
22931 (execute_pred_commoning_chain): Ditto.
22932 (prepare_initializers_chain_store_elim): Ditto.
22933 (prepare_finalizers): Ditto.
22934 (is_inv_store_elimination_chain): New function.
22935 (initialize_root_vars_store_elim_1): New function.
22937 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22939 * tree-predcom.c: Revise general description of the pass.
22940 (enum chain_type): New enum type for store elimination.
22941 (struct chain): New field supporting store elimination.
22942 (struct component): Ditto.
22943 (dump_chain): Dump store-stores chain.
22944 (release_chain): Release resources.
22945 (split_data_refs_to_components): Compute and create component
22946 contains only stores for elimination.
22947 (get_chain_last_ref_at): New function.
22948 (make_invariant_chain): Initialization.
22949 (make_rooted_chain): Specify chain type in parameter and record it.
22950 (add_looparound_copies): Skip for store-stores chain.
22951 (determine_roots_comp): Compute type of chain and pass it to
22953 (initialize_root_vars_store_elim_2): New function.
22954 (finalize_eliminated_stores): New function.
22955 (remove_stmt): Handle store for elimination.
22956 (execute_pred_commoning_chain): Execute predictive commoning on
22957 store-store chains.
22958 (determine_unroll_factor): Skip unroll for store-stores chain.
22959 (prepare_initializers_chain_store_elim): New function.
22960 (prepare_initializers_chain): Hanlde store-store chain.
22961 (prepare_finalizers_chain, prepare_finalizers): New function.
22962 (tree_predictive_commoning_loop): Return integer value indicating
22963 if loop is unrolled or lcssa form is corrupted.
22964 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22966 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22968 * tree-predcom.c (initialize_root): Delete.
22969 (execute_pred_commoning_chain): Initialize root vars and replace
22970 reference of non-combined chain directly, rather than call above
22973 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22975 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22976 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22978 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22980 * tree-predcom.c (struct chain): New field init_seq.
22981 (release_chain): Release init_seq.
22982 (prepare_initializers_chain): Record intialization stmts in above
22984 (insert_init_seqs): New function.
22985 (tree_predictive_commoning_loop): Call insert_init_seqs.
22987 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22989 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22991 2017-07-28 Richard Biener <rguenther@suse.de>
22993 * match.pd: Remove superfluous :c.
22994 * genmatch.c (simplify::id): Add member.
22995 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22997 (current_id): New global.
22998 (dt_node::parent): Move from ...
22999 (dt_operand::parent): ... here. Add for_id member.
23000 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23001 (decision_tree::find_node): Relax order requirement when
23002 merging DT_TRUE nodes to ones inbetween the current simplify
23003 and the one we try to merge with. Add diagnostic whenever
23004 we need to enforce pattern order by not merging.
23005 (decision_tree::insert): Set current_id.
23006 (decision_tree::print_node): Dump parent node and for_id.
23007 (parser::last_id): Add member.
23008 (parser::push_simplify): Assign unique id.
23009 (parser::parser): Initialize last_id.
23011 2017-07-28 Martin Liska <mliska@suse.cz>
23014 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23015 gimple_build_debug_bind.
23017 2017-07-28 Richard Biener <rguenther@suse.de>
23019 PR tree-optimization/81502
23020 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23022 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23024 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23025 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23026 for BIT_FIELD_REF args.
23027 * fold-const.c (make_bit_field_ref): Likewise.
23028 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23030 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23033 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23034 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23035 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23036 Or it into SANITIZER_UNDEFINED.
23037 * ubsan.c: Include gimple-fold.h and varasm.h.
23038 (ubsan_expand_ptr_ifn): New function.
23039 (instrument_pointer_overflow): New function.
23040 (maybe_instrument_pointer_overflow): New function.
23041 (instrument_object_size): Formatting fix.
23042 (pass_ubsan::execute): Call instrument_pointer_overflow
23043 and maybe_instrument_pointer_overflow.
23044 * internal-fn.c (expand_UBSAN_PTR): New function.
23045 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23046 * sanitizer.def (__ubsan_handle_pointer_overflow,
23047 __ubsan_handle_pointer_overflow_abort): New builtins.
23048 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23049 * internal-fn.def (UBSAN_PTR): New internal function.
23050 * opts.c (sanitizer_opts): Add pointer-overflow.
23051 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23052 * fold-const.c (build_range_check): Compute pointer range check in
23053 integral type if pointer arithmetics would be needed. Formatting
23056 2017-07-28 Martin Liska <mliska@suse.cz>
23059 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23060 parameters that are of a variable-length.
23062 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23064 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23066 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23067 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23068 (CRT_CALL_STATIC_FUNCTION): Likewise.
23069 (ASM_DEFAULT_SPEC): New define.
23070 (ASM_SPEC32): Likewise.
23071 (ASM_SPEC64): Likewise.
23072 (ASM_SPEC_COMMON): Likewise.
23073 (ASM_SPEC): Likewise.
23074 (INVALID_64BIT): Likewise.
23075 (LINK_OS_DEFAULT_SPEC): Likewise.
23076 (LINK_OS_SPEC32): Likewise.
23077 (LINK_OS_SPEC64): Likewise.
23078 (POWERPC_LINUX): Likewise.
23079 (PTRDIFF_TYPE): Likewise.
23080 (RESTORE_FP_PREFIX): Likewise.
23081 (RESTORE_FP_SUFFIX): Likewise.
23082 (SAVE_FP_PREFIX): Likewise.
23083 (SAVE_FP_SUFFIX): Likewise.
23084 (SIZE_TYPE): Likewise.
23085 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23086 (TARGET_64BIT): Likewise.
23087 (TARGET_64BIT): Likewise.
23088 (TARGET_AIX): Likewise.
23089 (WCHAR_TYPE_SIZE): Likewise.
23090 (WCHAR_TYPE): Undefine.
23091 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23092 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23093 (CPP_OS_RTEMS_SPEC): Delete.
23094 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23095 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23097 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23099 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23101 PR tree-optimization/81578
23102 * tree-parloops.c (build_new_reduction): Bail out if
23103 reduction_code isn't one of the standard OpenMP reductions.
23104 Move the details printing after that decision.
23106 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23108 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23109 related to reload_in_progress.
23110 (splat_input_operand): Likewise.
23111 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23113 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23115 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23116 (TARGET_INSTANTIATE_DECLS): Likewise.
23117 (legitimate_indexed_address_p): Delete reload_in_progress code.
23118 (rs6000_debug_legitimate_address_p): Likewise.
23119 (rs6000_eliminate_indexed_memrefs): Likewise.
23120 (rs6000_emit_le_vsx_store): Likewise.
23121 (rs6000_emit_move_si_sf_subreg): Likewise.
23122 (rs6000_emit_move): Likewise.
23123 (register_to_reg_type): Likewise.
23124 (rs6000_pre_atomic_barrier): Likewise.
23125 (rs6000_machopic_legitimize_pic_address): Likewise.
23126 (rs6000_allocate_stack_temp): Likewise.
23127 (rs6000_address_for_fpconvert): Likewise.
23128 (rs6000_address_for_altivec): Likewise.
23129 (rs6000_secondary_memory_needed_rtx): Delete function.
23130 (rs6000_check_sdmode): Likewise.
23131 (rs6000_alloc_sdmode_stack_slot): Likewise.
23132 (rs6000_instantiate_decls): Likewise.
23133 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23134 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23135 Delete reload_in_progress.
23136 (*vec_reload_and_plus_<mptrsize>): Likewise.
23137 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23138 (vsx_div_v2di): Likewise.
23139 (vsx_udiv_v2di): Likewise.
23141 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23143 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23144 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23145 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23146 (rs6000_debug_reg_global): Delete print of LRA status.
23147 (rs6000_option_override_internal): Delete dead LRA related code.
23148 (rs6000_lra_p): Delete function.
23149 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23151 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23153 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23154 * config/riscv/rtems.h: New file.
23156 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23157 Sudakshina Das <sudi.das@arm.com>
23159 * config/aarch64/aarch64.md
23160 (define_split for and<mode>3nr_compare): Move
23161 non aarch64_logical_operand to a register.
23162 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23163 register immediate operand to a register.
23164 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23166 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23168 PR middle-end/81564
23169 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23171 2017-07-27 Richard Biener <rguenther@suse.de>
23173 PR tree-optimization/81573
23174 PR tree-optimization/81494
23175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23176 multi defuse cycle case.
23178 2017-07-27 Richard Biener <rguenther@suse.de>
23180 PR tree-optimization/81571
23181 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23184 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23186 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23187 earlier and only if MASK_FPU is set. Adjust formatting.
23189 2017-07-27 Martin Liska <mliska@suse.cz>
23191 * opt-functions.awk: Add validation of value of Init.
23192 * optc-gen.awk: Pass new argument.
23194 2017-07-27 Martin Liska <mliska@suse.cz>
23196 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23197 Fix wrong condition.
23199 2017-07-27 Martin Liska <mliska@suse.cz>
23201 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23202 BBs and edges seen by autoFDO.
23204 2017-07-27 Richard Biener <rguenther@suse.de>
23206 PR tree-optimization/81502
23207 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23208 with incompatible but same sized type.
23209 (execute_update_addresses_taken): Likewise.
23211 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23213 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23214 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23216 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23219 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23220 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23221 Change s_operand to memory_operand.
23223 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23225 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23226 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23227 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23228 Emit instructions rather than returning an expression. Handle TFmode
23229 and KFmode by casting to TImode.
23230 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23231 (rs6000_emit_le_vsx_store): Likewise.
23232 * config/rs6000/vsx.md (VSX_TI): New iterator.
23233 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23234 (*vsx_le_undo_permute_<mode>): Likewise.
23235 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23236 emit the split sequence.
23237 (*vsx_le_perm_store_<mode>): Likewise.
23239 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23241 PR tree-optimization/81555
23242 PR tree-optimization/81556
23243 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23244 if true, force CHANGED for the recursive invocation.
23245 (reassociate_bb): Remember original length of ops array, pass
23246 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23248 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23249 attributes for noipa attribute. For naked attribute use
23250 lookup_attribute first before lookup_attribute_spec.
23251 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23253 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23254 for functions with noipa attribute.
23255 (cgraph_externally_visible_p): Return true for functions with noipa
23257 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23258 for functions with noipa attribute.
23259 * doc/extend.texi: Document noipa function attribute.
23260 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23261 also for functions with noipa attribute.
23262 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23264 2017-07-26 Andrew Pinski <apinski@cavium.com>
23266 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23267 vec_unalign_load_cost and vec_unalign_store_cost.
23269 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23271 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23272 -mvsx-small-integer option.
23273 (ISA_3_0_MASKS_IEEE): Likewise.
23274 (OTHER_VSX_VECTOR_MASKS): Likewise.
23275 (POWERPC_MASKS): Likewise.
23276 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23277 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23278 code, only testing for DImode being allowed in non-VSX floating
23280 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23281 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23283 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23284 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23285 TARGET_P8_VECTOR test.
23286 (rs6000_secondary_reload_simple_move): Likewise.
23287 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23288 since TARGET_P9_VECTOR was already tested.
23289 (rs6000_opt_masks): Remove -mvsx-small-integer.
23290 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23291 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23293 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23294 test for TARGET_VEXTRACTUB was used, and that uses
23296 (p9 extract splitter): Likewise.
23297 (vsx_extract_<mode>_di_p9): Likewise.
23298 (vsx_extract_<mode>_store_p9): Likewise.
23299 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23300 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23301 the elimination of TARGET_VSX_SMALL_INTEGER.
23302 (vsx_extract_<mode>_p8): Likewise.
23303 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23304 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23305 (vsx_set_<mode>_p9): Likewise.
23306 (vsx_set_v4sf_p9): Likewise.
23307 (vsx_set_v4sf_p9_zero): Likewise.
23308 (vsx_insert_extract_v4sf_p9): Likewise.
23309 (vsx_insert_extract_v4sf_p9_2): Likewise.
23310 * config/rs6000/rs6000.md (sign extend splitter): Change
23311 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23312 (floatsi<mode>2_lfiwax_mem): Likewise.
23313 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23314 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23315 since a test for TARGET_P9_VECTOR was used.
23316 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23317 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23318 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23319 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23320 TARGET_P8_VECTOR test.
23321 (fix_trunc<mode>si2_stfiwx): Likewise.
23322 (fix_trunc<mode>si2_internal): Likewise.
23323 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23324 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23326 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23327 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23328 TARGET_P8_VECTOR test.
23329 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23330 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23331 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23333 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23334 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23335 since a test for TARGET_P9_VECTOR was used.
23336 (splitter for loading small constants): Likewise.
23338 2017-07-26 Andrew Pinski <apinski@cavium.com>
23340 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23343 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23346 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23347 (fp_valid_at): Likewise.
23349 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23351 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23352 (qdf24xx_addrcost_table): Likewise.
23353 (cortexa57_tunings): Update to use generic_branch_cost.
23354 (cortexa72_tunings): Likewise.
23355 (cortexa73_tunings): Likewise.
23356 (qdf24xx_tunings): Likewise.
23358 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23360 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23361 (thunderx2t99_branch_cost): Likewise.
23362 (cortexa35_tunings): Update to use generic_branch_cost.
23363 (cortexa53_tunings): Likewise.
23364 (cortexa57_tunings): Likewise.
23365 (cortexa72_tunings): Likewise.
23366 (cortexa73_tunings): Likewise.
23367 (thunderx2t99_tunings): Likewise.
23369 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23371 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23372 (sparc_option_override): Honour MASK_FSMULD.
23373 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23374 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23375 * config/sparc/sparc.opt (mfsmuld): New option.
23376 * doc/invoke.texi (mfsmuld): Document option.
23378 2017-07-26 Marek Polacek <polacek@redhat.com>
23380 PR middle-end/70992
23381 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23383 2017-07-26 Richard Biener <rguenther@suse.de>
23385 * gimple-match-head.c (do_valueize): Return OP if valueize
23387 (get_def): New helper to get at the def stmt of a SSA name
23388 if valueize allows.
23389 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23390 do_valueize to get at the def stmt.
23391 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23393 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23395 PR middle-end/46932
23396 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23398 2017-07-26 Martin Liska <mliska@suse.cz>
23401 * function.c (expand_function_start): Make expansion of
23402 nonlocal_goto_save_area after parm_birth_insn.
23404 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23406 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23407 from all CPU target flags enable members.
23409 2017-07-26 Richard Biener <rguenther@suse.de>
23411 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23412 (decision_tree::gen): Make 'type' const.
23413 (write_predicate): Likewise.
23415 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23417 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23418 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23419 (rs6000_option_override_internal): Likewise.
23420 (rs6000_expand_vector_set): Likewise.
23421 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23422 (TARGET_UPPER_REGS_SF): Likewise.
23423 (TARGET_UPPER_REGS_DI): Likewise.
23424 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23425 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23426 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23427 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23428 (Splitters for DI constants in Altivec registers): Likewise.
23429 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23430 (vsx_set_v4sf_p9): Likewise.
23431 (vsx_set_v4sf_p9_zero): Likewise.
23432 (vsx_insert_extract_v4sf_p9): Likewise.
23433 (vsx_insert_extract_v4sf_p9_2): Likewise.
23435 2017-07-25 Carl Love <cel@us.ibm.com>
23437 * doc/extend.texi: Update the built-in documentation file for the
23438 existing built-in functions
23439 vector signed char vec_cnttz (vector signed char);
23440 vector unsigned char vec_cnttz (vector unsigned char);
23441 vector signed short vec_cnttz (vector signed short);
23442 vector unsigned short vec_cnttz (vector unsigned short);
23443 vector signed int vec_cnttz (vector signed int);
23444 vector unsigned int vec_cnttz (vector unsigned int);
23445 vector signed long long vec_cnttz (vector signed long long);
23446 vector unsigned long long vec_cnttz (vector unsigned long long);
23448 2017-07-25 Andrew Pinski <apinski@cavium.com>
23450 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23451 accesses where the use is for the first operand of a BIT_INSERT.
23453 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23456 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23457 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23459 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23461 * config/i386/gstabs.h: Delete.
23462 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23464 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23466 * config/i386/i386.c (ix86_decompose_address): Do not check for
23467 register RTX when looking at index_reg or base_reg.
23468 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23470 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23472 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23473 to update EH info here.
23475 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23477 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23479 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23481 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23483 2017-07-25 Torsten Duwe <duwe@suse.de>
23485 * common.opt: Introduce -fpatchable-function-entry
23486 command line option, and its variables function_entry_patch_area_size
23487 and function_entry_patch_area_start.
23488 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23489 including a two-value parser.
23490 * target.def (print_patchable_function_entry): New target hook.
23491 * targhooks.h (default_print_patchable_function_entry): New function.
23492 * targhooks.c (default_print_patchable_function_entry): Likewise.
23493 * toplev.c (process_options): Switch off IPA-RA if
23494 patchable function entries are being generated.
23495 * varasm.c (assemble_start_function): Look at the
23496 patchable-function-entry command line switch and current
23497 function attributes and maybe generate NOP instructions by
23498 calling the print_patchable_function_entry hook.
23499 * doc/extend.texi: Document patchable_function_entry attribute.
23500 * doc/invoke.texi: Document -fpatchable_function_entry
23501 command line option.
23502 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23504 * doc/tm.texi: Re-generate.
23506 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23509 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23510 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23512 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23514 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23516 2017-07-25 Richard Biener <rguenther@suse.de>
23518 PR tree-optimization/81455
23519 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23520 not walk in cycles when looking for guards.
23522 2017-07-25 Richard Biener <rguenther@suse.de>
23524 PR tree-optimization/81529
23525 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23526 when optimizing backedge uses.
23528 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23530 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23532 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23533 to dl_section_ref. On AIX, append an expression to subtract
23534 the size of the section length to dl_section_ref.
23536 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23538 * configure.ac: If any of the config.* scripts fail, exit 1.
23539 * configure: Regenerate.
23541 2017-07-25 Richard Biener <rguenther@suse.de>
23543 PR middle-end/81546
23544 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23545 of immediate uses, be more verbose on errors.
23547 2017-07-25 Richard Biener <rguenther@suse.de>
23549 PR tree-optimization/81510
23550 * tree-vect-loop.c (vect_is_simple_reduction): When the
23551 reduction stmt is not inside the loop bail out.
23553 2017-07-25 Richard Biener <rguenther@suse.de>
23555 PR tree-optimization/81303
23556 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23557 profitability check against LOOP_VINFO_NITERSM1.
23559 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23561 * domwalk.c (cmp_bb_postorder): Simplify.
23562 (sort_bbs_postorder): New function. Use it...
23563 (dom_walker::walk): ...here to optimize common cases.
23565 2017-07-25 Martin Liska <mliska@suse.cz>
23568 * ipa-visibility.c (function_and_variable_visibility): Make the
23569 redirection just on target that supports aliasing.
23570 Fix GNU coding style.
23572 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23575 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23577 * config/arm/rtems.h: Likewise.
23578 * config/bfin/rtems.h: Likewise.
23579 * config/i386/rtemself.h: Likewise.
23580 * config/lm32/rtems.h: Likewise.
23581 * config/m32c/rtems.h: Likewise.
23582 * config/m68k/rtemself.h: Likewise.
23583 * config/microblaze/rtems.h: Likewise.
23584 * config/mips/rtems.h: Likewise.
23585 * config/moxie/rtems.h: Likewise.
23586 * config/nios2/rtems.h: Likewise.
23587 * config/powerpcspe/rtems.h: Likewise.
23588 * config/rs6000/rtems.h: Likewise.
23589 * config/rtems.h: Likewise.
23590 * config/sh/rtems.h: Likewise.
23591 * config/sh/rtemself.h: Likewise.
23592 * config/sparc/rtemself.h: Likewise.
23594 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23597 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23598 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23599 * tree-ssa-structalias.c (alias_get_name): Same.
23601 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23604 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23605 instructions if no du chain is found.
23607 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23609 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23611 2017-07-25 Richard Biener <rguenther@suse.de>
23613 PR middle-end/81505
23614 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23617 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23619 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23620 upper-regs options.
23621 (ISA_2_7_MASKS_SERVER): Likewise.
23622 (ISA_3_0_MASKS_IEEE): Likewise.
23623 (OTHER_P8_VECTOR_MASKS): Likewise.
23624 (OTHER_VSX_VECTOR_MASKS): Likewise.
23625 (POWERPC_MASKS): Likewise.
23626 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23627 duplicate list of options.
23628 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23629 explicit -mupper-regs options.
23630 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23631 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23632 alias for -mupper-regs-df.
23633 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23634 (rs6000_init_hard_regno_mode_ok): Likewise.
23635 (rs6000_option_override_internal): Likewise.
23636 (rs6000_opt_masks): Likewise.
23637 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23638 options in terms of whether -mvsx or -mpower8-vector was used.
23639 (TARGET_UPPER_REGS_DI): Likewise.
23640 (TARGET_UPPER_REGS_SF): Likewise.
23641 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23642 -mupper-regs-* options.
23644 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23646 * passes.c (emergency_dump_function): Print some empty lines and a
23647 header before the RTL dump.
23649 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23651 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23653 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23656 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23657 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23659 2017-07-24 Carl Love <cel@us.ibm.com>
23661 * config/rs6000/rs6000-c.c: Add support for built-in functions
23662 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23663 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23664 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23665 vec_extract_fp_from_shortl): Add defines for the two builtins.
23666 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23667 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23669 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23670 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23671 * doc/extend.texi: Update the built-in documentation file for the
23672 new built-in function.
23674 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23677 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23679 * doc/generic.texi: Likewise.
23680 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23681 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23683 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23685 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23686 (aarch64_mls_elt_merge<mode>): Likewise.
23688 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23690 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23691 having __cxa_atexit.
23693 2017-07-23 Michael Collison <michael.collison@arm.com>
23695 * config/arm/arm.c (arm_option_override): Deprecate
23696 use of -mstructure-size-boundary.
23697 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23698 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23700 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23703 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23704 Reduce cost estimate for direct moves.
23706 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23709 * config/i386/i386.c (ix86_option_override_internal): Disable
23710 BMI, BMI2 and TBM instructions for -m16.
23712 2017-07-21 Carl Love <cel@us.ibm.com>
23714 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23715 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23716 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23717 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23718 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23719 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23720 VMULOSW): New enum "unspec" values.
23721 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23722 altivec_vmulosw): New patterns.
23723 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23724 VMULOSW): Add definitions.
23726 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23728 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23729 (qdf24xx): Likewise.
23730 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23731 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23732 (AARCH64_FL_V8_1): Renumber.
23733 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23734 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23735 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23736 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23737 rdma to feature modifiers list.
23739 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23741 PR middle-end/56727
23742 * ipa-visibility (function_and_variable_visibility): Convert
23743 recursive PLT call to direct call if appropriate.
23745 2017-07-21 Andrew Pinski <apinski@cavium.com>
23747 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23748 operand 1 to see if the types precision matches.
23749 * fold-const.c (operand_equal_p): Likewise.
23751 2017-07-21 Richard Biener <rguenther@suse.de>
23753 PR tree-optimization/81303
23754 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23755 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23756 (vect_peeling_hash_get_lowest_cost): Adjust.
23757 (vect_enhance_data_refs_alignment): Likewise. Use
23758 vect_get_peeling_costs_all_drs to compute the penalty for no
23759 peeling to match up costs.
23761 2017-07-21 Richard Biener <rguenther@suse.de>
23763 PR tree-optimization/81500
23764 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23765 we didn't identify a reduction path.
23767 2017-07-21 Tom de Vries <tom@codesourcery.com>
23768 Cesar Philippidis <cesar@codesourcery.com>
23770 PR gcov-profile/81442
23771 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23774 2017-07-21 Tom de Vries <tom@codesourcery.com>
23777 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23779 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23780 nvptx_override_options_after_change.
23782 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23784 * dwarf2out.c (output_file_names): Avoid double testing for
23785 dwarf_version >= 5.
23787 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23789 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23791 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23793 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23795 (try_crossjump_to_edge): Do not punt on partitioned functions.
23797 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23799 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23800 Put all BBs reachable only via paths crossing cold region to cold
23802 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23804 2016-07-21 Richard Biener <rguenther@suse.de>
23806 PR tree-optimization/81303
23807 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23808 into account prologue and epilogue iterations when raising
23809 min_profitable_iters to sth at least covering one vector iteration.
23811 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23813 * config/arm/arm.c (arm_test_cpu_arch_dat):
23816 2017-07-20 Nathan Sidwell <nathan@acm.org>
23818 Remove TYPE_METHODS.
23819 * tree.h (TYPE_METHODS): Delete.
23820 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23821 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23822 (dbxout_type_methods): Scan TYPE_FIELDS.
23823 (dbxout_type): Don't check TYPE_METHODS here.
23824 * function.c (use_register_for_decl): Always ignore register for
23825 class types when not optimizing.
23826 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23827 * tree.c (free_lang_data_in_type): Stitch out member functions and
23828 templates from TYPE_FIELDS.
23829 (build_distinct_type_copy, verify_type_variant,
23830 verify_type): Member fns are on TYPE_FIELDS.
23831 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23832 * tree-pretty-print.c (dump_generic_node): Likewise.
23834 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23837 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23838 V2TImode and V4TImode.
23839 (ix86_expand_vector_extract): Likewise.
23840 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23841 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23842 (ssescalarmode): Handle V4TImode and V2TImode.
23843 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23844 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23845 (VEXTRACTI128_MODE): New mode iterator.
23846 (splitter for *vec_extractv?ti first element): New.
23847 (VEC_INIT_MODE): New mode iterator.
23848 (vec_init<mode>): Consolidate 3 expanders into one using
23849 VEC_INIT_MODE mode iterator.
23851 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23853 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23854 non_spilled_static_chain_regno_p.
23856 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23858 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23860 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23862 * bb-reorder.c (connect_traces): Allow copying of blocks within
23865 2017-07-20 Richard Biener <rguenther@suse.de>
23867 * gimple.h (gimple_phi_result): Add gphi * overload.
23868 (gimple_phi_result_ptr): Likewise.
23869 (gimple_phi_arg): Likewise. Adjust index assert to only
23870 allow actual argument accesses rather than all slots available
23872 (gimple_phi_arg_def): Add gphi * overload.
23873 * tree-phinodes.c (make_phi_node): Initialize only actual
23875 (resize_phi_node): Clear memory not covered by old node,
23876 do not initialize excess argument slots.
23877 (reserve_phi_args_for_new_edge): Initialize new argument slot
23880 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23882 PR tree-optimization/81388
23884 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23886 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23887 by removing computation of may_be_zero.
23889 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23890 Tom de Vries <tom@codesourcery.com>
23892 PR middle-end/81030
23893 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23894 when gimple level profile disagrees with what RTL expander did.
23896 2017-07-20 Richard Biener <rguenther@suse.de>
23898 PR tree-optimization/61171
23899 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23900 (vect_analyze_stmt): Add slp instance parameter.
23901 (vectorizable_reduction): Likewise.
23902 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23903 (vect_is_simple_reduction): Deal with chains not detected
23904 as SLP reduction chain, specifically not properly associated
23905 chains containing a mix of plus/minus.
23906 (get_reduction_op): Remove.
23907 (get_initial_defs_for_reduction): Simplify, pass in whether
23908 this is a reduction chain, pass in the SLP node for the PHIs.
23909 (vect_create_epilog_for_reduction): Get the SLP instance as
23911 (vectorizable_reduction): Get the SLP instance as arg.
23912 During analysis remember the SLP node with the PHIs in the
23913 instance. Simplify getting at the vectorized reduction PHIs.
23914 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23915 through SLP instance.
23916 (vect_slp_analyze_operations): Likewise.
23917 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23918 (vect_transform_stmt): Likewise.
23920 2017-07-20 Tom de Vries <tom@codesourcery.com>
23922 PR tree-optimization/81489
23923 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23924 read of phi arg location to before loop that modifies phi.
23926 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23928 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23931 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23933 PR middle-end/81331
23934 * except.c (execute): Fix ordering issue.
23936 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23938 PR rtl-optimization/81423
23939 * combine.c (make_compound_operation_int): Don't try to optimize
23940 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23942 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23944 PR rtl-optimization/81423
23945 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23946 with a constant that is -1 in the truncated to mode.
23948 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23950 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23951 (determine_unlikely_bbs): ... here.
23952 * predict.h (propagate_unlikely_bbs_forward): Declare.
23953 * cfgexpand.c (pass_expand::execute): Use it.
23954 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23956 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23957 propagate_unlikely_bbs_forward.
23959 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23961 PR middle-end/81331
23962 * except.c (maybe_add_nop_after_section_switch): New function.
23965 2017-07-19 Tom de Vries <tom@codesourcery.com>
23967 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23969 2017-07-19 Tom de Vries <tom@codesourcery.com>
23971 * gimple.h (gimple_phi_arg): Make assert more strict.
23973 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23975 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23976 * config/rs6000/mmintrin.h: New file.
23977 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23979 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23981 PR tree-optimization/81346
23982 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23984 2017-07-19 Tom de Vries <tom@codesourcery.com>
23986 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23988 2017-07-19 Tom de Vries <tom@codesourcery.com>
23990 * config/nvptx/nvptx-modes.def: Add V2DImode.
23991 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23992 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23993 (nvptx_output_mov_insn): Handle lack of mov.b128.
23994 (nvptx_print_operand): Handle 'H' and 'L' codes.
23995 (nvptx_vector_mode_supported): Allow V2DImode.
23996 (nvptx_preferred_simd_mode): New function.
23997 (nvptx_data_alignment): New function.
23998 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23999 nvptx_preferred_simd_mode.
24000 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24002 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24004 2017-07-19 Tom de Vries <tom@codesourcery.com>
24006 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24007 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24008 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24009 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24010 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24011 (mov<VECIM>_insn): New define_insn.
24012 (define_expand "mov<VECIM>): New define_expand.
24014 2017-07-19 Tom de Vries <tom@codesourcery.com>
24016 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24018 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24020 PR tree-optimization/81346
24021 * fold-const.h (fold_div_compare, range_check_type): Declare.
24022 * fold-const.c (range_check_type): New function.
24023 (build_range_check): Use range_check_type.
24024 (fold_div_compare): No longer static, rewritten into
24025 a match.pd helper function.
24026 (fold_comparison): Don't call fold_div_compare here.
24027 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24028 as helper function.
24030 2017-07-19 Nathan Sidwell <nathan@acm.org>
24032 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24033 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24034 * tree.c (find_decls_types_r, verify_type): Use
24035 TYPE_{MIN,MAX}_VALUE_RAW.
24036 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24037 (hash_tree): Likewise.
24038 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24040 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24043 2017-07-18 Tom de Vries <tom@codesourcery.com>
24045 PR middle-end/81464
24046 * omp-expand.c (expand_omp_for_static_chunk): Handle
24047 equal-argument loop exit phi.
24049 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24052 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24053 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24054 operand 2 predicate.
24055 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24056 operand 2 predicate.
24057 (ror,rol -> rorx splitters): Use const_int_operand as
24058 operand 2 predicate.
24060 2017-06-18 Richard Biener <rguenther@suse.de>
24062 PR tree-optimization/81410
24063 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24064 the gap in the ! slp_perm SLP case after each group.
24066 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24068 PR middle-end/81463
24069 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24072 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24074 PR middle-end/81462
24075 * predict.c (set_even_probabilities): Cleanup; do not affect
24076 probabilities that are already known.
24077 (combine_predictions_for_bb): Call even when count is set.
24079 2017-07-18 Nathan Sidwell <nathan@acm.org>
24081 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24084 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24087 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24088 optimization for loop niter analysis.
24090 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24093 * config/avr/avr.c (avr_optimize_casesi): Don't use
24094 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24096 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24098 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24099 body_cost_vec from _vect_peel_extended_info.
24100 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24101 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24104 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24106 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24108 2017-07-18 Richard Biener <rguenther@suse.de>
24110 PR tree-optimization/80620
24111 PR tree-optimization/81403
24112 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24113 info when re-using a VN table entry.
24115 2017-07-18 Richard Biener <rguenther@suse.de>
24117 PR tree-optimization/81418
24118 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24119 vectype_in. Verify that with lane-reducing reduction operations
24120 we have a single def-use cycle.
24122 2017-07-17 Carl Love <cel@us.ibm.com>
24124 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24126 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24127 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24128 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24129 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24130 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24131 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24132 VMULOSW): New enum "unspec" values.
24133 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24134 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24135 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24136 altivec_vmulosw): New patterns.
24137 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24138 VMULOSW): Add definitions.
24140 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24142 * config/alpha/alpha.c: Include predict.h.
24144 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24146 * tree-vrp.c (compare_assert_loc): Fix comparison function
24147 to return predictable results.
24149 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24151 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24153 (subdi3): Likewise.
24154 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24155 * doc/invoke.texi (mexpand-adddi): Update text.
24157 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24159 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24160 that also clobbers the CC register. The old expand code is moved
24162 (*arc_clzsi2): ... here.
24163 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24164 the CC register. The old expand code is moved to ...
24165 (arc_ctzsi2): ... here.
24167 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24169 * config/arc/arc.opt (mindexed-loads): Use initial value
24170 TARGET_INDEXED_LOADS_DEFAULT.
24171 (mauto-modify-reg): Use initial value
24172 TARGET_AUTO_MODIFY_REG_DEFAULT.
24173 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24174 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24175 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24176 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24178 2017-07-17 Martin Liska <mliska@suse.cz>
24181 * opts.c (finish_options): Do not allow -fgnu-tm
24182 w/ -fsanitize={kernel-,}address. Say sorry.
24184 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24187 * tree-loop-distribution.c (classify_partition): Only assert on
24188 numer of iterations.
24189 (merge_dep_scc_partitions): Delete prameter. Update function call.
24190 (distribute_loop): Remove code handling loop with unknown niters.
24191 (pass_loop_distribution::execute): Skip loop with unknown niters.
24193 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24196 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24197 function sort_partitions_by_post_order.
24199 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24201 PR tree-optimization/81374
24202 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24203 the max index of basic blocks, rather than number of basic blocks.
24205 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24207 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24209 (arc_legitimate_pic_operand_p): Likewise.
24210 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24212 (arc_needs_pcl_p): Likewise.
24213 (arc_legitimate_pc_offset_p): Likewise.
24214 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24215 function is also used in constrains.md.
24216 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24217 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24218 PLUS. Only return true/false in known cases, otherwise assert.
24219 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24220 is already called in arc_legitimate_constant_p.
24221 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24223 (LEGITIMATE_PIC_OPERAND_P): Use
24224 arc_raw_symbolic_reference_mentioned_p function.
24225 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24230 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24231 Andrew Burgess <andrew.burgess@embecosm.com>
24233 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24234 (arc_return_address_register): New function.
24235 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24236 (arc_handle_fndecl_attribute): Add naked attribute.
24237 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24238 (TARGET_WARN_FUNC_RETURN): Likewise.
24239 (arc_allocate_stack_slots_for_args): New function.
24240 (arc_warn_func_return): Likewise.
24241 (machine_function): Change type fn_type.
24242 (arc_compute_function_type): Consider new naked function type,
24243 change function return type.
24244 (arc_must_save_register): Adapt to handle new
24245 arc_compute_function_type's return type.
24246 (arc_expand_prologue): Likewise.
24247 (arc_expand_epilogue): Likewise.
24248 (arc_return_address_regs): Delete.
24249 (arc_return_address_register): New function.
24250 (arc_epilogue_uses): Use above function.
24251 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24252 (arc_function_type): Change encoding, add naked type.
24253 (ARC_INTERRUPT_P): Change to handle the new encoding.
24254 (ARC_FAST_INTERRUPT_P): Likewise.
24255 (ARC_NORMAL_P): Define.
24256 (ARC_NAKED_P): Likewise.
24257 (arc_compute_function_type): Delete prototype.
24258 * config/arc/arc.md (in_ret_delay_slot): Use
24259 arc_return_address_register function.
24260 (simple_return): Likewise.
24261 (p_return_i): Likewise.
24263 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24265 PR tree-optimization/81428
24266 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24267 can't be built for those types.
24269 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24271 Remove stuff dead since r239246.
24273 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24274 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24275 (avr_inform_devices): Remove dead stuff.
24277 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24279 * config/arm/arm_neon.h: Fix softp typo.
24281 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24283 PR tree-optimization/81365
24284 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24285 aggregate moves onto bb predecessor edges, make sure there are no
24286 loads that could alias the lhs in between the start of bb and the
24289 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24292 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24293 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24294 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24296 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24298 PR tree-optimization/81396
24299 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24300 (init_symbolic_number): Initialize it to 1.
24301 (perform_symbolic_merge): Add n_ops from both operands into the new
24303 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24304 without base_addr as useless if they need more than one operation.
24305 (bswap_replace): Handle !bswap case for NULL base_addr.
24307 2017-07-17 Tom de Vries <tom@codesourcery.com>
24310 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24313 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24315 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24316 conditional builtin define __FIX_LEON3FT_B2BST.
24318 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24320 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24321 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24324 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24326 PR rtl-optimization/81424
24327 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24328 to remove potential trapping from operands if -fnon-call-exceptions.
24330 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24333 profile_proability for scalling.
24334 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24336 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24338 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24340 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24342 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24343 fixpoint arithmetics.
24345 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24347 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24348 fixpoint arithmetics.
24350 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24352 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24353 fixpoint arithmetics.
24355 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24357 * profile-count.h (profile_probability::from_reg_br_prob_note,
24358 profile_probability::to_reg_br_prob_note): New functions.
24359 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24360 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24361 * predict.c (probability_reliable_p): Update.
24362 (edge_probability_reliable_p): Update.
24363 (br_prob_note_reliable_p): Update.
24364 (invert_br_probabilities): Update.
24365 (add_reg_br_prob_note): New function.
24366 (combine_predictions_for_insn): Update.
24367 * asan.c (asan_clear_shadow): Update.
24368 * cfgbuild.c (compute_outgoing_frequencies): Update.
24369 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24370 (update_br_prob_note): Update.
24371 (rtl_verify_edges): Update.
24372 (purge_dead_edges): Update.
24373 (fixup_reorder_chain): Update.
24374 * emit-rtl.c (try_split): Update.
24375 * ifcvt.c (cond_exec_process_insns): Update.
24376 (cond_exec_process_if_block): Update.
24377 (dead_or_predicable): Update.
24378 * internal-fn.c (expand_addsub_overflow): Update.
24379 (expand_neg_overflow): Update.
24380 (expand_mul_overflow): Update.
24381 * loop-doloop.c (doloop_modify): Update.
24382 * loop-unroll.c (compare_and_jump_seq): Update.
24383 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24384 * predict.h: Update.
24385 * reorg.c (mostly_true_jump): Update.
24387 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24388 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24389 * config/arc/arc.c: (emit_unlikely_jump): Update.
24390 * config/arm/arm.c: (emit_unlikely_jump): Update.
24391 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24392 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24393 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24394 (ix86_print_operand): Update.
24395 (ix86_split_fp_branch): Update.
24396 (predict_jump): Update.
24397 * config/ia64/ia64.c (ia64_print_operand): Update.
24398 * config/mmix/mmix.c (mmix_print_operand): Update.
24399 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24400 (rs6000_expand_split_stack_prologue): Update.
24401 * config/rs6000/rs6000.c: Update.
24402 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24403 (s390_expand_vec_movstr): Update.
24404 (s390_expand_cs_tdsi): Update.
24405 (s390_expand_split_stack_prologue): Update.
24406 * config/sh/sh.c (sh_print_operand): Update.
24407 (expand_cbranchsi4): Update.
24408 (expand_cbranchdi4): Update.
24409 * config/sparc/sparc.c (output_v9branch): Update.
24410 * config/spu/spu.c (get_branch_target): Update.
24411 (ea_load_store_inline): Update.
24412 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24413 * config/tilepro/tilepro.c: Update.
24415 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24417 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24418 (gimplify_save_expr): Likewise.
24420 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24422 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24424 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24426 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24428 * timevar.def (TV_IPA_FNSUMMARY): Define.
24430 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24432 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24433 to back store errata sensitive sequence from being generated.
24434 (sqrtdf2_fix): Likewise.
24436 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24438 * tree-ssa-threadupdate.c (compute_path_counts,
24439 update_joiner_offpath_counts): Use profile_probability.
24441 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24444 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24446 * config/arm/arm-c.c (arm_cpu_builtins): Define
24447 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24449 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24452 array entries to represent __ieee128 versions of the
24453 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24454 scalar_extract_sig, and scalar_insert_exp built-in functions.
24455 (altivec_resolve_overloaded_builtin): Add special case handling
24456 for the __builtin_scalar_insert_exp function, as represented by
24457 the P9V_BUILTIN_VEC_VSIEDP constant.
24458 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24459 exponent support for __ieee128 argument.
24460 (VSESQP): Add scalar extract signature support for __ieee128
24462 (VSTDCNQP): Add scalar test negative support for __ieee128
24464 (VSIEQP): Add scalar insert exponent support for __int128 argument
24465 with __ieee128 result.
24466 (VSIEQPF): Add scalar insert exponent support for __ieee128
24467 argument with __ieee128 result.
24468 (VSTDCQP): Add scalar test data class support for __ieee128
24470 (VSTDCNQP): Add overload support for scalar test negative with
24471 __ieee128 argument.
24472 (VSTDCQP): Add overload support for scalar test data class
24473 __ieee128 argument.
24474 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24475 UNSPEC_VSX_SXSIGDP.
24476 (UNSPEC_VSX_SIEXPQP): New constant.
24477 (xsxexpqp): New insn for VSX scalar extract exponent quad
24479 (xsxsigqp): New insn for VSX scalar extract significand quad
24481 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24482 precision with floating point argument.
24483 (xststdcqp): New expand for VSX scalar test data class quad
24485 (xststdcnegqp): New expand for VSX scalar test negative quad
24487 (xststdcqp): New insn to match expansions for VSX scalar test data
24488 class quad precision and VSX scalar test negative quad precision.
24489 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24490 special case operand checking to enforce that second operand of
24491 VSX scalar test data class with quad precision argument is a 7-bit
24493 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24494 prototypes and descriptions of __ieee128 versions of
24495 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24496 scalar_test_data_class, and scalar_test_neg built-in functions.
24498 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24500 PR tree-optimization/81162
24501 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24502 replace a negate with an add.
24504 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24506 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24508 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24510 * config/arm/arm-c.c (arm_cpu_builtins): Define
24511 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24513 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24515 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24516 (armv8-r): Set ARM Cortex-R52 as default CPU.
24517 * config/arm/arm-tables.opt: Regenerate.
24518 * config/arm/arm-tune.md: Regenerate.
24519 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24521 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24522 extension for -mcpu=cortex-r52.
24524 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24526 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24527 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24528 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24529 (fp-armv8): Define it as FP_ARMv8 only.
24530 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24531 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24533 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24534 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24535 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24536 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24537 than TARGET_FPU_ARMV8.
24538 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24539 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24540 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24542 * config/arm/neon.md (neon_vrint): Likewise.
24543 (neon_vcvt): Likewise.
24544 (neon_<fmaxmin_op><mode>): Likewise.
24545 (<fmaxmin><mode>3): Likewise.
24546 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24547 * config/arm/predicates.md (arm_cond_move_operator): Check against
24548 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24550 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24552 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24553 to top of function.
24555 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24557 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24558 loop in comment with memset.
24560 2017-07-14 Martin Liska <mliska@suse.cz>
24562 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24563 * dwarf2out.c (is_java): Remove the function.
24564 (output_pubname): Remove usage of the function.
24565 (lower_bound_default): Remove usage of DW_LANG_Java.
24566 (gen_compile_unit_die): Likewise.
24567 * gcc.c: Remove compiler defaults for .java and .zip files.
24568 * gimple-expr.c (remove_suffix): Change as there's no longer
24569 extension than 4-letter one.
24570 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24571 (gimplify_save_expr): Likewise.
24572 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24573 as it's possible even for other languages than Java.
24574 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24575 * lto-opts.c (lto_write_options): Remove reference to Java.
24576 * opts.c (strip_off_ending): Update file extension handling.
24577 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24578 * tree-eh.c (lower_resx): Likewise.
24579 * tree.c (free_lang_data_in_type): Remove dead code.
24580 (find_decls_types_r): Likewise.
24581 (build_common_builtin_nodes): Remove Java from a comment.
24582 (verify_type): Remove dead code.
24583 * varasm.c (assemble_external): Remove Java from a comment.
24585 2017-07-14 Martin Liska <mliska@suse.cz>
24587 * opts.c (finish_options): Add quotes.
24588 (common_handle_option): Likewise.
24590 2017-07-14 Martin Liska <mliska@suse.cz>
24592 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24593 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24594 Remove N_SO_PASCAL.
24595 * dwarf2out.c (lower_bound_default): Do not handle
24597 (gen_compile_unit_die): Likewise.
24598 * gcc.c: Remove default extension binding for GNU Pascal.
24599 * stmt.c: Remove Pascal language from a comment.
24600 * xcoffout.c: Likewise.
24602 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24605 * diagnostic-show-locus.c (fixit_cmp): New function.
24606 (layout::layout): Sort m_fixit_hints.
24607 (column_range::column_range): Assert that the values are valid.
24608 (struct char_span): New struct.
24609 (correction::overwrite): New method.
24610 (struct source_line): New struct.
24611 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24612 calls in terms of classes source_line and char_span, and
24613 correction::overwrite.
24614 (selftest::test_overlapped_fixit_printing_2): New function.
24615 (selftest::diagnostic_show_locus_c_tests): Call it.
24617 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24619 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24620 early if there is no lhs.
24622 2017-07-13 Martin Liska <mliska@suse.cz>
24624 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24625 (gen_reference_type_die): Likewise.
24626 * stor-layout.c: Remove Pascal-related comment.
24628 2017-07-13 Martin Liska <mliska@suse.cz>
24630 * opts.c (finish_options): Add quotes to error messages.
24631 (parse_sanitizer_options): Likewise.
24633 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24635 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24637 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24639 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24641 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24643 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24645 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24647 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24650 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24651 provides the hardware capability bits, define the macro
24652 __BUILTIN_CPU_SUPPORTS__.
24653 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24654 if GLIBC does not provide the hardware capability bits. Add a
24655 gcc_unreachable call if the built-in cpu function is neither
24656 __builtin_cpu_is nor __builtin_cpu_supports.
24657 (rs6000_get_function_versions_dispatcher): Change the warning
24658 that an old GLIBC is used which does not export the capability
24659 bits to be an error.
24660 * doc/extend.texi (target_clones attribute): Document the
24661 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24662 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24663 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24664 the macros defined by GCC if the newer GLIBC is available.
24666 2017-07-12 Jeff Law <law@redhat.com>
24668 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24669 remaining includes slightly.
24670 * config/riscv/riscv-builtins.c: Include profile-count.h.
24672 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24675 * config/avr/avr.c (avr_set_current_function): In diagnostic
24676 messages: Quote keywords and (parts of) identifiers.
24677 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24680 2017-07-12 Carl Love <cel@us.ibm.com>
24682 * config/rs6000/rs6000-c.c: Add support for built-in functions
24683 vector bool char vec_revb (vector bool char);
24684 vector bool short vec_revb (vector short char);
24685 vector bool int vec_revb (vector bool int);
24686 vector bool long long vec_revb (vector bool long long);
24687 * doc/extend.texi: Update the built-in documentation file for the
24688 new built-in functions.
24690 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24692 * config/s390/s390.md: Remove movcc splitter.
24694 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24696 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24697 load/store on condition.
24699 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24702 * config/avr/avr.c (avr_encode_section_info)
24703 [progmem && !TREE_READONLY]: Error if progmem object needs
24706 2017-07-11 Michael Collison <michael.collison@arm.com>
24708 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24711 2017-07-11 Carl Love <cel@us.ibm.com>
24713 * config/rs6000/rs6000-c.c: Add support for builtins
24714 vector unsigned int vec_parity_lsbb (vector signed int);
24715 vector unsigned int vec_parity_lsbb (vector unsigned int);
24716 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24717 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24718 vector unsigned long long vec_parity_lsbb (vector signed long long);
24719 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24720 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24721 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24722 * doc/extend.texi: Update the built-in documentation file for the
24723 new built-in functions.
24725 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24727 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24728 (layout::m_primary_loc): New field.
24729 (layout::layout): Initialize new field. Move location filtering
24730 logic from here to...
24731 (layout::maybe_add_location_range): ...this new method. Add
24732 support for filtering to just the lines already specified by other
24734 (layout::will_show_line_p): New method.
24735 (gcc_rich_location::add_location_if_nearby): New method.
24736 (selftest::test_add_location_if_nearby): New test function.
24737 (selftest::diagnostic_show_locus_c_tests): Call it.
24738 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24741 2017-07-11 Tom de Vries <tom@codesourcery.com>
24743 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24744 (bb_first_real_insn): New function.
24745 (nvptx_single): Add extra initialization of broadcasted condition
24748 2017-07-11 Nathan Sidwell <nathan@acm.org>
24750 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24752 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24754 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24755 Binutils doc as TEXI will mess them up.
24756 * doc/invoke.texi (AVR Options): Same here.
24758 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24760 * config/sparc/sparc.opt (mfix-ut700): New option.
24761 (mfix-gr712rc): Likewise.
24762 (sparc_fix_b2bst): New variable.
24763 * doc/invoke.texi (SPARC options): Document them.
24764 (ARM options): Fix warnings.
24765 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24766 instructions to prevent sequences that can trigger the store-store
24767 errata for certain LEON3FT processors.
24768 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24769 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24770 * config/sparc/sparc.md (fix_b2bst): New attribute.
24771 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24773 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24776 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24778 (*rsqrtsf2_sse): Ditto.
24780 (div<mode>3): Macroize insn from divdf3 and divsf3
24781 using MODEF mode iterator.
24783 2017-07-10 Martin Sebor <msebor@redhat.com>
24785 PR tree-optimization/80397
24786 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24787 instead of testing for equality to INTEGER_TYPE.
24789 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24791 * config.gcc: Remove uclibc from arc target spec.
24793 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24795 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24797 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24800 * lto-wrapper.c (remove_option): New function.
24801 (merge_and_complain): Merge PIC/PIE options more realistically.
24803 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24805 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24809 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24810 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24811 * config.in: Regenerate.
24812 * configure: Regenerate.
24813 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24814 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24815 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24816 (TARGET_GASISR_PROLOGUES): ...target mask.
24817 * common/config/avr/avr-common.c
24818 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24819 Set -mgas-isr-prologues.
24820 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24821 INSERT_PASS_BEFORE for it.
24822 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24823 * config/avr/avr.c (avr_option_override)
24824 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24825 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24826 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24827 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24828 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24830 (make_avr_pass_pre_proep): New function.
24831 (emit_push_sfr) <treg>: Add argument to function and use it
24832 instead of TMP_REG.
24833 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24834 and set machine->gasisr.yes.
24835 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24836 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24837 __gcc_isr.n_pushed to .L__stack_usage.
24838 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24839 (avr_asm_final_postscan_insn): ...this new static function.
24840 * config/avr/avr.h (machine_function)
24841 <is_no_gccisr, use_L__stack_usage>: New fields.
24842 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24843 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24844 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24845 (gasisr, *gasisr): New expander and insn.
24846 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24847 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24848 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24850 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24852 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24855 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24857 Move jump-tables out of .text again.
24860 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24861 (ASM_OUTPUT_ADDR_VEC): New function.
24862 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24863 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24864 INSN_ADDRESSes as asm comment.
24865 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24866 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24867 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24868 * config/avr/avr.md (*tablejump): Adjust comment.
24869 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24870 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24872 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24873 (avr_output_addr_vec): New proto.
24874 (avr_log_t) <insn_addresses>: New field.
24876 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24879 * config/i386/i386.c (ix86_function_arg_advance): Set
24880 outgoing_args_on_stack to true if there are outgoing arguments
24882 (ix86_function_arg): Likewise.
24883 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24884 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24885 * config/i386/i386.h (machine_function): Add
24886 outgoing_args_on_stack.
24888 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24890 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24891 supporting pthreds.
24892 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24894 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24896 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24897 (REAL_H): Remove $(MACHMODE_H).
24898 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24900 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24901 $(MACHMODE_H) and double-int.h.
24902 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24904 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24907 2017-07-07 Andrew Pinski <apinski@cavium.com>
24909 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24910 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24912 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24914 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24915 Add warning if GCC was not configured to link against a GLIBC that
24916 exports the hardware capability bits.
24917 (make_resolver_func): Make resolver function private and not a
24918 COMDAT function. Create the name with clone_function_name instead
24919 of make_unique_name.
24922 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24923 correct operand in doing the split.
24925 2017-07-07 Carl Love <cel@us.ibm.com>
24927 * config/rs6000/rs6000-c: Add support for built-in function
24928 vector unsigned short vec_pack_to_short_fp32 (vector float,
24930 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24931 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24932 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24933 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24934 (convert_4f32_8i16): Add define_expand.
24935 * doc/extend.texi: Update the built-in documentation file for the
24936 new built-in function.
24938 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24940 * config/sparc/m8.md: New file.
24941 * config/sparc/sparc.md: Include m8.md.
24943 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24945 * config/sparc/sparc.opt: New option -mvis4b.
24946 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24947 (sparc_option_override): Handle VIS4B.
24948 (enum sparc_builtins): Define
24949 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24950 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24951 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24952 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24953 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24954 (check_constant_argument): New function.
24955 (sparc_vis_init_builtins): Define builtins
24956 __builtin_vis_dictunpack{8,16,32},
24957 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24958 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24959 __builtin_vis_fpcmpde{8,16,32}shl and
24960 __builtin_vis_fpcmpur{8,16,32}shl.
24961 (sparc_expand_builtin): Check that the constant operands to
24962 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24963 constant and in range.
24964 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24966 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24967 (SPARC_IMM5_P): Likewise.
24968 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24969 (enabled): Handle vis4b.
24970 (UNSPEC_DICTUNPACK): New unspec.
24971 (UNSPEC_FPCMPSHL): Likewise.
24972 (UNSPEC_FPUCMPSHL): Likewise.
24973 (UNSPEC_FPCMPDESHL): Likewise.
24974 (UNSPEC_FPCMPURSHL): Likewise.
24975 (cpu_feature): New CPU feature `vis4b'.
24976 (dictunpack{8,16,32}): New insns.
24977 (FPCSMODE): New mode iterator.
24978 (fpcscond): New code iterator.
24979 (fpcsucond): Likewise.
24980 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24981 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24982 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24983 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24984 * config/sparc/constraints.md: Define constraints `q' for unsigned
24985 2-bit integer constants and `t' for unsigned 5-bit integer
24987 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24989 (imm5_operand_dictunpack16): Likewise.
24990 (imm5_operand_dictunpack32): Likewise.
24991 (imm2_operand): Likewise.
24992 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24993 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24994 ditunpack* and fpcmp*shl builtins.
24996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24998 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24999 * config.in: Add HAVE_AS_SPARC6 define.
25000 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25002 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25004 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25005 (CPP_CPU_SPEC): Handle m8.
25006 (ASM_CPU_SPEC): Likewise.
25007 * config/sparc/sparc-opts.h (enum processor_type): Add
25009 * config/sparc/sparc.c (m8_costs): New struct.
25010 (sparc_option_override): Handle TARGET_CPU_m8.
25011 (sparc32_initialize_trampoline): Likewise.
25012 (sparc64_initialize_trampoline): Likewise.
25013 (sparc_issue_rate): Likewise.
25014 (sparc_register_move_cost): Likewise.
25015 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25016 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25017 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25018 (CPP_CPU_SPEC): Handle M8.
25019 (ASM_CPU_SPEC): Likewise.
25020 (AS_M8_FLAG): Define.
25021 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25022 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25023 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25025 * configure: Regenerate.
25026 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25029 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25031 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25033 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25034 ("*movdi_insn_sp32"): Do not set v3pipe.
25035 ("*movsi_insn"): Likewise.
25036 ("*movdi_insn_sp64"): Likewise.
25037 ("*movsf_insn"): Likewise.
25038 ("*movdf_insn_sp32"): Likewise.
25039 ("*movdf_insn_sp64"): Likewise.
25040 ("*zero_extendsidi2_insn_sp64"): Likewise.
25041 ("*sign_extendsidi2_insn"): Likewise.
25042 ("*mov<VM32:mode>_insn"): Likewise.
25043 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25044 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25045 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25046 ("<vlop:code><VL:mode>3"): Likewise.
25047 ("*not_<vlop:code><VL:mode>3"): Likewise.
25048 ("*nand<VL:mode>_vis"): Likewise.
25049 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25050 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25051 ("one_cmpl<VL:mode>2"): Likewise.
25052 ("faligndata<VM64:mode>_vis"): Likewise.
25053 ("alignaddrsi_vis"): Likewise.
25054 ("alignaddrdi_vis"): Likweise.
25055 ("alignaddrlsi_vis"): Likewise.
25056 ("alignaddrldi_vis"): Likewise.
25057 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25058 ("bmaskdi_vis"): Likewise.
25059 ("bmasksi_vis"): Likewise.
25060 ("bshuffle<VM64:mode>_vis"): Likewise.
25061 ("cmask8<P:mode>_vis"): Likewise.
25062 ("cmask16<P:mode>_vis"): Likewise.
25063 ("cmask32<P:mode>_vis"): Likewise.
25064 ("pdistn<P:mode>_vis"): Likewise.
25065 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25067 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25069 * config/sparc/sparc.md ("subtype"): New insn attribute.
25070 ("*wrgsr_sp64"): Set insn subtype.
25071 ("*rdgsr_sp64"): Likewise.
25072 ("alignaddrsi_vis"): Likewise.
25073 ("alignaddrdi_vis"): Likewise.
25074 ("alignaddrlsi_vis"): Likewise.
25075 ("alignaddrldi_vis"): Likewise.
25076 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25077 ("fexpand_vis"): Likewise.
25078 ("fpmerge_vis"): Likewise.
25079 ("faligndata<VM64:mode>_vis"): Likewise.
25080 ("bshuffle<VM64:mode>_vis"): Likewise.
25081 ("cmask8<P:mode>_vis"): Likewise.
25082 ("cmask16<P:mode>_vis"): Likewise.
25083 ("cmask32<P:mode>_vis"): Likewise.
25084 ("fchksm16_vis"): Likewise.
25085 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25086 ("fmean16_vis"): Likewise.
25087 ("fp<plusminus_insn>64_vis"): Likewise.
25088 ("<plusminus_insn>v8qi3"): Likewise.
25089 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25090 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25091 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25092 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25093 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25094 ("*movqi_insn"): Likewise.
25095 ("*movhi_insn"): Likewise.
25096 ("*movsi_insn"): Likewise.
25097 ("movsi_pic_gotdata_op"): Likewise.
25098 ("*movdi_insn_sp32"): Likewise.
25099 ("*movdi_insn_sp64"): Likewise.
25100 ("movdi_pic_gotdata_op"): Likewise.
25101 ("*movsf_insn"): Likewise.
25102 ("*movdf_insn_sp32"): Likewise.
25103 ("*movdf_insn_sp64"): Likewise.
25104 ("*zero_extendhisi2_insn"): Likewise.
25105 ("*zero_extendqihi2_insn"): Likewise.
25106 ("*zero_extendqisi2_insn"): Likewise.
25107 ("*zero_extendqidi2_insn"): Likewise.
25108 ("*zero_extendhidi2_insn"): Likewise.
25109 ("*zero_extendsidi2_insn_sp64"): Likewise.
25110 ("ldfsr"): Likewise.
25111 ("prefetch_64"): Likewise.
25112 ("prefetch_32"): Likewise.
25113 ("tie_ld32"): Likewise.
25114 ("tie_ld64"): Likewise.
25115 ("*tldo_ldub_sp32"): Likewise.
25116 ("*tldo_ldub1_sp32"): Likewise.
25117 ("*tldo_ldub2_sp32"): Likewise.
25118 ("*tldo_ldub_sp64"): Likewise.
25119 ("*tldo_ldub1_sp64"): Likewise.
25120 ("*tldo_ldub2_sp64"): Likewise.
25121 ("*tldo_ldub3_sp64"): Likewise.
25122 ("*tldo_lduh_sp32"): Likewise.
25123 ("*tldo_lduh1_sp32"): Likewise.
25124 ("*tldo_lduh_sp64"): Likewise.
25125 ("*tldo_lduh1_sp64"): Likewise.
25126 ("*tldo_lduh2_sp64"): Likewise.
25127 ("*tldo_lduw_sp32"): Likewise.
25128 ("*tldo_lduw_sp64"): Likewise.
25129 ("*tldo_lduw1_sp64"): Likewise.
25130 ("*tldo_ldx_sp64"): Likewise.
25131 ("*mov<VM32:mode>_insn"): Likewise.
25132 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25133 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25135 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25137 * config/sparc/sparc.md ("type"): New insn type viscmp.
25138 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25140 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25141 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25142 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25143 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25145 ("n7_vis_logical_11cycle"): Likewise.
25146 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25147 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25148 * config/sparc/niagara.md ("niag_vis"): Likewise.
25149 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25150 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25152 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25154 * config/sparc/sparc.md: New instruction type `bmask'.
25155 (bmaskdi_vis): Use the `bmask' type.
25156 (bmasksi_vis): Likewise.
25157 * config/sparc/ultra3.md (us3_array): Likewise.
25158 * config/sparc/niagara7.md (n7_array): Likewise.
25159 * config/sparc/niagara4.md (n4_array): Likewise.
25160 * config/sparc/niagara2.md (niag2_vis): Likewise.
25161 (niag3_vis): Likewise.
25162 * config/sparc/niagara.md (niag_vis): Likewise.
25164 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25166 * ipa-comdats.c: Remove optimize check from gate.
25167 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25168 for functions not optimized.
25169 (ipa_fn_summary_read): Skip optimize check.
25170 (ipa_fn_summary_write): Likewise.
25171 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25173 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25175 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25176 (check_callers): Give up on caller which is not optimized.
25177 (inline_small_functions): Likewise.
25178 (ipa_inline): Do not give up when not optimizing.
25179 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25180 away unoptimizes cdtors.
25181 (whole_program_function_and_variable_visibility): Do
25182 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25183 * ipa.c (process_references): Do not check optimize.
25184 (symbol_table::remove_unreachable_nodes): Update optimize check.
25185 (set_writeonly_bit): Update optimize check.
25186 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25187 (pass_ipa_single_use::gate): Remove.
25189 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25191 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25192 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25193 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25194 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25195 permute_load, permute_store, adjust_extract, adjust_splat,
25196 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25197 replace_swap_with_copy, dump_swap_insn_table,
25198 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25199 recombine_lvx_pattern, recombine_stvx_pattern,
25200 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25201 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25202 to file rs6000-p8swap.c.
25203 * config/rs6000/rs6000-p8swap.c: New file.
25204 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25205 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25206 and rs6000*-*-* targets.
25208 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25210 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25212 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25214 * lto-wrapper.c (merge_and_complain): Do not merge
25215 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25216 fsigned_zeros, ftrapping_math, fwrapv.
25217 (append_compiler_options): Do not track these options.
25218 (append_linker_options): Likewie
25220 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25222 * cgraphunit.c (cgraph_node::finalize_function): When
25223 !flag_toplevel_reorde set no_reorder flag.
25224 (varpool_node::finalize_decl): Likewise.
25225 (symbol_table::compile): Drop no toplevel reorder path.
25227 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25229 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25230 edges; zero probability is not better than uninitialized.
25232 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25234 * asan.h (asan_sanitize_allocas_p): Declare.
25235 * asan.c (asan_sanitize_allocas_p): New function.
25236 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25237 (handle_builtin_alloca): Likewise.
25238 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25239 if !asan_sanitize_allocas_p.
25240 * params.def (asan-instrument-allocas): Add new option.
25241 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25242 * opts.c (common_handle_option): Disable allocas sanitization for
25245 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25247 * asan.c: Include gimple-fold.h.
25248 (get_last_alloca_addr): New function.
25249 (handle_builtin_stackrestore): Likewise.
25250 (handle_builtin_alloca): Likewise.
25251 (asan_emit_allocas_unpoison): Likewise.
25252 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25253 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25254 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25255 (instrument_builtin_call): Pass gimple iterator to
25256 get_mem_refs_of_builtin_call.
25257 (last_alloca_addr): New global.
25258 * asan.h (asan_emit_allocas_unpoison): Declare.
25259 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25260 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25261 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25262 if function calls alloca.
25263 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25264 * gimple-fold.h (replace_call_with_value): Declare.
25265 * internal-fn.c: Include asan.h.
25266 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25267 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25269 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25271 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25272 (C_SELFTEST_FLAGS): New.
25273 (CPP_SELFTEST_FLAGS): New.
25274 (SELFTEST_DEPS): New, from deps of s-selftest.
25275 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25276 (CPP_SELFTEST_DEPS): New.
25277 (selftest): Add dependency on s-selftest-c++.
25278 (s-selftest): Rename to...
25279 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25280 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25281 than SELFTEST_FLAGS.
25282 (selftest-gdb): Rename to...
25283 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25285 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25286 (selftest-valgrind): Rename to...
25287 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25289 (selftest-valgrind): Reintroduce as an alias for
25290 selftest-c-valgrind.
25291 (s-selftest-c++): New.
25292 (selftest-c++-gdb): New.
25293 (selftest-c++-valgrind): New.
25295 2017-07-06 Olivier Hainque <hainque@adacore.com>
25297 * gcc.c (process_command): When deciding if undefined variables
25298 should be ignored when processing specs, accept "gcc -v" as well.
25300 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25302 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25303 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25305 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25307 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25308 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25309 * config/arm/arm-tables.opt: Regenerate.
25310 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25312 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25314 2017-07-06 Carl Love <cel@us.ibm.com>
25316 * ChangeLog: Clean up from mid air collision
25318 2017-07-06 Carl Love <cel@us.ibm.com>
25320 * config/rs6000/rs6000-c.c: Add support for built-in functions
25321 vector signed int vec_subc (vector signed int, vector signed int);
25322 vector signed __int128 vec_subc (vector signed __int128,
25323 vector signed __int128);
25324 vector unsigned __int128 vec_subc (vector unsigned __int128,
25325 vector unsigned __int128);
25326 vector signed int vec_sube (vector signed int, vector signed int,
25327 vector signed int);
25328 vector unsigned int vec_sube (vector unsigned int,
25329 vector unsigned int,
25330 vector unsigned int);
25331 vector signed __int128 vec_sube (vector signed __int128,
25332 vector signed __int128,
25333 vector signed__int128);
25334 vector unsigned __int128 vec_sube (vector unsigned __int128,
25335 vector unsigned __int128,
25336 vector unsigned __int128);
25337 vector signed int vec_subec (vector signed int, vector signed int,
25338 vector signed int);
25339 vector unsigned int vec_subec (vector unsigned int,
25340 vector unsigned int,
25341 vector unsigned int);
25342 vector signed __int128 vec_subec (vector signed __int128,
25343 vector signed __int128,
25344 vector signed__int128);
25345 vector unsigned __int128 vec_subec (vector unsigned __int128,
25346 vector unsigned __int128,
25347 vector unsigned __int128);
25348 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25349 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25350 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25351 BU_ALTIVEC_OVERLOAD_X definitions.
25352 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25353 * doc/extend.texi: Update the built-in documentation file for the new
25354 built-in functions.
25356 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25359 * diagnostic-show-locus.c (layout::layout): Use start and finish
25360 spelling location for the start and finish of each range.
25361 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25362 Add unused aspect param.
25363 * input.c (expand_location_1): Add "aspect" param, and use it
25364 to access the correct part of the location.
25365 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25367 (expand_location_to_spelling_point): Likewise.
25368 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25369 param, and pass it to expand_location_1.
25371 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25373 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25374 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25375 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25376 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25377 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25378 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25379 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25380 _mm_maskz_getmant_ss): New intrinsics.
25381 (__builtin_ia32_getexpss128_mask): Changed to ...
25382 __builtin_ia32_getexpss128_round ... this.
25383 (__builtin_ia32_getexpsd128_mask): Changed to ...
25384 __builtin_ia32_getexpsd128_round ... this.
25385 * config/i386/i386-builtin-types.def
25386 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25387 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25388 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25389 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25390 __builtin_ia32_getmantss_mask_round): New builtins.
25391 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25392 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25393 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25394 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25395 * config/i386/sse.md
25396 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25397 avx512f_sgetexp<mode><mask_scalar_name>
25398 <round_saeonly_scalar_name> ... this.
25399 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25400 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25401 vgetexp<ssescalarmodesuffix>
25402 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25403 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25404 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25405 avx512f_vgetmant<mode><mask_scalar_name>
25406 <round_saeonly_scalar_name> ... this.
25407 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25408 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25409 vgetmant<ssescalarmodesuffix>
25410 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25411 %0<mask_scalar_operand4>, %1, %2
25412 <round_saeonly_scalar_mask_op4>, %3} ... this.
25413 * config/i386/subst.md (mask_scalar_operand4,
25414 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25415 round_saeonly_scalar_nimm_predicate): New subst attributes.
25417 2017-07-06 Julia Koval <julia.koval@intel.com>
25419 * config/i386/i386.c (ix86_erase_embedded_rounding):
25420 Remove code for old rounding pattern.
25422 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25424 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25426 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25428 * doc/sourcebuild.texi (Test Directives, Variants of
25429 dg-require-support): Add documentation for dg-require-stack-check.
25431 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25433 * config/i386/subst.md (mask_scalar, round_scalar,
25434 round_saeonly_scalar): New meta-templates.
25435 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25436 round_scalar_mask_operand3, round_scalar_mask_op3,
25437 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25438 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25439 round_saeonly_scalar_constraint,
25440 round_saeonly_scalar_prefix): New subst attribute.
25441 * config/i386/sse.md
25442 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25443 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25444 <round_scalar_name> ... this.
25445 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25446 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25447 <round_scalar_name> ... this.
25448 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25449 <sse>_vm<code><mode>3<mask_scalar_name>
25450 <round_saeonly_scalar_name> ... this.
25451 (v<plusminus_mnemonic><ssescalarmodesuffix>
25452 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25453 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25454 v<plusminus_mnemonic><ssescalarmodesuffix>
25455 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25456 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25457 (v<multdiv_mnemonic><ssescalarmodesuffix>
25458 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25459 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25460 v<multdiv_mnemonic><ssescalarmodesuffix>
25461 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25462 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25463 (v<maxmin_float><ssescalarmodesuffix>
25464 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25465 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25466 v<maxmin_float><ssescalarmodesuffix>
25467 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25468 %0<mask_scalar_operand3>, %1, %<iptr>2
25469 <round_saeonly_scalar_mask_op3>} ... this.
25471 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25473 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25474 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25476 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25477 Alan Hayward <alan.hayward@arm.com>
25478 David Sherwood <david.sherwood@arm.com>
25480 * combine.c (simplify_if_then_else): Remove "enum" before
25482 * compare-elim.c (can_eliminate_compare): Likewise.
25483 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25485 (aarch64_lookup_simd_builtin_type): Likewise.
25486 (aarch64_simd_builtin_type): Likewise.
25487 (aarch64_init_simd_builtin_types): Likewise.
25488 (aarch64_simd_expand_args): Likewise.
25489 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25491 (aarch64_reverse_mask): Likewise.
25492 (aarch64_simd_emit_reg_reg_move): Likewise.
25493 (aarch64_gen_adjusted_ldpstp): Likewise.
25494 (aarch64_ccmp_mode_to_code): Likewise.
25495 (aarch64_operands_ok_for_ldpstp): Likewise.
25496 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25497 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25499 (aarch64_min_divisions_for_recip_mul): Likewise.
25500 (aarch64_reassociation_width): Likewise.
25501 (aarch64_get_condition_code_1): Likewise.
25502 (aarch64_simd_emit_reg_reg_move): Likewise.
25503 (aarch64_simd_attr_length_rglist): Likewise.
25504 (aarch64_reverse_mask): Likewise.
25505 (aarch64_operands_ok_for_ldpstp): Likewise.
25506 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25507 (aarch64_gen_adjusted_ldpstp): Likewise.
25508 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25510 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25511 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25512 (arm_lookup_simd_builtin_type): Likewise.
25513 (arm_simd_builtin_type): Likewise.
25514 (arm_init_simd_builtin_types): Likewise.
25515 (arm_expand_builtin_args): Likewise.
25516 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25517 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25518 (ft32_setup_incoming_varargs): Likewise.
25519 (ft32_function_arg): Likewise.
25520 (ft32_function_arg_advance): Likewise.
25521 (ft32_pass_by_reference): Likewise.
25522 (ft32_arg_partial_bytes): Likewise.
25523 (ft32_valid_pointer_mode): Likewise.
25524 (ft32_addr_space_pointer_mode): Likewise.
25525 (ft32_addr_space_legitimate_address_p): Likewise.
25526 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25528 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25529 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25530 (iamcu_alignment): Likewise.
25531 (canonicalize_vector_int_perm): Likewise.
25532 (ix86_noce_conversion_profitable_p): Likewise.
25533 (ix86_mpx_bound_mode): Likewise.
25534 (ix86_operands_ok_for_move_multiple): Likewise.
25535 * config/microblaze/microblaze-protos.h
25536 (microblaze_expand_conditional_branch_reg): Likewise.
25537 * config/microblaze/microblaze.c
25538 (microblaze_expand_conditional_branch_reg): Likewise.
25539 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25541 (rs6000_reassociation_width): Likewise.
25542 (rs6000_invalid_binary_op): Likewise.
25543 (fusion_p9_p): Likewise.
25544 (emit_fusion_p9_load): Likewise.
25545 (emit_fusion_p9_store): Likewise.
25546 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25548 (riscv_hard_regno_mode_ok_p): Likewise.
25549 (riscv_address_insns): Likewise.
25550 (riscv_split_symbol): Likewise.
25551 (riscv_legitimize_move): Likewise.
25552 (riscv_function_value): Likewise.
25553 (riscv_hard_regno_nregs): Likewise.
25554 (riscv_expand_builtin): Likewise.
25555 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25556 (riscv_build_integer): Likewise.
25557 (riscv_split_integer): Likewise.
25558 (riscv_legitimate_constant_p): Likewise.
25559 (riscv_cannot_force_const_mem): Likewise.
25560 (riscv_regno_mode_ok_for_base_p): Likewise.
25561 (riscv_valid_base_register_p): Likewise.
25562 (riscv_valid_offset_p): Likewise.
25563 (riscv_valid_lo_sum_p): Likewise.
25564 (riscv_classify_address): Likewise.
25565 (riscv_legitimate_address_p): Likewise.
25566 (riscv_address_insns): Likewise.
25567 (riscv_load_store_insns): Likewise.
25568 (riscv_force_binary): Likewise.
25569 (riscv_split_symbol): Likewise.
25570 (riscv_force_address): Likewise.
25571 (riscv_legitimize_address): Likewise.
25572 (riscv_move_integer): Likewise.
25573 (riscv_legitimize_const_move): Likewise.
25574 (riscv_legitimize_move): Likewise.
25575 (riscv_address_cost): Likewise.
25576 (riscv_subword): Likewise.
25577 (riscv_output_move): Likewise.
25578 (riscv_canonicalize_int_order_test): Likewise.
25579 (riscv_emit_int_order_test): Likewise.
25580 (riscv_function_arg_boundary): Likewise.
25581 (riscv_pass_mode_in_fpr_p): Likewise.
25582 (riscv_pass_fpr_single): Likewise.
25583 (riscv_pass_fpr_pair): Likewise.
25584 (riscv_get_arg_info): Likewise.
25585 (riscv_function_arg): Likewise.
25586 (riscv_function_arg_advance): Likewise.
25587 (riscv_arg_partial_bytes): Likewise.
25588 (riscv_function_value): Likewise.
25589 (riscv_pass_by_reference): Likewise.
25590 (riscv_setup_incoming_varargs): Likewise.
25591 (riscv_print_operand): Likewise.
25592 (riscv_elf_select_rtx_section): Likewise.
25593 (riscv_save_restore_reg): Likewise.
25594 (riscv_for_each_saved_reg): Likewise.
25595 (riscv_register_move_cost): Likewise.
25596 (riscv_hard_regno_mode_ok_p): Likewise.
25597 (riscv_hard_regno_nregs): Likewise.
25598 (riscv_class_max_nregs): Likewise.
25599 (riscv_memory_move_cost): Likewise.
25600 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25601 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25602 (rl78_addr_space_address_mode): Likewise.
25603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25606 (rs6000_reassociation_width): Likewise.
25607 (rs6000_invalid_binary_op): Likewise.
25608 (fusion_p9_p): Likewise.
25609 (emit_fusion_p9_load): Likewise.
25610 (emit_fusion_p9_store): Likewise.
25611 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25612 (ok_for_simple_move_operands): Likewise.
25613 (ok_for_simple_move_strict_operands): Likewise.
25614 (ok_for_simple_arith_logic_operands): Likewise.
25615 (visium_legitimize_reload_address): Likewise.
25616 (visium_select_cc_mode): Likewise.
25617 (output_cbranch): Likewise.
25618 (visium_split_double_move): Likewise.
25619 (visium_expand_copysign): Likewise.
25620 (visium_expand_int_cstore): Likewise.
25621 (visium_expand_fp_cstore): Likewise.
25622 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25623 (visium_function_arg): Likewise.
25624 (visium_function_arg_advance): Likewise.
25625 (visium_libcall_value): Likewise.
25626 (visium_setup_incoming_varargs): Likewise.
25627 (visium_legitimate_constant_p): Likewise.
25628 (visium_legitimate_address_p): Likewise.
25629 (visium_legitimize_address): Likewise.
25630 (visium_secondary_reload): Likewise.
25631 (visium_register_move_cost): Likewise.
25632 (visium_memory_move_cost): Likewise.
25633 (prepare_move_operands): Likewise.
25634 (ok_for_simple_move_operands): Likewise.
25635 (ok_for_simple_move_strict_operands): Likewise.
25636 (ok_for_simple_arith_logic_operands): Likewise.
25637 (visium_function_value_1): Likewise.
25638 (rtx_ok_for_offset_p): Likewise.
25639 (visium_legitimize_reload_address): Likewise.
25640 (visium_split_double_move): Likewise.
25641 (visium_expand_copysign): Likewise.
25642 (visium_expand_int_cstore): Likewise.
25643 (visium_expand_fp_cstore): Likewise.
25644 (visium_split_cstore): Likewise.
25645 (visium_select_cc_mode): Likewise.
25646 (visium_split_cbranch): Likewise.
25647 (output_cbranch): Likewise.
25648 (visium_print_operand_address): Likewise.
25649 * expmed.c (flip_storage_order): Likewise.
25650 * expmed.h (emit_cstore): Likewise.
25651 (flip_storage_order): Likewise.
25652 * genrecog.c (validate_pattern): Likewise.
25653 * hsa-gen.c (gen_hsa_addr): Likewise.
25654 * internal-fn.c (expand_arith_overflow): Likewise.
25655 * ira-color.c (allocno_copy_cost_saving): Likewise.
25656 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25657 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25658 (process_invariant_for_inheritance): Likewise.
25659 * lra-eliminations.c (move_plus_up): Likewise.
25660 * omp-low.c (lower_oacc_reductions): Likewise.
25661 * simplify-rtx.c (simplify_subreg): Likewise.
25662 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25663 (TARGET_CHKP_BOUND_MODE): Likewise..
25664 * targhooks.c (default_chkp_bound_mode): Likewise.
25665 (default_setup_incoming_vararg_bounds): Likewise.
25666 * targhooks.h (default_chkp_bound_mode): Likewise.
25667 (default_setup_incoming_vararg_bounds): Likewise.
25668 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25669 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25670 (have_whole_vector_shift): Likewise.
25671 * tree-vect-stmts.c (vectorizable_load): Likewise.
25672 * doc/tm.texi: Regenerate.
25674 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25676 Graceful degrade if Binutils PR21472 is not available.
25679 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25680 .rodata in flash test fails.
25681 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25682 * confgure: Regenerate.
25683 * config.in: Regenerate.
25684 * config/avr/avr.c (avr_asm_named_section)
25685 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25686 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25687 (avr_asm_init_sections): Same.
25689 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25691 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25692 (fma<VH:mode>4_intrinsic): Likewise.
25693 (*fmsub<VCVTF:mode>4): Likewise.
25694 (*fmsub<VH:mode>4_intrinsic): Likewise.
25696 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25699 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25700 Don't depend on "optimize > 0".
25701 (out_movhi_r_mr, out_movqi_mr_r): Same.
25702 (out_movhi_mr_r, out_movqi_r_mr): Same.
25703 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25704 io_address_operand on "optimize > 0".
25706 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25708 * tree-loop-distribution.c: Add general explanantion on the pass.
25709 (generate_loops_for_partition): Mark distributed loop.
25710 (pg_add_dependence_edges): New parameter. Handle alias data
25711 dependence specially and record it in the parameter if asked.
25712 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25713 (init_partition_graph_vertices, add_partition_graph_edge): New.
25714 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25715 (free_partition_graph_vdata, build_partition_graph): New.
25716 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25717 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25718 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25719 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25720 (version_for_distribution_p, finalize_partitions): New.
25721 (distribute_loop): Handle alias data dependence specially. Factor
25722 out loop fusion code as functions and call these functions.
25724 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25726 * tree-loop-distribution.c (classify_partition): New parameter and
25727 better handle reduction statement.
25728 (rdg_build_partitions): Revise comment.
25729 (distribute_loop): Compute statements in all partitions and pass it
25730 to classify_partition.
25732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25734 * tree-loop-distribution.c (enum partition_type): New.
25735 (struct partition): New field type.
25736 (partition_merge_into): Add parameter. Update partition type.
25737 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25738 (build_rdg_partition_for_vertex): Compute partition type.
25739 (rdg_build_partitions): Dump partition type.
25740 (distribute_loop): Update calls to partition_merge_into.
25742 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25744 * tree-loop-distribution.c (struct ddr_hasher): New.
25745 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25747 (classify_partition): Call get_data_dependence.
25748 (pg_add_dependence_edges): Ditto.
25749 (distribute_loop): Release data dependence hash table.
25751 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25753 * tree-loop-distribution.c (ref_base_address): Delete.
25754 (similar_memory_accesses): Rename ...
25755 (share_memory_accesses): ... to this. Check if partitions access
25756 the same memory reference.
25757 (distribute_loop): Call share_memory_accesses.
25759 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25761 * tree-loop-distribution.c (struct partition): New field recording
25762 its data reference.
25763 (partition_alloc, partition_free): Init and release data refs.
25764 (partition_merge_into): Merge data refs.
25765 (build_rdg_partition_for_vertex): Collect data refs for partition.
25766 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25768 (distribute_loop): Remve data refs from vertice data of partition
25771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25773 * tree-loop-distribution.c (params.h): Include header file.
25774 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25775 (datarefs_vec): New global var.
25776 (create_rdg_vertices): Use datarefs_vec directly.
25777 (free_rdg): Don't free data references.
25778 (build_rdg): Update use. Don't free data references.
25779 (distribute_loop): Compute global variable for data references.
25780 Bail out if there are too many data references.
25782 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25784 * tree-loop-distribution.c (loop_nest): New global var.
25785 (build_rdg): Use loop directly, rather than loop nest.
25786 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25788 (distribute_loop): Compute global variable loop nest. Update use.
25790 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25792 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25793 (partition_merge_into): New parameter. Dump reason for fusion.
25794 (distribute_loop): Update use of partition_merge_into.
25796 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25798 * tree-loop-distribution.c (bb_top_order_index): New.
25799 (bb_top_order_index_size, bb_top_order_cmp): New.
25800 (stmts_from_loop): Use topological order.
25801 (pass_loop_distribution::execute): Compute and release topological
25802 order for basic blocks.
25804 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25806 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25809 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25811 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25812 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25813 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25814 * internal-fn.def (LOOP_DIST_ALIAS): New.
25815 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25816 (fold_loop_internal_call): ... this.
25817 (vect_loop_dist_alias_call): New function.
25818 (set_uid_loop_bbs): Call fold_loop_internal_call.
25819 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25822 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25825 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25826 Require dead FLAGS_REG at the beginning of a peephole.
25828 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25831 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25832 arguments in the call to __builtin_ia32_sbb_u32.
25833 (_subborrow_u64): Swap _X and _Y arguments in the call to
25834 __builtin_ia32_sbb_u64.
25836 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25839 * tree-vrp.c (compare_assert_loc): Turn into a function template
25840 with stable template parameter. Only test if a->e is NULL,
25841 !a->e == !b->e has been verified already. Use e == NULL or
25842 e != NULL instead of e or ! e tests. If stable is true, don't use
25843 iterative_hash_expr, on the other side allow a or b or both NULL
25844 and sort the NULLs last.
25845 (process_assert_insertions): Sort using compare_assert_loc<false>
25846 instead of compare_assert_loc, later sort using
25847 compare_assert_loc<true> before calling process_assert_insertions_for
25848 in a loop. Use break instead of continue once seen NULL pointer.
25850 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25852 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25853 Cortex-R7 and Cortex-R8 processors.
25855 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25857 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25858 uninitialized while src is not.
25860 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25862 * common/config/arm/arm-common.c: Adjust include path for
25864 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25865 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25866 (arm-cpu-data.h): Likewise.
25867 (arm-cpu-cdata.h): Likewise.
25868 * config/arm/arm-cpu.h: Delete.
25869 * config/arm/arm-cpu-cdata.h: Delete.
25870 * config/arm/arm-cpu-data.h: Delete.
25872 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25874 * config/arm/arm-cpus.in (cortex-a55): New.
25875 (cortex-a75): Likewise.
25876 (cortex-a75.cortex-a55): Likewise.
25877 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25879 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25880 * config/arm/arm-cpu-cdata.h: Regenerate.
25881 * config/arm/arm-cpu-data.h: Regenerate.
25882 * config/arm/arm-cpu.h: Regenerate.
25883 * config/arm/arm-tables.opt: Regenerate.
25884 * config/arm/arm-tune.md: Regenerate.
25886 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25888 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25890 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25892 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25895 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25897 PR tree-optimization/81292
25898 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25899 full_string_p, also call adjust_related_strinfos if the adjustment
25900 is simple, otherwise invalidate related strinfos.
25902 2017-07-04 Martin Liska <mliska@suse.cz>
25905 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25906 newly created variable as DECL_IGNORED_P.
25908 2017-07-04 Martin Liska <mliska@suse.cz>
25911 * ipa-inline.c (inline_small_functions):
25912 Use xstrdup_for_dump.
25914 2017-07-04 Tom de Vries <tom@codesourcery.com>
25916 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25918 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25921 * config/darwin.c (darwin_function_switched_text_sections):
25924 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25926 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25928 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25930 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25932 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25934 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25935 min_profitable_iters, and th as inclusive lower bounds.
25936 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25937 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25938 for min_profitable_iters and min_profitable_estimate.
25939 (vect_transform_loop): Treat th as an inclusive lower bound.
25940 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25942 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25945 * config/darwin.c (darwin_function_switched_text_sections):
25946 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25947 in two pieces, and suppress the use of buf.
25949 2017-07-03 Nathan Sidwell <nathan@acm.org>
25951 * hash-table.h (hash_table_mod1): Fix indentation.
25953 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25955 PR middle-end/81290
25956 * predict.c (force_edge_cold): Be more careful about propagation
25958 * profile-count.h (profile_probability::guessed,
25959 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25961 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25963 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25965 * doc/invoke.texi (rcpc architecture extension): Document it.
25967 2017-07-03 Richard Biener <rguenther@suse.de>
25969 PR tree-optimization/60510
25970 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25971 the scalar reduction PHI and use it.
25972 (vectorizable_reduction): Properly guard the single_defuse_cycle
25973 path for non-SLP reduction chains where we cannot use it.
25974 Rework reduc_def/index and vector type deduction. Rework
25975 vector operand gathering during reduction op code-gen.
25976 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25977 chains dissolve the chain and leave it to non-SLP reduction
25980 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25982 * tree-data-ref.h (dr_alignment): Declare.
25983 * tree-data-ref.c (dr_alignment): New function.
25984 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25985 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25987 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25989 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25991 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25992 and base_misalignment fields.
25993 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25994 * tree-data-ref.c: Include builtins.h.
25995 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25996 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25997 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25998 * tree-vect-data-refs.c: Include tree-cfg.h.
25999 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26000 fields instead of calculating an alignment here.
26001 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26002 innermost_loop_behavior fields.
26004 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26006 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26008 (DR_STEP_ALIGNMENT): New macro.
26009 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26010 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26011 (create_data_ref): Print it.
26012 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26013 to tell whether the step preserves vector (mis)alignment.
26014 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26015 Move the check for an integer step and generalise to all INTEGER_CST.
26016 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26017 Print the outer step alignment.
26019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26021 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26022 with offset_alignment.
26023 (DR_ALIGNED_TO): Delete.
26024 (DR_OFFSET_ALIGNMENT): New macro.
26025 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26026 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26027 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26028 (create_data_ref): Likewise.
26029 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26030 (vect_analyze_data_refs): Likewise.
26031 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26032 creating dummy innermost behavior.
26034 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26036 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26037 with a "innermost_loop_behavior *" and refeence tree.
26038 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26039 (create_data_ref): Update call accordingly.
26040 * tree-predcom.c (find_looparound_phi): Likewise.
26042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26044 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26045 fields with dr_wrt_vec_loop.
26046 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26047 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26048 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26049 (vect_dr_behavior): New function.
26050 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26051 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26052 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26053 track whether the step preserves the misalignment.
26054 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26055 Use vect_dr_behavior.
26056 (vect_setup_realignment): Update call accordingly.
26057 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26058 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26059 call to vect_create_addr_base_for_vector_ref.
26060 (vect_create_cond_for_align_checks): Likewise.
26061 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26062 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26063 (vect_recog_mask_conversion_pattern): Likewise.
26064 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26065 (new_stmt_vec_info): Remove redundant zeroing.
26067 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26069 * common/config/arm/arm-common.c (arm_be8_option): New function.
26070 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26071 (ISA_ARMv6): Add isa_bit_be8.
26072 * config/arm/arm.h (arm_be8_option): Add prototype.
26073 (BE8_SPEC_FUNCTION): New define.
26074 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26075 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26076 (mlittle-endian): Similarly.
26077 (mbe8, mbe32): New options.
26078 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26079 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26081 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26083 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26085 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26087 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26088 (cleanup_tree_cfg_bb): Use it.
26089 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26091 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26093 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26096 * loop-doloop.c (add_test): Update profile.
26098 2017-07-03 Martin Liska <mliska@suse.cz>
26101 * sanopt.c (rewrite_usage_of_param): New function.
26102 (sanitize_rewrite_addressable_params): Likewise.
26103 (pass_sanopt::execute): Call rewrite_usage_of_param.
26105 2017-07-03 Richard Biener <rguenther@suse.de>
26107 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26108 back to using VIEW_CONVERT_EXPR.
26110 2017-07-03 Martin Liska <mliska@suse.cz>
26113 * doc/extend.texi: Document when a resolver function is
26114 generated for target_clones.
26116 2017-07-03 Martin Liska <mliska@suse.cz>
26118 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26119 and shadow memory of auto variables which are subject of
26120 use-after-scope sanitization.
26121 (asan_expand_mark_ifn): Add do set only when is_poison.
26123 2016-07-03 Richard Biener <rguenther@suse.de>
26125 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26127 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26128 (vectorizable_reduction): Perform reduction PHI creation when
26129 visiting a reduction PHI and adjust and simplify code generation
26130 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26131 (vect_transform_loop): Visit reduction PHIs.
26132 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26133 defs into the SLP tree.
26134 (vect_build_slp_tree): Reduction defs terminate the recursion.
26135 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26137 (vect_get_vec_defs_for_stmt_copy): Export.
26138 (vect_get_vec_defs): Likewise.
26139 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26141 (vect_get_vec_defs_for_stmt_copy): Declare.
26142 (vect_get_vec_defs): Likewise.
26144 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26146 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26147 parameter with a "loop" parameter and use it instead of the
26148 loop containing DR_STMT. Don't check simple_iv when doing
26149 BB analysis. Describe the two analysis modes in the comment.
26151 2017-07-03 Tom de Vries <tom@codesourcery.com>
26153 PR tree-optimization/69468
26154 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26155 (find_same_succ_bb): Handle ignore_edge_flags.
26157 2017-07-03 Tom de Vries <tom@codesourcery.com>
26159 PR tree-optimization/81192
26160 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26162 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26164 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26166 2017-07-03 Tom de Vries <tom@codesourcery.com>
26168 PR tree-optimization/81192
26169 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26170 BB_SAME_SUCC (bb) == NULL.
26172 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26174 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26177 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26179 * dumpfile.c: Include profile-count.h
26180 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26182 (insert_cond_bb): Update profile.
26183 * tree-cfg.h (insert_cond_bb): Update prototype.
26184 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26185 * tree-dump.c: Do not include tree-cfg.
26187 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26189 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26191 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26193 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26196 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26198 * tree-complex.c (expand_complex_div_wide): update profile.
26200 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26201 Alan Hayward <alan.hayward@arm.com>
26202 David Sherwood <david.sherwood@arm.com>
26204 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26205 (CORETYPES_H): New define.
26206 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26207 (insn-modes-inline.h, s-modes-inline-h): New rules.
26208 (generated_files): Add insn-modes-inline.h.
26209 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26210 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26211 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26212 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26213 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26214 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26215 (build/gencodes.o, build/genconditions.o): Likewise.
26216 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26217 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26218 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26219 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26220 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26221 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26222 * coretypes.h: Include everything up to real.h for generators.
26223 Include insn-modes.h first. Include wide-int-print.h after
26224 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26225 * machmode.h: Don't include insn-modes.h here.
26226 * function-tests.c: Remove includes of signop.h, machmode.h,
26227 double-int.h and wide-int.h.
26229 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26231 * optc-save-gen.awk: Likewise.
26232 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26233 * godump.c: Remove include of wide-int-print.h.
26234 * pretty-print.h: Likewise.
26235 * wide-int-print.cc: Likewise.
26236 * wide-int.cc: Likewise.
26237 * hash-map-tests.c: Remove include of signop.h.
26238 * hash-set-tests.c: Likewise.
26239 * rtl-tests.c: Likewise.
26240 * mkconfig.sh: Remove include of machmode.h.
26241 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26243 (emit_insn_modes_inline_h): ...this new function. Emit the code
26244 into an insn-modes-inline.h header file, adding appropriate
26245 include guards and end comments.
26246 (emit_insn_modes_c_header): Remove include of machmode.h.
26247 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26249 (main): Handle -i flag and call emit_insn_modes_inline_h when
26252 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26254 * tree-ssa-strlen.c (strinfo): Rename the length field to
26255 nonzero_chars. Add a full_string_p field.
26256 (compare_nonzero_chars, zero_length_string_p): New functions.
26257 (get_addr_stridx): Add an offset_out parameter.
26258 Use compare_nonzero_chars.
26259 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26260 (new_strinfo): Update after above changes to strinfo.
26261 (set_endptr_and_length): Set full_string_p.
26262 (get_string_length): Update after above changes to strinfo.
26263 (unshare_strinfo): Update call to new_strinfo.
26264 (maybe_invalidate): Likewise.
26265 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26266 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26267 as a uhwi instead of an shwi. Update after above changes to
26268 strinfo and new_strinfo.
26269 (zero_length_string): Assert that chainsi contains full strings.
26270 Use zero_length_string_p. Update call to new_strinfo.
26271 (adjust_related_strinfos): Update after above changes to strinfo.
26272 Copy full_string_p from origsi.
26273 (adjust_last_stmt): Use zero_length_string_p.
26274 (handle_builtin_strlen): Update after above changes to strinfo and
26275 new_strinfo. Install the lhs as the string length if the previous
26276 entry didn't describe a full string.
26277 (handle_builtin_strchr): Update after above changes to strinfo
26279 (handle_builtin_strcpy): Likewise.
26280 (handle_builtin_strcat): Likewise.
26281 (handle_builtin_malloc): Likewise.
26282 (handle_pointer_plus): Likewise.
26283 (handle_builtin_memcpy): Likewise. Track nonzero characters
26284 that aren't necessarily followed by a nul terminator.
26285 (handle_char_store): Likewise.
26287 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26289 PR tree-optimization/80769
26290 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26291 for malloc and calloc. Document the new invariant that all related
26292 strinfos have delayed lengths or none do.
26293 (verify_related_strinfos): Move earlier in file.
26294 (set_endptr_and_length): New function, split out from...
26295 (get_string_length): ...here. Also set the lengths of related
26297 (zero_length_string): Assert that chainsi has known (rather than
26299 (adjust_related_strinfos): Likewise.
26301 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26303 PR tree-optimization/81136
26304 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26305 assert that two references with the same misalignment have the same
26306 compile-time misalignment if those compile-time misalignments
26309 2017-07-01 Andi Kleen <ak@linux.intel.com>
26311 * print-tree.c (print_node): Print all attributes.
26313 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26315 * cfg.c (scale_bbs_frequencies): New function.
26316 * cfg.h (scale_bbs_frequencies): Declare it.
26317 * cfgloopanal.c (single_likely_exit): Cleanup.
26318 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26320 (scale_loop_profile): Likewise.
26321 (loop_version): Likewise.
26322 (create_empty_loop_on_edge): Update.
26323 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26324 scale_loop_frequencies, scale_loop_profile, loopify,
26325 loop_version): Update prototypes.
26326 * modulo-sched.c (sms_schedule): Update.
26327 * predict.c (unlikely_executed_edge_p): Also check probability.
26328 (probably_never_executed_edge_p): Fix typo.
26329 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26330 * tree-parloops.c (gen_parallel_loop): Update.
26331 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26333 * tree-ssa-loop-split.c (split_loop): Update.
26334 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26335 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26336 (vect_loop_versioning): Update.
26337 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26339 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26341 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26343 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26345 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26346 to keep profile consistent.
26348 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26350 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26351 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26352 * profile-count.h (max_safe_multiplier): Make unsigned.
26353 (profile_count::guessed_zero): New.
26355 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26357 * bb-reorder.c (fix_up_crossing_landing_pad,
26358 fix_crossing_conditional_branches): Use make_single_succ_edge
26359 to keep profile consistent.
26361 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26363 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26366 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26369 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26370 the right scopes, make sure cond_jump isn't preserved between multiple
26371 iterations. Search for fallthru edge whenever there are 3+ edges and
26372 use find_fallthru_edge for it.
26374 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26376 Patch by Alexander Monakov <amonakov@ispras.ru>
26377 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26378 probabilities consistently.
26380 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26382 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26383 emit_cmp_and_jump_insns.
26385 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26388 * tree-inline.c (expand_call_inline): Combine profile statuses.
26390 2017-06-30 Andrew Pinski <apinski@cavium.com>
26392 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26393 fold_stmt returned true.
26395 2017-06-30 Nathan Sidwell <nathan@acm.org>
26397 * ggc.h (empty_string): Delete.
26398 * cfgexpand.c (expand_asm_stmt): Use plain "".
26399 * optabs.c (expand_asm_memory_barrier): Likewise.
26400 * stringpool.c (empty_string): Delete.
26401 (digit_vector, digit_string): Delete.
26402 (ggc_alloc_string): Use plain "", don't optimize single digit
26403 strings. Use ggc_alloc_atomic.
26405 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26407 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26408 comparison set and one other set, use the cost of the non-comparison
26411 2017-06-30 Nathan Sidwell <nathan@acm.org>
26413 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26416 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26418 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26419 loops. Remove now unneeded calls to gimple_switch_set_label() that
26420 just set removed labels to NULL_TREE.
26422 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26424 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26425 (set_range_info): ...here. Only call set_range_info_raw if domain
26427 (set_nonzero_bits): Call set_range_info_raw.
26428 * tree-ssanames.h (set_range_info_raw): New.
26430 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26433 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26434 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26435 of nonimmediate_operand and <store_mask_constraint> instead of m
26436 for the input operand. For V8FI iterator, always split if input
26437 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26438 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26439 <store_mask_predicate> instead of register_operand and
26440 <store_mask_constraint> instead of v for the input operand. Make
26441 sure both operands aren't MEMs for if not <mask_applied>.
26443 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26445 * lto-wrapper.c (copy_file) Close both file descriptors before
26448 2017-06-30 Martin Liska <mliska@suse.cz>
26451 * multiple_target.c (create_dispatcher_calls): Make ifunc
26452 also for function that don't have calls or are not referenced.
26454 2017-06-30 Richard Biener <rguenther@suse.de>
26456 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26457 analyze the first scalar stmt. Move vector type computation
26458 for the BB case here from ...
26459 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26460 live operation processing in the SLP case properly.
26462 2017-06-30 Richard Biener <rguenther@suse.de>
26464 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26466 2017-06-30 Martin Liska <mliska@suse.cz>
26469 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26470 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26472 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26474 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26477 2017-06-30 Richard Biener <rguenther@suse.de>
26479 PR tree-optimization/81249
26480 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26481 condition reduction result to original scalar type.
26483 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26485 * profile-count.h (enum profile_quality): Fix typos and whitespace
26488 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26490 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26491 type for branch probabilities.
26493 2017-06-29 Julian Brown <julian@codesourcery.com>
26494 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26496 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26497 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26498 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26499 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26501 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26503 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26504 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26505 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26506 CC usage from generic code to here.
26507 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26508 CC usage into the target macros.
26510 2017-06-29 Maya Rashish <coypu@sdf.org>
26512 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26515 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26517 * arm/arm-builtins.c: Include profile-count.h
26518 * except.c (sjlj_emit_function_enter): Use
26519 profile_probability::unlikely.
26521 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26523 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26524 and tocrel_offset be pointer args rather than implicitly using
26526 (legitimate_constant_pool_address_p, rs6000_emit_move,
26527 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26528 tocrel_offset and use in toc_relative_expr_p call.
26529 (print_operand, print_operand_address): Use static tocrel_base_oac
26530 and tocrel_offset_oac.
26531 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26534 2017-06-29 Maya Rashish <coypu@sdf.org>
26536 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26538 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26540 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26541 objects, take into account only the alignment of 'op0' and 'mode1' if
26544 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26546 * ccmp.c (ccmp_tree_comparison_p): New function.
26547 (ccmp_candidate_p): Update to use above function.
26548 (get_compare_parts): New function.
26549 (expand_ccmp_next): Update to use new functions.
26550 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26552 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26553 take mode as argument.
26554 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26555 * expr.c (expand_expr_real_1): Pass mode as argument.
26557 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26559 * combine.c (combine_instructions): Print insns to dump_file, together
26562 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26564 * asan.c (asan_emit_stack_protection): Update.
26565 (create_cond_insert_point): Update.
26566 * auto-profile.c (afdo_propagate_circuit): Update.
26567 * basic-block.h (struct edge_def): Turn probability to
26568 profile_probability.
26569 (EDGE_FREQUENCY): Update.
26570 * bb-reorder.c (find_traces_1_round): Update.
26571 (better_edge_p): Update.
26572 (sanitize_hot_paths): Update.
26573 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26574 (make_single_succ_edge): Update.
26575 (check_bb_profile): Update.
26576 (dump_edge_info): Update.
26577 (update_bb_profile_for_threading): Update.
26578 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26579 probabilitycount to 0.
26580 * cfgbuild.c (compute_outgoing_frequencies): Update.
26581 * cfgcleanup.c (try_forward_edges): Update.
26582 (outgoing_edges_match): Update.
26583 (try_crossjump_to_edge): Update.
26584 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26585 (expand_gimple_tailcall): Update.
26586 (construct_init_block): Use make_single_succ_edge.
26587 (construct_exit_block): Use make_single_succ_edge.
26588 * cfghooks.c (verify_flow_info): Update.
26589 (redirect_edge_succ_nodup): Update.
26590 (split_edge): Update.
26591 (account_profile_record): Update.
26592 * cfgloopanal.c (single_likely_exit): Update.
26593 * cfgloopmanip.c (scale_loop_profile): Update.
26594 (set_zero_probability): Remove.
26595 (duplicate_loop_to_header_edge): Update.
26596 * cfgloopmanip.h (loop_version): Update prototype.
26597 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26598 (force_nonfallthru_and_redirect): Update.
26599 (update_br_prob_note): Update.
26600 (rtl_verify_edges): Update.
26601 (purge_dead_edges): Update.
26602 (rtl_lv_add_condition_to_bb): Update.
26603 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26604 * cgraphunit.c (init_lowered_empty_function): Update.
26605 (cgraph_node::expand_thunk): Update.
26606 * cilk-common.c: Include profile-count.h
26607 * dojump.c (inv): Remove.
26608 (jumpifnot): Update.
26609 (jumpifnot_1): Update.
26610 (do_jump_1): Update.
26612 (do_jump_by_parts_greater_rtx): Update.
26613 (do_compare_rtx_and_jump): Update.
26614 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26615 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26616 * dwarf2cfi.c: Include profile-count.h
26617 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26618 (sjlj_emit_dispatch_table): Likewise.
26619 * explow.c: Include profile-count.h
26620 * expmed.c (emit_store_flag_force): Update.
26621 (do_cmp_and_jump): Update.
26622 * expr.c (compare_by_pieces_d::generate): Update.
26623 (compare_by_pieces_d::finish_mode): Update.
26624 (emit_block_move_via_loop): Update.
26625 (store_expr_with_bounds): Update.
26626 (store_constructor): Update.
26627 (expand_expr_real_2): Update.
26628 (expand_expr_real_1): Update.
26629 * expr.h (try_casesi, try_tablejump): Update prototypes.
26630 * gimple-pretty-print.c (dump_probability): Update.
26631 (dump_profile): New.
26632 (dump_gimple_label): Update.
26633 (dump_gimple_bb_header): Update.
26634 * graph.c (draw_cfg_node_succ_edges): Update.
26635 * hsa-gen.c (convert_switch_statements): Update.
26636 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26637 (find_if_case_1): Update.
26638 (find_if_case_2): Update.
26639 * internal-fn.c (expand_arith_overflow_result_store): Update.
26640 (expand_addsub_overflow): Update.
26641 (expand_neg_overflow): Update.
26642 (expand_mul_overflow): Update.
26643 (expand_vector_ubsan_overflow): Update.
26644 * ipa-cp.c (good_cloning_opportunity_p): Update.
26645 * ipa-split.c (split_function): Use make_single_succ_edge.
26646 * ipa-utils.c (ipa_merge_profiles): Update.
26647 * loop-doloop.c (add_test): Update.
26648 (doloop_modify): Update.
26649 * loop-unroll.c (compare_and_jump_seq): Update.
26650 (unroll_loop_runtime_iterations): Update.
26651 * lra-constraints.c (lra_inheritance): Update.
26652 * lto-streamer-in.c (input_cfg): Update.
26653 * lto-streamer-out.c (output_cfg): Update.
26654 * mcf.c (adjust_cfg_counts): Update.
26655 * modulo-sched.c (sms_schedule): Update.
26656 * omp-expand.c (expand_omp_for_init_counts): Update.
26657 (extract_omp_for_update_vars): Update.
26658 (expand_omp_ordered_sink): Update.
26659 (expand_omp_for_ordered_loops): Update.
26660 (expand_omp_for_generic): Update.
26661 (expand_omp_for_static_nochunk): Update.
26662 (expand_omp_for_static_chunk): Update.
26663 (expand_cilk_for): Update.
26664 (expand_omp_simd): Update.
26665 (expand_omp_taskloop_for_outer): Update.
26666 (expand_omp_taskloop_for_inner): Update.
26667 * omp-simd-clone.c (simd_clone_adjust): Update.
26668 * optabs.c (expand_doubleword_shift): Update.
26669 (expand_abs): Update.
26670 (emit_cmp_and_jump_insn_1): Update.
26671 (expand_compare_and_swap_loop): Update.
26672 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26673 * predict.c (predictable_edge_p): Update.
26674 (edge_probability_reliable_p): Update.
26675 (set_even_probabilities): Update.
26676 (combine_predictions_for_insn): Update.
26677 (combine_predictions_for_bb): Update.
26678 (propagate_freq): Update.
26679 (estimate_bb_frequencies): Update.
26680 (force_edge_cold): Update.
26681 * profile-count.c (profile_count::dump): Add missing space into dump.
26682 (profile_count::debug): Add newline.
26683 (profile_count::differs_from_p): Explicitly convert to unsigned.
26684 (profile_count::stream_in): Update.
26685 (profile_probability::dump): New member function.
26686 (profile_probability::debug): New member function.
26687 (profile_probability::differs_from_p): New member function.
26688 (profile_probability::differs_lot_from_p): New member function.
26689 (profile_probability::stream_in): New member function.
26690 (profile_probability::stream_out): New member function.
26691 * profile-count.h (profile_count_quality): Rename to ...
26692 (profile_quality): ... this one.
26693 (profile_probability): New.
26694 (profile_count): Update.
26695 * profile.c (compute_branch_probabilities): Update.
26696 * recog.c (peep2_attempt): Update.
26697 * sched-ebb.c (schedule_ebbs): Update.
26698 * sched-rgn.c (find_single_block_region): Update.
26699 (compute_dom_prob_ps): Update.
26700 (schedule_region): Update.
26701 * sel-sched-ir.c (compute_succs_info): Update.
26702 * stmt.c (struct case_node): Update.
26703 (do_jump_if_equal): Update.
26704 (get_outgoing_edge_probs): Update.
26705 (conditional_probability): Update.
26706 (emit_case_dispatch_table): Update.
26707 (expand_case): Update.
26708 (expand_sjlj_dispatch_table): Update.
26709 (emit_case_nodes): Update.
26710 * targhooks.c: Update.
26711 * tracer.c (better_p): Update.
26712 (find_best_successor): Update.
26713 * trans-mem.c (expand_transaction): Update.
26714 * tree-call-cdce.c: Update.
26715 * tree-cfg.c (gimple_split_edge): Upate.
26716 (move_sese_region_to_fn): Upate.
26717 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26718 * tree-eh.c (lower_resx): Upate.
26719 (cleanup_empty_eh_move_lp): Upate.
26720 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26721 * tree-inline.c (copy_edges_for_bb): Update.
26722 (copy_cfg_body): Update.
26723 * tree-parloops.c (gen_parallel_loop): Update.
26724 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26725 (gimple_gen_time_profiler): Update.
26726 * tree-ssa-dce.c (remove_dead_stmt): Update.
26727 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26728 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26729 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26730 (unloop_loops): Update.
26731 (try_peel_loop): Update.
26732 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26733 * tree-ssa-loop-split.c (connect_loops): Update.
26734 (split_loop): Update.
26735 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26736 (hoist_guard): Update.
26737 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26738 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26739 (value_replacement): Update.
26740 * tree-ssa-reassoc.c (branch_fixup): Update.
26741 * tree-ssa-tail-merge.c (replace_block_by): Update.
26742 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26743 (create_edge_and_update_destination_phis): Update.
26744 (compute_path_counts): Update.
26745 (recompute_probabilities): Update.
26746 (update_joiner_offpath_counts): Update.
26747 (freqs_to_counts_path): Update.
26748 (duplicate_thread_path): Update.
26749 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26750 (struct switch_conv_info): Update.
26751 (gen_inbound_check): Update.
26752 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26753 (vect_do_peeling): Update.
26754 (vect_loop_versioning): Update.
26755 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26756 (optimize_mask_stores): Update.
26757 * ubsan.c (ubsan_expand_null_ifn): Update.
26758 * value-prof.c (gimple_divmod_fixed_value): Update.
26759 (gimple_divmod_fixed_value_transform): Update.
26760 (gimple_mod_pow2): Update.
26761 (gimple_mod_pow2_value_transform): Update.
26762 (gimple_mod_subtract): Update.
26763 (gimple_mod_subtract_transform): Update.
26764 (gimple_ic): Update.
26765 (gimple_stringop_fixed_value): Update.
26766 (gimple_stringops_transform): Update.
26767 * value-prof.h: Update.
26769 2017-06-29 Carl Love <cel@us.ibm.com>
26771 * config/rs6000/rs6000-c.c: Add support for built-in functions
26772 vector signed int vec_signed (vector float);
26773 vector signed long long vec_signed (vector double);
26774 vector signed int vec_signed2 (vector double, vector double);
26775 vector signed int vec_signede (vector double);
26776 vector signed int vec_signedo (vector double);
26777 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26778 instruction generator.
26779 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26780 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26781 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26783 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26784 vunsignede_v2df): Add define_expands.
26785 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26786 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26787 VEC_UNSIGNEDO): Add definitions.
26788 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26789 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26790 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26791 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26792 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26793 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26794 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26795 vec_unsignede, vec_unsignedo): Add builtin defines.
26796 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26798 * doc/extend.texi: Update the built-in documentation file for the
26799 new built-in functions.
26801 2017-06-29 Richard Biener <rguenther@suse.de>
26803 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26804 reduction chains to LOOP_VINFO_REDUCTIONS.
26805 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26806 SLP reductions after processing reduction chains.
26808 2017-06-29 Nathan Sidwell <nathan@acm.org>
26810 * builtins.c (fold_builtin_FUNCTION): Use
26811 lang_hooks.decl_printable_name.
26813 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26815 PR middle-end/81194
26816 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26817 with only one label.
26818 * stmt.c (expand_case): Assert NCASES is greater than one.
26820 2017-06-29 Richard Biener <rguenther@suse.de>
26822 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26824 (group_case_labels): Likewise.
26825 (find_taken_edge): Push sanity checking on val to workers...
26826 (find_taken_edge_cond_expr): ... here
26827 (find_taken_edge_switch_expr): ... and here, handle cases
26828 with just a default label.
26829 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26830 (group_case_labels): Likewise.
26831 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26832 group_case_labels does anything cleanup the CFG again.
26834 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26836 PR tree-optimization/81196
26837 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26838 exit condition comparing two IVs.
26840 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26842 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26843 profile to the dummy entry at the end of the list of architectures.
26844 * config/arm/arm-cpu-cdata.h: Regenerated.
26846 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26847 Michael Collison <michael.collison@arm.com>
26850 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26852 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26853 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26854 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26855 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26856 of shift when the shift amount is masked with constant equal to
26857 the size of the mode.
26858 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26861 2017-06-29 Martin Liska <mliska@suse.cz>
26863 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26865 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26868 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26871 (edge_growth_cache_entry::edge_growth_cache_entry): New
26873 (reset_edge_growth_cache): Update to use constructor.
26875 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26877 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26878 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26879 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26881 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26883 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26884 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26886 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26888 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26889 (*-linux-uclibc*): Add t-uclibc tmake_file.
26890 * config/t-musl: New.
26891 * config/t-uclibc: New.
26893 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26895 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26897 (gen_comm_data): Emit architectural setting of arch_prof.
26898 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26900 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26901 (armv8-m.base, armv8-m.main): Likewise.
26902 * arm-protos.h (arm_build_target): Add profile field.
26903 (arch_option): Likewise.
26904 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26906 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26907 arm_active_target.profile.
26909 2017-06-28 Richard Biener <rguenther@suse.de>
26911 PR middle-end/81227
26912 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26913 TYPE_OVERFLOW_WRAPS.
26914 * match.pd (negate_expr_p): Likewise.
26915 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26916 fold_build2, not fold_binary.
26918 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26920 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26921 Convert memory address to Pmode.
26922 (aarch64_print_operand): Assert MEM operands are always Pmode.
26924 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26927 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26928 Remove redundant if.
26929 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26930 * config/arm/aarch-common-protos.h
26931 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26932 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26934 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26937 * multiple_target.c (create_dispatcher_calls): Set the default
26938 clone to be static, not public.
26940 2017-06-28 Richard Biener <rguenther@suse.de>
26942 * tree-vect-loop.c (vectorizable_reduction): Move special
26943 cond reduction IV var creation ...
26944 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26945 parameter. Use STMT_VINFO_VECTYPE.
26946 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26949 2017-06-28 Martin Liska <mliska@suse.cz>
26952 * ipa-visibility.c (non_local_p): Handle visibility.
26954 2017-06-28 Martin Liska <mliska@suse.cz>
26957 * common.opt: Add IntegerRange to various options.
26958 * opt-functions.awk (integer_range_info): New function.
26959 * optc-gen.awk: Add integer_range_info to cl_options struct.
26960 * opts-common.c (decode_cmdline_option): Handle
26961 CL_ERR_INT_RANGE_ARG.
26962 (cmdline_handle_error): Likewise.
26963 * opts.c (print_filtered_help): Show valid interval in
26964 when --help is provided.
26965 * opts.h (struct cl_option): Add range_min and range_max fields.
26966 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26968 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26970 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26971 (x * C EQ/NE y * C): New transformation.
26973 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26975 * genmultilib (combination_space): Accept '+' in option names.
26977 2017-06-28 Martin Liska <mliska@suse.cz>
26980 * asan.c (instrument_derefs): Bail out inner references
26981 that are hard register variables.
26983 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26986 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26987 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26989 2017-06-28 Richard Biener <rguenther@suse.de>
26991 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26992 (vect_get_slp_defs): Adjust.
26993 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26995 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26997 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26998 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26999 (vectorizable_reduction): Adjust.
27000 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27002 (vect_get_slp_defs): Likewise.
27003 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27004 (vectorizable_bswap): Adjust.
27005 (vectorizable_call): Likewise.
27006 (vectorizable_conversion): Likewise.
27007 (vectorizable_assignment): Likewise.
27008 (vectorizable_shift): Likewise.
27009 (vectorizable_operation): Likewise.
27010 (vectorizable_store): Likewise.
27011 (vectorizable_condition): Likewise.
27012 (vectorizable_comparison): Likewise.
27014 2017-06-28 Michael Collison <michael.collison@arm.com>
27017 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27019 (arm_gen_movmemqi): Removed unused variable 'i'.
27020 Convert 'for' loop into 'while' loop.
27021 (arm_expand_prologue): Remove last unnecessary set of insn.
27022 (thumb_pop): Remove unused variable 'pushed_words'.
27023 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27025 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27027 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27028 * config/s390/s390-protos.h: Add prototype of
27029 s390_rel_address_ok_p.
27030 * config/s390/s390.c (s390_got_symbol): New function.
27031 (s390_rel_address_ok_p): New function.
27032 (legitimize_pic_address): Use s390_rel_address_ok_p.
27033 (s390_load_got): Use s390_got_symbol.
27034 (s390_option_override): Issue error if
27035 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27036 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27038 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27040 2017-06-27 Andrew Pinski <apinski@cavium.com>
27042 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27043 (X * copysign (1.0, X)): New pattern.
27044 (X * copysign (1.0, -X)): New pattern.
27045 (copysign (-1.0, CST)): New pattern.
27047 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27049 * genmultilib (combination_space): Remove variable.
27050 Validate reuse rules against regular expression for any sequence
27051 of multilib options in any order.
27053 2017-06-27 Michael Collison <michael.collison@arm.com>
27055 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27056 call aarch64_split_simd_combine.
27057 * (aarch64_combine_internal<mode>): Delete pattern.
27058 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27059 Allow register and subreg operands.
27061 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27063 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27064 specific need, just fallback on defaults.
27065 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27067 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27068 Olivier Hainque <hainque@adacore.com>
27070 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27072 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27073 targets. Pick a default if no particular attempt applied.
27074 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27077 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27079 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27080 (x86_64-wrs-vxworks7): Likewise.
27082 2017-06-27 Marek Polacek <polacek@redhat.com>
27085 * ubsan.c (instrument_null): Check get_base_address's result for null.
27087 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27089 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27091 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27093 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27094 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27095 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27096 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27097 New function types.
27098 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27099 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27100 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27101 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27102 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27103 BUILT_IN_FEUPDATEENV): New builtins.
27104 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27105 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27106 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27107 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27109 (builtin_structptr_types): Adjust size.
27110 * tree.c (builtin_structptr_types): Add four entries.
27112 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27113 Olivier Hainque <hainque@adacore.com>
27115 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27116 (TLS_SYM): New local macro, forcing reference to __tls__ on
27117 link command lines for VxWorks 7 RTPs, triggering initialization
27119 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27120 OS features TLS support, true for RTPs on VxWorks 7.
27121 * config/vxworks.c (vxworks_override_options): Setup emutls
27124 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27126 * predict.c (test_prediction_value_range): Use -1U instead of -1
27127 to avoid narrowing conversion warning.
27128 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27129 to avoid narrowing conversion warning.
27130 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27132 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27134 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27136 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27137 64bit configurations.
27138 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27139 (SIZE_TYPE): Likewise.
27140 * config/vxworks.c (vxworks_emutls_var_fields): Use
27141 long_unsigned_type_node instead of unsigned_type_node as the offset
27142 field type, which is "pointer" mode in emutls.c.
27144 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27147 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27149 PR middle-end/81207
27150 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27151 gimple_vuse copying separately from gimple_vdef copying.
27153 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27155 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27158 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27159 Olivier Hainque <hainque@adacore.com>
27161 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27162 for all vxworks7 targets.
27163 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27164 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27165 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27166 variations for VX6/VX7 and 32/64bits later on in ...
27167 (VXWORKS_LIB_SPEC): Leverage new macros.
27168 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27169 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27171 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27173 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27174 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27176 2017-06-26 Carl Love <cel@us.ibm.com>
27178 * config/rs6000/rs6000-c.c: Add support for built-in functions
27179 vector bool char vec_reve (vector bool char);
27180 vector signed char vec_reve (vector signed char);
27181 vector unsigned char vec_reve (vector unsigned char);
27182 vector bool int vec_reve (vector bool int);
27183 vector signed int vec_reve (vector signed int);
27184 vector unsigned int vec_reve (vector unsigned int);
27185 vector bool long long vec_reve (vector bool long long);
27186 vector signed long long vec_reve (vector signed long long);
27187 vector unsigned long long vec_reve (vector unsigned long long);
27188 vector bool short vec_reve (vector bool short);
27189 vector signed short vec_reve (vector signed short);
27190 vector double vec_reve (vector double);
27191 vector float vec_reve (vector float);
27192 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27193 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27194 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27195 (altivec_vreve): New pattern.
27196 * config/rs6000/altivec.h (vec_reve): New define.
27197 * doc/extend.texi (vec_rev): Update the built-in documentation file
27198 for the new built-in functions.
27200 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27202 PR tree-optimization/71815
27203 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27205 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27207 (slsr_process_phi): Likewise.
27208 (replace_uncond_cands_and_profitable_phis): Don't replace a
27209 multiply candidate with a stride of 1 (copy or cast).
27210 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27212 (lowest_cost_path): Likewise.
27213 (total_savings): Likewise.
27215 2017-06-26 Richard Biener <rguenther@suse.de>
27218 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27219 Use def_builtin_pure for all gather builtins.
27221 2017-06-26 Richard Biener <rguenther@suse.de>
27223 PR tree-optimization/81203
27224 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27225 non-dominating BBs.
27227 2017-06-26 Marek Polacek <polacek@redhat.com>
27230 * doc/invoke.texi: Document -Wmultistatement-macros.
27232 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27234 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27235 arm_neon_ok_no_float_abi effective target.
27237 2017-06-26 Richard Biener <rguenther@suse.de>
27239 PR tree-optimization/80928
27240 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27241 (copy_bbs): Set BB_DUPLICATED flag early.
27242 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27244 (execute_on_shrinking_pred): Likewise.
27245 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27246 BB_DUPLICATED blocks.
27247 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27248 iterate over all PHIs considering removal of *gsi.
27250 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27252 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27255 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27257 * config/rs6000/rs6000-string.c: (expand_block_clear,
27258 do_load_for_compare, select_block_compare_mode,
27259 compute_current_alignment, expand_block_compare,
27260 expand_strncmp_align_check, expand_strn_compare,
27261 expand_block_move, rs6000_output_load_multiple)
27262 Move functions related to string/block move/compare
27263 to a separate file.
27264 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27265 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27266 for this function which is now used in two files.
27267 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27268 * config.gcc: Add rs6000-string.o to extra_objs for
27269 targets powerpc*-*-* and rs6000*-*-*.
27271 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27274 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27275 32-bit, since indexed is not valid for DImode.
27276 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27277 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27278 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27279 (define_peephole2 for Altivec d-form store): Likewise.
27282 * multiple_target.c (create_dispatcher_calls): Only create the
27283 dispatcher call if the function is the default clone of a
27284 versioned function.
27286 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27288 PR middle-end/80902
27289 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27290 a call, force the call to not be a tail call.
27292 2017-06-23 Jeff Law <law@redhat.com>
27294 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27297 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27299 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27300 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27301 handling for early expansion of vector shifts (sl,sr,sra,rl).
27302 (builtin_function_type): Add vector shift right instructions
27303 to the unsigned argument list.
27305 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27307 rtl-optimizatoin/79286
27308 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27309 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27310 trap. PIC register plus a const unspec without offset can never trap.
27312 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27314 * tree.h (builtin_structptr_type): New type.
27315 (builtin_structptr_types): Declare new array.
27316 * tree.c (builtin_structptr_types): New array.
27317 (free_lang_data, build_common_tree_nodes): Use it.
27319 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27322 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27325 2017-06-22 Matt Turner <mattst88@gmail.com>
27327 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27328 Lake models to skylake case. Assume skylake for unknown
27329 models with clflushopt.
27331 2017-06-22 Jeff Law <law@redhat.com>
27333 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27334 frame sizes that do not satisfy aarch64_uimm12_shift.
27336 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27338 * profile-count.h (apply_probability,
27339 apply_scale, probability_in): Fix checks for zero.
27341 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27343 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27344 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27346 2016-06-22 Richard Biener <rguenther@suse.de>
27348 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27349 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27350 REDUC_MAX_EXPR support.
27351 (vectorizable_reduction): Likewise.
27352 (vect_create_epilog_for_reduction): Likewise.
27354 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27356 * match.pd (A / (1 << B) -> A >> B): New.
27357 * generic-match-head.c: Include optabs-tree.h.
27358 * gimple-match-head.c: Likewise.
27359 * optabs-tree.h (target_supports_op_p): New.
27360 * optabs-tree.c (target_supports_op_p): New.
27362 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27364 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27365 $gcc_cv_ld --help output.
27366 (gcc_cv_ld_demangle): Likewise.
27367 (gcc_cv_ld_eh_frame_hdr): Likewise.
27368 (gcc_cv_ld_pie): Likewise.
27369 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27370 (gcc_cv_ld_buildid): Likewise.
27371 (gcc_cv_ld_sysroot): Likewise.
27372 (ld_bndplt_support): Likewise.
27373 (ld_pushpopstate_support): Likewise.
27374 * configure: Regenerate.
27375 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27377 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27380 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27381 operands indexes to avoid gap between operands and match_dups.
27383 2017-06-21 Andrew Pinski <apinski@cavium.com>
27385 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27386 Increment Arith_shift and Arith_shift_reg by 1.
27387 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27389 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27390 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27391 (aarch64_strip_extend): Add new argument and test for it.
27392 (aarch64_cheap_mult_shift_p): New function.
27393 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27394 add a cost if it is true.
27395 Update calls to aarch64_strip_extend.
27396 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27398 2017-06-21 Andrew Pinski <apinski@cavium.com>
27400 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27402 (thunderxt88): Likewise.
27403 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27404 (thunderx_prefetch_tune): New variable.
27405 (thunderx2t99_prefetch_tune): Update for the correct values.
27406 (thunderxt88_tunings): New variable.
27407 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27408 generic_prefetch_tune.
27409 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27411 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27413 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27414 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27415 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27416 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27417 (aarch64_atomic_cas<mode>, GPI): Likewise.
27419 2017-06-21 Martin Liska <mliska@suse.cz>
27421 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27422 statements on cold and hot labels.
27423 * predict.c (tree_estimate_probability_bb): Remove the
27424 prediction from this place.
27426 2017-06-21 Martin Liska <mliska@suse.cz>
27428 PR tree-optimization/79489
27429 * gimplify.c (maybe_add_early_return_predict_stmt): New
27431 (gimplify_return_expr): Call the function.
27432 * predict.c (tree_estimate_probability_bb): Remove handling
27434 * predict.def: Update comment about early return predictor.
27435 * gimple-predict.h (is_gimple_predict): New function.
27436 * predict.def: Change default value of early return to 66.
27437 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27439 * passes.def: Put pass_strip_predict_hints to the beginning of
27442 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27444 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27445 FUNCTION_DECL declarations.
27446 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27448 (dwaf2out_decl): Likewise.
27449 * godump.c (go_early_global_decl): Skip call to the real debug hook
27450 for FUNCTION_DECL declarations.
27451 * passes.c (rest_of_decl_compilation): Skip call to the
27452 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27453 -fdump-go-spec is passed.
27455 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27457 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27458 Reorder for compactness.
27459 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27460 (def_builtin_pure, def_builtin_pure2): New functions.
27461 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27463 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27465 * match.pd (nop_convert): New predicate.
27466 ((A +- CST1) +- CST2): Allow some NOP conversions.
27468 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27471 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27472 with ctors/dtors if GOVD_SHARED is set.
27474 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27476 * config/aarch64/aarch64.md (movti_aarch64):
27477 Emit mov rather than orr.
27478 (movtf_aarch64): Likewise.
27479 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27480 Emit mov rather than orr.
27482 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27484 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27485 Swap alternatives, make integer dup more expensive.
27487 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27489 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27490 Return true for non-tls symbols.
27492 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27494 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27495 (cortex-a75): Likewise.
27496 (cortex-a75.cortex-a55): Likewise.
27497 * config/aarch64/aarch64-tune.md: Regenerate.
27498 * doc/invoke.texi (-mtune): Document new values for -mtune.
27500 2017-06-21 Tom de Vries <tom@codesourcery.com>
27502 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27503 stack_size feature.
27504 (Effective-Target Keywords, Other attributes): Suggest using
27505 dg-add-options stack_size feature to get stack limit in stack_size
27506 effective target documentation.
27508 2017-06-21 Julian Brown <julian@codesourcery.com>
27509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27511 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27512 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27513 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27515 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27516 attribute type list for neon_multiply.
27517 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27518 attribute type list for neon_multiply.
27519 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27520 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27521 attribute type list for neon_multiply.
27522 * config/arm/types.md (crypto_pmull): Add.
27523 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27524 attribute type list.
27526 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27528 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27531 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27533 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27534 to make sure not to dereference a NULL cost_classes_ptr pointer.
27536 2017-06-20 Carl Love <cel@us.ibm.com>
27538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27539 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27540 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27541 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27542 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27543 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27544 VMULOSW): New enum "unspec" values.
27545 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27546 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27547 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27548 altivec_vmulosw): New patterns.
27549 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27550 VMULOSW): Add definitions.
27552 2017-06-20 Julia Koval <julia.koval@intel.com>
27554 * config/i386/i386.c: Fix rounding expand for new pattern.
27555 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27557 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27559 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27560 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27562 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27564 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27567 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27569 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27570 architecture, then by implementer ID.
27571 * config/aarch64/aarch64-tune.md: Regenerate.
27573 2017-06-20 Richard Biener <rguenther@suse.de>
27575 PR middle-end/81097
27576 * fold-const.c (split_tree): Fold to type before negating.
27578 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27580 * diagnostic-show-locus.c
27581 (selftest::test_fixit_deletion_affecting_newline): New function.
27582 (selftest::diagnostic_show_locus_c_tests): Call it.
27584 2017-06-20 Andreas Schwab <schwab@suse.de>
27587 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27590 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27592 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27593 __ARM_FEATURE_COPROC according to support.
27595 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27597 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27598 Rewritten to avoid overflow for > 32-bit pointers.
27601 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27602 by removing enum keyword.
27603 (ubsan_type_descriptor): Likewise. Formatting fix.
27606 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27607 splitter): Require TARGET_SSE2 in the condition.
27609 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27612 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27613 for doing vector set of SFmode on ISA 3.0.
27614 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27615 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27617 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27618 SFmode value into a V4SF variable that was extracted from another
27619 V4SF variable without converting the element to double precision
27620 and back to single precision vector format.
27621 (vsx_insert_extract_v4sf_p9_2): Likewise.
27623 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27625 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27626 in UWHI to avoid undefined overflow.
27629 * ubsan.h (enum ubsan_encode_value_phase): New.
27630 (ubsan_encode_value): Change second argument to
27631 enum ubsan_encode_value_phase with default value of
27632 UBSAN_ENCODE_VALUE_GENERIC.
27633 * ubsan.c (ubsan_encode_value): Change second argument to
27634 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27635 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27636 create_tmp_var_raw instead of create_tmp_var and use a
27638 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27639 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27640 ubsan_encode_value callers.
27643 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27644 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27645 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27647 2017-06-19 Richard Biener <rguenther@suse.de>
27649 PR middle-end/81118
27650 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27651 estimates if we changed anything.
27653 2017-06-19 Richard Biener <rguenther@suse.de>
27655 PR tree-optimization/80887
27656 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27657 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27658 simplified lookups, then reset mprts_hook.
27659 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27661 (try_to_simplify): Likewise.
27663 2017-06-19 Martin Liska <mliska@suse.cz>
27666 * gimplify.c (gimplify_switch_expr):
27667 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27669 2017-06-19 Martin Liska <mliska@suse.cz>
27671 * doc/install.texi: Document that PGO runs in 4 stages.
27673 2017-06-19 Martin Liska <mliska@suse.cz>
27676 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27677 to dispatcher function name.
27678 * multiple_target.c (replace_function_decl): New function.
27679 (create_dispatcher_calls): Redirect both edges and references.
27681 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27683 * profile-count.c (profile_count::dump): Dump quality.
27684 (profile_count::differs_from_p): Update for unsigned val.
27685 * profile-count.h (profile_count_quality): New enum.
27686 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27688 2017-06-19 Richard Biener <rguenther@suse.de>
27690 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27691 struct function as arg.
27692 (estimate_numbers_of_iterations): Export overload with loop arg.
27693 (free_numbers_of_iterations_estimates_loop): Use an overload of
27694 free_numbers_of_iterations_estimates instead.
27695 * tree-cfg.c (remove_bb): Adjust.
27696 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27697 * tree-parloops.c (gen_parallel_loop): Likewise.
27698 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27700 (tree_unroll_loops_completely): Likewise.
27701 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27702 Use an overload instead and export.
27703 (estimated_loop_iterations): Adjust.
27704 (max_loop_iterations): Likewise.
27705 (likely_max_loop_iterations): Likewise.
27706 (estimate_numbers_of_iterations): Take struct function as arg
27708 (loop_exits_before_overflow): Adjust.
27709 (free_numbers_of_iterations_estimates_loop): Use an overload.
27710 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27711 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27713 2017-06-19 Richard Biener <rguenther@suse.de>
27716 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27717 RANGE_EXPR conservatively.
27719 2017-06-16 Carl Love <cel@us.ibm.com>
27721 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27722 definitions for vec_float, vec_float2, vec_floato,
27723 vec_floate built-ins.
27724 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27725 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27727 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27728 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27729 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27730 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27731 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27732 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27733 vec_floato): Add builtin defines.
27734 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27735 Update the built-in documentation file for the new built-in
27738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27740 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27741 (mthumb): Mark as the negative of -marm.
27743 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27745 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27747 (ARM Options, -mtune): Document that this accepts the same
27748 extension options as -mcpu.
27749 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27753 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27754 permitted extensions.
27756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27758 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27759 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27760 (armv8-m.main): Add option +nodsp.
27761 * config/arm/arm-cpu-cdata.h: Regenerated.
27763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27765 * config/arm/t-fuchsia: New file.
27766 * config.gcc (arm*-*-fuchsia*): Use it.
27768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27770 * config/arm/t-symbian: Rewrite for new option infrastructure.
27772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27774 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27775 (MULTILIB_REQUIRED): Likewise.
27777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27779 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27780 (MULTILIB_RESUE): Likewise.
27781 (MULTILIB_MATCHES): Likewise.
27782 (MULTLIB_REQUIRED): Likewise.
27784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27786 * config/arm/t-rtems: Rewrite for new option framework.
27788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27790 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27791 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27792 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27793 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27794 * config/arm/t-multilib: ... here.
27795 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27796 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27797 armv7-a and armv8*-a when A-profile libraries have not been built.
27798 * config/arm/t-rmprofile: Rewrite.
27800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27802 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27803 with a backslash. Remove the backslash after substituting unescaped
27805 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27809 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27810 the driver to pass through the default setting of -mfloat-abi.
27811 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27813 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27814 (all_feat_combs): New rule.
27815 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27817 * config/arm/t-aprofile: Rewrite.
27819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27821 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27822 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27823 fpu is not specified by the user/command-line.
27824 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27825 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27826 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27827 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27828 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27829 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27833 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27834 * config/arm/t-arm-elf: Rewritten.
27836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27838 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27839 have some floating-point instructions.
27840 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27841 (TARGET_MAYBE_HARD_FLOAT): New macro.
27842 * config/arm/arm-builtins.c (arm_init_builtins): Use
27843 TARGET_MAYBE_HARD_FLOAT.
27844 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27848 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27849 (configargs.h): Include it.
27850 (arm_print_hint_for_fpu_option): New function.
27851 (arm_parse_fpu_option): New function.
27852 (candidate_extension): New class.
27853 (arm_canon_for_multilib): New function.
27854 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27855 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27856 (ARCH_CANONICAL_SPECS): New macro.
27857 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27859 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27861 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27862 are set after handling multilib fragments. Set target_cpu_default2
27865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27867 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27869 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27870 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27872 (check_arch): Likewise.
27873 * config/arm/arm.c (arm_configure_build_target): Handle
27874 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27875 options in the default.
27877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27879 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27880 when an option is an alias of another.
27881 * config/arm/parsecpu.awk (optalias): New parser token.
27882 (gen_comm_data): Mark non-alias options as such. Emit entries
27883 for extension aliases.
27884 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27885 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27886 (armv6kz, armv6zk, armv6t2): Likewise.
27887 (armv7): Make vfpv3-d16 an alias.
27888 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27890 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27891 Sort in canonical order.
27892 (armv8-a): Sort in canonical order.
27893 (armv8.1-a, armv8.2-a): Likewise.
27894 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27896 (cortex-a9): Sort in canonical order.
27897 * config/arm/arm.c (selftests.h): Include it.
27898 (arm_test_cpu_arch_data): New function.
27899 (arm_run_self_tests): New function.
27900 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27901 (targetm): Move declaration to the end of the file.
27902 * arm-cpu-cdata.h: Regenerated.
27904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27906 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27907 call to target_mode_check describing the type of option passed.
27908 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27909 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27910 arm_parse_cpu_option_name to match parameters against list of
27912 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27913 arm_arch_core_flags data structure.
27914 * config/arm/arm-cpu_cdata.h: Regenerated.
27916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27918 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27920 (arm_print_hint_for_cpu_option): Likewise.
27921 (arm_print_hint_for_arch_option): Likewise.
27922 (arm_parse_cpu_option_name): Likewise.
27923 (arm_parse_arch_option_name): Likewise.
27924 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27925 of entries in the all_fpus list.
27926 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27927 (arm_parse_cpu_option_name): Declare.
27928 (arm_parse_arch_option_name): Declare.
27929 (arm_parse_option_features): Declare.
27930 (arm_intialize_isa): Declare.
27931 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27933 (gen_comm_data): ... here. Make definitions non-static.
27934 * config/arm/arm-cpu-data.h: Regenerated.
27935 * config/arm/arm-cpu-cdata.h: Regenerated.
27937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27939 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27940 (cpu_arch_extension): New structure.
27941 (cpu_arch_option, arch_option, cpu_option): New structures.
27942 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27943 architecture types.
27944 (gen_data): Generate new format data tables.
27945 * config/arm/arm.c (cpu_tune): New structure.
27946 (cpu_option, processors): Delete.
27947 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27948 (arm_print_hint_for_cpu_option): ... this and ...
27949 (arm_print_hint_for_arch_option): ... this.
27950 (arm_parse_arch_cpu_name): Delete. Replace with ...
27951 (arm_parse_cpu_option_name): ... this and ...
27952 (arm_parse_arch_option_name): ... this.
27953 (arm_unrecognized_feature): Change type of target parameter to
27955 (arm_parse_arch_cpu_features): Delete. Replace with ...
27956 (arm_parse_option_features): ... this.
27957 (arm_configure_build_target): Rework to use new configuration data
27959 (arm_print_tune_info): Rework for new configuration data tables.
27960 * config/arm/arm-cpu-data.h: Regenerated.
27961 * config/arm/arm-cpu.h: Regenerated.
27963 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27965 * Makefile.in (OBJS): Move sbitmap.o from here ...
27966 (OBJS-libcommon): ... to here.
27968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27970 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27971 (ISA_ALL_CRYPTO): New macro.
27972 (ISA_ALL_SIMD): New macro
27973 (ISA_ALL_FP): New macro.
27974 * config/arm/arm.c (fpu_bitlist): Update initializer.
27975 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27977 (arm9e): Add fpu. Add option for nofp
27978 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27979 (arm926ej-s, arm1026ej-s): Likewise.
27980 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27981 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27982 neon-fp16, neon-vfpv4, nofp and nosimd.
27983 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27984 (cortex-a8): Add fpu. Add option for nofp.
27985 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27986 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27987 (cortex-r4f): Add fpu.
27988 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27989 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27991 (cortex-r8): Likewise.
27992 (cortex-m4): Add fpu. Add option for nofp.
27993 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27994 (cortex-a17.cortex-a7): Likewise.
27995 (cortex-a32): Add fpu. Add options for crypto and nofp.
27996 (cortex-a35, cortex-a53): Likewise.
27997 (cortex-a57): Add fpu. Add option for crypto.
27998 (cortex-a72, cortex-a73): Likewise.
27999 (exynos-m1): Likewise.
28000 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28001 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28002 (cortex-m33): Add fpu. Add option for nofp.
28003 * config/arm/arm-cpu-cdata.h: Regenerated
28004 * config/arm/arm-cpu-data.h: Regenerated.
28006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28008 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28009 (armv5te, armv5tej): Likewise.
28010 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28011 (armv7): Add options fp and vfpv3-d16.
28012 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28013 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28015 (armv7ve): Likewise.
28016 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28017 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28018 (armv8-a): Add nocrypto option.
28019 (armv8.1-a, armv8.2-a): Likewise.
28020 (armv8-m.main): add options fp, fp.dp and nofp.
28022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28024 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28026 (armv8-a+crc): Delete.
28027 (armv8.1-a): Add options simd, crypto and nofp.
28028 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28029 (armv8.2-a+fp16): Delete.
28030 (armv8-m.main): Add option dsp.
28031 (armv8-m.main+dsp): Delete.
28032 (cortex-a8): Add fpu. Add nofp option.
28033 (cortex-a9): Add fpu. Add nofp and nosimd options.
28034 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28035 link to main cpu and architecture data structures.
28036 (gen_comm_data): Only put isa attributes from the main architecture
28038 (option): New statement for architecture and CPU entries.
28039 * arm.c (struct cpu_option): New structure.
28040 (struct processors): Add entry for options.
28041 (arm_unrecognized_feature): New function.
28042 (arm_parse_arch_cpu_name): Ignore any characters after the first
28044 (arm_parse_arch_cpu_feature): New function.
28045 (arm_configure_build_target): Separate out any CPU and architecture
28046 features and parse separately. Don't error out if -mfpu=auto is
28047 used with only an architecture string.
28048 (arm_print_asm_arch_directives): New function.
28049 (arm_file_start): Call it.
28050 * config/arm/arm-cpu-cdata.h: Regenerated.
28051 * config/arm/arm-cpu-data.h: Likewise.
28052 * config/arm/arm-tables.opt: Likewise.
28054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28056 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28057 assembler when it is not -mfpu=auto.
28059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28061 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28062 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28063 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28064 (ASM_CPU_SPEC): Rewrite.
28065 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28066 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28067 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28068 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28069 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28070 copied string is NUL-terminated. Also strip any characters prefixed
28072 (arm_rewrite_selected_arch): New function.
28073 (arm_rewrite_march): New function.
28075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28077 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28078 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28079 (march, mcpu, mtune): Convert to string-based options.
28080 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28081 (arm_parse_arch_cpu_name): New function.
28082 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28083 identify selected architecture or CPU.
28084 (arm_option_save): New function.
28085 (TARGET_OPTION_SAVE): Redefine.
28086 (arm_option_restore): Restore string options.
28087 (arm_option_print): Print string options.
28089 2017-06-16 Martin Sebor <msebor@redhat.com>
28091 PR tree-optimization/80933
28092 PR tree-optimization/80934
28093 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28094 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28095 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28096 (gimple_fold_builtin): Call them.
28098 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28100 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28101 as unlikely; update profile.
28103 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28105 * predict.c (force_edge_cold): Handle declaring edges impossible
28108 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28110 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28112 (try_unroll_loop_completely): Fix reporting.
28114 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28116 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28118 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28121 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28122 if given a non-constant argument for an intrinsic which requires a
28125 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28127 * profile.c (compare_freqs): New function.
28128 (branch_prob): Sort edge list.
28129 (find_spanning_tree): Assume that the list is priority sorted.
28131 2017-06-16 Richard Biener <rguenther@suse.de>
28133 PR tree-optimization/81090
28134 * passes.def (pass_record_bounds): Remove.
28135 * tree-pass.h (make_pass_record_bounds): Likewise.
28136 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28137 make_pass_record_bounds): Likewise.
28138 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28139 not free niter estimates at the beginning but at the end.
28140 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28142 2017-06-16 Richard Biener <rguenther@suse.de>
28144 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28145 initializer to workaround ICE in host GCC 4.8.
28147 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28149 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28151 (clone_inlined_nodes): Update.
28153 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28155 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28156 prefetch settings, and enable prefetching by default at -O3.
28158 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28160 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28161 Set flag_prefetch_loop_arrays according to tuning data.
28163 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28165 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28166 New tune structure.
28167 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28168 [Unrelated to main purpose of the patch] Place the pointer field last
28169 to enable type checking errors when tune structure are wrongly merged.
28170 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28171 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28172 (thunderx2t99_prefetch_tune): New tune constants.
28173 (tune_params *_tunings): Update all tunings (no functional change).
28174 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28175 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28176 from tunings structures.
28178 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28181 * ubsan.c (instrument_null): Add T argument, use it instead
28182 of computing it based on IS_LHS.
28183 (instrument_object_size): Likewise.
28184 (pass_ubsan::execute): Adjust instrument_null and
28185 instrument_object_size callers to pass gimple_get_lhs or
28186 gimple_assign_rhs1 result to it. Use instrument_null instead of
28187 calling get_base_address and instrument_mem_ref. Handle
28188 aggregate call arguments for object-size sanitization.
28190 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28192 PR tree-optimization/81089
28193 * tree-vrp.c (is_masked_range_test): Validate operands of
28196 2017-06-15 Martin Sebor <msebor@redhat.com>
28199 * dumpfile.c (dump_register): Avoid calling memset to initialize
28200 a class with a default ctor.
28201 * gcc.c (struct compiler): Remove const qualification.
28202 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28203 * hash-table.h: Ditto.
28204 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28206 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28207 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28209 * params.h (struct param_info): Make struct members non-const.
28210 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28211 with default initialization.
28212 * vec.h (vec_copy_construct, vec_default_construct): New helper
28214 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28215 with vec_copy_construct.
28216 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28217 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28218 * doc/invoke.texi (-Wclass-memaccess): Document.
28220 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28222 * emit-rtl.h (is_leaf): Update comment about local
28223 register allocator.
28225 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28228 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28229 for a variable to have a section before checking if the section has a
28231 Set section to.persistent if persistent attribute is set.
28232 Warn if .persistent attribute is used on an automatic variable.
28234 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28236 PR rtl-optimization/80474
28237 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28239 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28241 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28244 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28246 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28247 (casesi): Emit bounds checking as RTL.
28248 (casesi_internal_mips16_<mode>): Remove bounds checking.
28250 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28252 * config/xtensa/xtensa.c (xtensa_option_override): Append
28253 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28254 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28255 xtensa_doloop_hooks): Define unconditionally.
28256 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28258 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28259 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28260 for it in xtensa_option_override.
28261 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28262 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28264 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28266 * doc/cppopts.texi: Document '-' special value to -MF.
28268 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28270 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28271 (cortex_a53_fconst): Likewise.
28272 (cortex_a53_fpmul): Likewise.
28273 (cortex_a53_f_load_64): Likewise.
28274 (cortex_a53_f_load_many): Likewise.
28275 (cortex_a53_advsimd_alu): Likewise.
28276 (cortex_a53_advsimd_alu_q): Likewise.
28277 (cortex_a53_advsimd_mul): Likewise.
28278 (cortex_a53_advsimd_mul_q): Likewise.
28279 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28280 Add missing fmul, r2f_cvt and fconst cases.
28282 2017-06-14 Richard Biener <rguenther@suse.de>
28284 PR middle-end/81088
28285 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28287 (fold_binary_loc): When associating do not treat pre-existing
28288 TREE_OVERFLOW on literal constants as a reason to allow
28289 TREE_OVERFLOW on associated literal constants.
28291 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28293 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28294 (MASK_FEATURES): New macro.
28295 * config/sparc/sparc.c (sparc_option_override): Remove the special
28296 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28298 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28300 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28301 a division of 0 if non-call exceptions are enabled.
28303 2017-06-14 Andrew Pinski <apinski@cavium.com>
28304 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28307 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28308 Improve vector initialization code gen for only variable case.
28310 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28312 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28314 2017-06-14 Richard Biener <rguenther@suse.de>
28316 PR tree-optimization/81083
28317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28320 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28322 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28323 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28324 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28325 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28326 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28327 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28329 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28331 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28332 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28334 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28336 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28338 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28340 * config/rs6000/t-rtems: Don't handle SPE.
28342 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28344 * config/rs6000/t-linux: Don't handle SPE.
28346 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28348 * config/rs6000/eabispe.h: Delete file.
28350 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28352 * config/rs6000/t-spe: Delete file.
28354 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28356 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28357 (rs6000_legitimate_offset_address_p): Return false for anything in
28358 V2SImode or V2SFmode.
28360 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28362 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28363 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28364 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28366 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28367 (rs6000_legitimate_offset_address_p): Ditto.
28368 (rs6000_emit_move): Ditto.
28369 (rs6000_init_builtins): Remove V4HI_type_node.
28371 2017-06-13 Martin Liska <mliska@suse.cz>
28374 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28375 (gate_asan): Likewise.
28376 * asan.h (asan_no_sanitize_address_p): Remove the function.
28377 (sanitize_flags_p): New function.
28378 * builtins.def: Fix coding style.
28379 * common.opt: Use renamed enum value.
28380 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28381 * doc/extend.texi: Document no_sanitize attribute.
28382 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28383 to SANITIZE_UNDEFINED_NONDEFAULT.
28384 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28385 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28386 Use sanitize_flags_p.
28387 * gimplify.c (gimplify_function_tree): Likewise.
28388 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28389 * opts.c (parse_no_sanitize_attribute): New function.
28390 (common_handle_option): Use renamed enum value.
28391 * opts.h (parse_no_sanitize_attribute): Declare.
28392 * tree.c (sanitize_flags_p): New function.
28393 * tree.h: Declared here.
28394 * tsan.c: Use sanitize_flags_p.
28395 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28396 (instrument_mem_ref): Likewise.
28397 (instrument_bool_enum_load): Likewise.
28398 (do_ubsan_in_current_function): Remove the function.
28399 (pass_ubsan::execute): Use sanitize_flags_p.
28400 * ubsan.h: Remove do_ubsan_in_current_function
28401 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28402 (dump_function_to_file): Use it here.
28404 2017-06-13 Martin Jambor <mjambor@suse.cz>
28406 PR tree-optimization/80803
28407 PR tree-optimization/81063
28408 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28409 (propagate_subaccesses_across_link): Enqueue subtree whenever
28410 necessary instead of relying on the caller.
28412 2017-06-13 Martin Jambor <mjambor@suse.cz>
28414 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28415 that have a first_link.
28416 (sort_and_splice_var_accesses): Do not check first_link before
28418 (subtree_mark_written_and_enqueue): Likewise.
28419 (propagate_all_subaccesses): Likewise and do not stop at first
28420 parent with a first_link.
28422 2017-06-13 Martin Jambor <mjambor@suse.cz>
28424 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28427 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28429 * match.pd: New pattern.
28431 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28433 * tree-vrp.c (is_masked_range_test): New function.
28434 (register_edge_assert_for): Determine ranges for
28437 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28439 PR tree-optimization/67328
28440 * fold-const.c (maskable_range_p): New function.
28441 (build_range_check): Generate bittests if possible.
28443 2017-06-13 Martin Liska <mliska@suse.cz>
28445 * gimple-pretty-print.c (dump_probability): Add new argument.
28446 (dump_edge_probability): Dump both probability and count.
28447 (dump_gimple_label): Likewise.
28448 (dump_gimple_bb_header): Likewise.
28450 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28453 * config/avr/avr-devices.c: Fix indentation.
28454 * config/avr/gen-avr-mmcu-specs.c: Dito.
28456 2017-06-13 Richard Biener <rguenther@suse.de>
28458 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28459 instead get vector type from stmt_info.
28460 (vectorizable_reduction): Adjust. Remove dead code.
28462 2017-06-13 Richard Biener <rguenther@suse.de>
28464 PR middle-end/81065
28465 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28466 case of C * (x * C2 + C3).
28467 (fold_addr_of_array_ref_difference): Properly fold index difference.
28469 2017-06-12 David S. Miller <davem@davemloft.net>
28472 * config/sparc/sparc.md (return expander): Emit frame blockage if
28473 function uses alloca.
28475 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28477 * combine.c (make_field_assignment): Check len rather than the mode
28478 precision when calling force_to_mode.
28480 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28482 Support multilibs and devices that see flash in RAM address range.
28485 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28486 (avr_mcu_t) <flash_pm_offset>: New field.
28487 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28488 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28489 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28490 (AVR_TINY_PM_OFFSET): Remove macro.
28491 * config/avr/avr.opt (-mshort-calls): New option.
28492 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28493 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28494 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28495 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28496 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28497 instead of avr_arch->have_jmp_call.
28498 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28499 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28500 avr_arch->flash_pm_offset to define.
28501 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28502 new field flash_pm_offset. Add entry for avrxmega3.
28503 (avr_texinfo): Add entry for avrxmega3.
28504 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28505 attiny212, attiny214,
28506 attiny412, attiny414, attiny416, attiny417,
28507 attiny814, attiny816, attiny817,
28508 attiny1614, attiny1616, attiny1617,
28509 attiny3214, attiny3216, attiny3217.
28510 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28511 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28512 (avr_print_operand_address) [AVR_TINY]: Same.
28513 (avr_asm_init_sections) <readonly_data_section>: Only patch
28514 callback if avr_arch->flash_pm_offset = 0.
28515 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28516 for rodata if avr_arch->flash_pm_offset != 0.
28517 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28518 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28519 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28520 (m_options): Append opt_rcall.
28521 (m_dirnames): Append dir_rcall.
28522 * config/avr/t-multilib: Regenerate.
28524 * configure.ac [target=avr]: Check whether avrxmega3 default
28525 linker description file works as needed.
28526 * configure: Regenerate.
28527 * doc/avr-mmcu.texi: Regenerate.
28528 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28529 <__AVR_ARCH__>: Document avrxmega3 and 103.
28530 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28531 <__AVR_SHORT_CALLS__>: Document it.
28532 <__AVR_PM_BASE_ADDRESS__>: Document it.
28533 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28534 (AVR Variable Attributes) <progmem>: Document this is
28535 not needed for avrxmega3.
28536 (AVR Named Address Spaces) <__flash>: Dito.
28538 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28540 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28542 2017-06-12 Doug Rupp <rupp@adacore.com>
28544 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28545 Append vxworks-stdint.h to the tm_file list.
28546 * config/vxworks-stdint.h: New file.
28548 2017-06-12 Martin Liska <mliska@suse.cz>
28550 PR tree-optimization/81041
28551 * tree-profile.c (gimple_gen_ic_func_profiler):
28552 Create an extra BB in profile-generate
28553 (gimple_gen_time_profiler): Likewise.
28555 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28557 PR tree-optimization/81003
28558 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28559 (update_range_test): Use it instead of force_gimple_operand_gsi.
28561 2017-06-12 Richard Biener <rguenther@suse.de>
28563 PR tree-optimization/81053
28564 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28565 with backedge value not defined in loop. Simplify def stmt
28568 2017-06-11 Tom de Vries <tom@codesourcery.com>
28571 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28573 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28574 nvptx_cannot_force_const_mem.
28576 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28578 * opts.c (finish_options): Move test for flag_split_stack after
28579 it has been initialized.
28581 2017-06-11 Jason Merrill <jason@redhat.com>
28583 * tree.h (id_equal): New.
28584 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28585 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28586 instead of strcmp of IDENTIFIER_POINTER.
28588 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28590 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28591 (mark_all_inlined_calls_cdtor): Fix formating.
28592 (inline_transform): Rescale profile before inlining.
28594 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28596 * cgraph.h (cgraph_edge::clone): Update prototype.
28597 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28598 (cgraph_node::create_clone): Update.
28599 (cgraph_node::create_version_clone): Update.
28600 * tree-inline.c (copy_bb): Update.
28601 (expand_call_inline): Update.
28603 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28605 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28606 factored out from ...
28607 (rs6000_emit_prologue): ... here.
28609 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28611 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28612 factored out from ...
28613 (rs6000_emit_prologue): ... here.
28615 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28617 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28619 (handle_missing_profiles): Fix computation of tp_first_run.
28620 (counts_to_freqs): Do not touch freqs when count is 0.
28622 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28624 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28627 2017-06-10 Tom de Vries <tom@codesourcery.com>
28629 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28630 attributes): Document signal effective target.
28632 2017-06-10 Tom de Vries <tom@codesourcery.com>
28634 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28635 Document effective target stack_size.
28637 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28639 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28640 to the edit_context if they can be auto-applied.
28642 2017-06-9 Ian Lance Taylor <iant@golang.org>
28644 * opts.c (finish_options): If -fsplit-stack, disable implicit
28645 -forder-blocks-and-partition.
28646 * doc/invoke.texi (Optimize Options): Document that when using
28647 -fsplit-stack -forder-blocks-and-partition is not implicitly
28650 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28652 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28653 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28654 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28655 * builtins.def (abort, trap, unreachable): Declare cold.
28656 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28657 * tree-core.h (ECF_COLD): New.
28658 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28659 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28661 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28663 * predict.c (unlikely_executed_stmt_p): Cleanup.
28665 2017-06-09 Richard Biener <rguenther@suse.de>
28667 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28668 model if the ref is always written to.
28670 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28672 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28674 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28676 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28679 2017-06-09 Tom de Vries <tom@codesourcery.com>
28682 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28683 "target cannot support label values" when encountering LABEL_REF.
28685 2017-06-09 Martin Liska <mliska@suse.cz>
28687 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28688 (gimple_gen_ic_func_profiler): Emit direct comparison
28689 of __gcov_indirect_call_callee with NULL.
28690 (gimple_gen_time_profiler): Change probability from
28691 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28693 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28695 * profile.c (edge_gcov_counts): Turn to pointer.
28696 (compute_branch_probabilities, compute_branch_probabilities): Update.
28697 (branch_prob): Do not clear edge_gcov_count.
28698 * profile.h (edge_gcov_counts): Turn to pointer.
28699 (edge_gcov_count): Update.
28701 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28703 * gimple.h (gimple_check_failed): Mark cold.
28705 2017-06-09 Richard Biener <rguenther@suse.de>
28707 PR tree-optimization/66623
28708 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28709 refactor check_reduction into two parts, properly computing
28710 whether we have to check reduction validity for outer loop
28713 2017-06-09 Richard Biener <rguenther@suse.de>
28715 PR tree-optimization/79483
28716 * graphite-scop-detection.c (order): New global.
28717 (get_order): Compute bb to order mapping that satisfies code
28718 generation constraints.
28719 (cmp_pbbs): New helper.
28720 (build_scops): Start domwalk at entry block, sort generated
28723 2017-06-09 Richard Biener <rguenther@suse.de>
28725 PR middle-end/81007
28726 * ipa-polymorphic-call.c
28727 (ipa_polymorphic_call_context::restrict_to_inner_class):
28728 Skip FIELD_DECLs with error_mark_node type.
28729 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28732 2017-06-09 Martin Liska <mliska@suse.cz>
28734 * predict.c (struct branch_predictor): New struct.
28735 (test_prediction_value_range): New test.
28736 (predict_c_tests): New function.
28737 * selftest-run-tests.c (selftest::run_tests): Run the function.
28738 * selftest.h: Declare new tests.
28740 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28743 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28744 gen_add3_insn did not fail.
28745 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28746 r0, construct that number in a temporary reg and add that reg to r0.
28747 If asked to put the result in r0 as well, fail.
28749 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28751 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28752 for early expansion of vec_eqv.
28754 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28756 PR middle-end/81005
28757 * ubsan.c (instrument_null): Avoid pointless code temporary.
28758 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28760 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28764 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28767 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28770 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28772 * predict.c (unlikely_executed_edge_p): Move ahead.
28773 (probably_never_executed_edge_p): Use it.
28775 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28777 PR middle-end/79988
28778 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28779 gimple_call_builtin_p call.
28781 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28783 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28784 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28785 rtl_check_failed_type2, rtl_check_failed_code1,
28786 rtl_check_failed_code2, rtl_check_failed_code_mode,
28787 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28788 rtvec_check_failed_bounds, rtl_check_failed_flag,
28789 _fatal_insn_not_found, _fatal_insn): Likewise.
28790 * tree.h (tree_contains_struct_check_failed,
28791 tree_check_failed, tree_not_check_failed,
28792 tree_class_check_failed, tree_range_check_failed,
28793 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28794 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28795 tree_operand_check_failed, omp_clause_check_failed,
28796 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28799 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28801 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28802 flag_branch_probabilities.
28803 * ipa-inline.c (edge_badness): Likewise.
28804 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28805 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28806 * predict.c (maybe_hot_frequency_p): Likewise.
28807 (probably_never_executed): Likewise.
28808 * sched-ebb.c (schedule_ebbs): Likewise.
28809 * sched-rgn.c (find_single_block_region): Likewise.
28810 * tracer.c (tail_duplicate): Likewise.
28812 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28814 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28815 longer requires x_flag_profile_use.
28817 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28819 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28820 instead of flag_reorder_blocks_and_partition.
28821 * dbxout.c (dbxout_function_end): Likewise.
28822 * dwarf2out.c (gen_subprogram_die): Likewise.
28823 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28824 * hw-doloop.c (reorg_loops): Likewise.
28825 * varasm.c (assemble_start_function,
28826 assemble_end_function): Likewise.
28827 (decide_function_section): Do not check for
28828 flag_reorder_blocks_and_partition.
28830 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28832 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28834 (chkp_get_hard_register_fake_addr_expr): Ditto.
28835 (chkp_build_addr_expr): Add check for hard reg case.
28836 (chkp_parse_array_and_component_ref): Ditto.
28837 (chkp_find_bounds_1): Ditto.
28838 (chkp_process_stmt): Don't generate bounds store for
28841 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28843 * predict.c (maybe_hot_bb_p): Do not check profile status.
28844 (maybe_hot_edge_p): Likewise.
28845 (probably_never_executed): Check for zero counts even if profile
28847 (unlikely_executed_edge_p): New function.
28848 (unlikely_executed_stmt_p): New function.
28849 (unlikely_executed_bb_p): New function.
28850 (set_even_probabilities): Use unlikely predicates.
28851 (combine_predictions_for_bb): Likewise.
28852 (predict_paths_for_bb): Likewise.
28853 (predict_paths_leading_to_edge): Likewise.
28854 (determine_unlikely_bbs): New function.
28855 (estimate_bb_frequencies): Use it.
28856 (compute_function_frequency): Use zero counts even if profile is
28858 * profile-count.h: Fix typo.
28860 2017-08-08 Julia Koval <julia.koval@intel.com>
28862 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28863 _mm512_mask_cvtsepi16_storeu_epi8,
28864 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28865 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28866 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28867 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28868 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28869 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28870 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28871 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28872 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28873 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28874 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28875 __builtin_ia32_pmovuswb256mem_mask,
28876 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28877 __builtin_ia32_pmovwb512mem_mask): New builtins.
28879 2017-08-08 Julia Koval <julia.koval@intel.com>
28881 PR target/73350,80862
28882 * config/i386/subst.md (round): Fix round pattern.
28883 * config/i386/i386.c (ix86_erase_embedded_rounding):
28884 Fix erasing rounding for the fixed pattern.
28886 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28888 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28890 2017-06-08 Martin Liska <mliska@suse.cz>
28892 PR gcov-profile/80911
28893 * gcov.c (block_info::block_info): New constructor.
28895 2017-06-07 Carl Love <cel@us.ibm.com>
28897 * config/rs6000/rs6000-c: The return type of the following
28898 built-in functions was implemented as int not long long. Fix sign
28899 of return value for the unsigned version of vec_mulo and vec_mule.
28900 vector unsigned long long vec_bperm (vector unsigned long long,
28901 vector unsigned char)
28902 vector signed long long vec_mule (vector signed int,
28904 vector unsigned long long vec_mule (vector unsigned int,
28905 vector unsigned int)
28906 vector signed long long vec_mulo (vector signed int,
28908 vector unsigned long long vec_mulo (vector unsigned int,
28909 vector unsigned int)
28910 * doc/extend.texi: Fix the documentation for the built-in
28913 2017-06-07 Carl Love <cel@us.ibm.com>
28916 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28919 2017-06-07 Carl Love <cel@us.ibm.com>
28921 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28922 support, Generate doublehv for signed int/float for BE case only.
28924 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28926 * doc/invoke.texi (mcx16): Rewrite.
28928 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28930 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28931 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28932 *mov<mode>_softfloat, and an anonymous splitter): Use
28933 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28935 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28937 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28939 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28940 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28941 (rs6000_debug_reg_global): Adjust.
28942 (rs6000_init_hard_regno_mode_ok): Adjust.
28943 (rs6000_dbx_register_number): Adjust.
28944 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28945 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28946 Remove SPE_ACC and SPEFSCR.
28947 (REG_ALLOC_ORDER): Ditto.
28948 (FRAME_POINTER_REGNUM): Change to 111.
28949 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28950 (REG_CLASS_NAMES): Ditto.
28951 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28952 (REGISTER_NAMES): Ditto.
28953 (ADDITIONAL_REG_NAMES): Ditto.
28954 (rs6000_reg_names): Ditto.
28955 * config/rs6000/rs6000.md: Renumber some register number
28958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28960 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28962 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28963 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28965 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28966 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28967 Delete the SPE high registers.
28968 (REG_ALLOC_ORDER): Ditto.
28969 (enum reg_class): Remove SPE_HIGH_REGS.
28970 (REG_CLASS_NAMES): Ditto.
28971 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28972 (REGISTER_NAMES): Ditto.
28973 (rs6000_reg_names): Ditto.
28974 * doc/tm.texi.in: Remove SPE as example.
28975 * doc/tm.texi: Regenerate.
28977 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28979 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28980 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28981 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28982 * config/rs6000/rs6000.md (type): Remove "brinc".
28984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28986 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28987 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28988 * config/rs6000/linuxspe.h: Delete file.
28989 * config/rs6000/rs6000.md: Don't include spe.md.
28990 * config/rs6000/spe.h: Delete file.
28991 * config/rs6000/spe.md: Delete file.
28992 * config/rs6000/t-rs6000: Remove spe.md.
28994 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28996 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28997 (reg_or_none500mem_operand): Delete.
28998 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28999 instead of reg_or_none500mem_operand.
29001 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29003 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29004 handling of SPE flags.
29005 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29007 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29009 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29011 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29012 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29013 paired_divv2sf3): Similar.
29014 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29015 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29016 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29018 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29019 Rename the paired_* instruction patterns.
29020 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29022 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29023 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29024 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29025 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29026 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29027 PAIRED_VECTOR_MODE.
29028 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29029 (spe_func_has_64bit_regs_p): Delete.
29030 (spe_expand_predicate_builtin): Delete.
29031 (spe_expand_evsel_builtin): Delete.
29032 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29033 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29034 (invalid_e500_subreg): Delete.
29035 (rs6000_legitimize_address): Always force_reg op2 as well, for
29036 paired single memory accesses.
29037 (rs6000_member_type_forces_blk): Delete.
29038 (rs6000_spe_function_arg): Delete.
29039 (rs6000_expand_unop_builtin): Delete SPE handling.
29040 (rs6000_expand_binop_builtin): Ditto.
29041 (spe_expand_stv_builtin): Delete.
29042 (bdesc_2arg_spe): Delete.
29043 (spe_expand_builtin): Delete.
29044 (spe_expand_predicate_builtin): Delete.
29045 (spe_expand_evsel_builtin): Delete.
29046 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29047 (spe_init_builtins): Delete.
29048 (spe_func_has_64bit_regs_p): Delete.
29049 (savres_routine_name): Delete "info" parameter. Adjust callers.
29050 (rs6000_emit_stack_reset): Ditto.
29051 (rs6000_dwarf_register_span): Delete.
29052 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29053 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29054 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29056 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29058 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29059 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29060 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29061 mulv2sf3, divv2sf3): Delete expanders.
29063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29065 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29067 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29069 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29070 * config/rs6000/rs6000.c: Ditto.
29072 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29074 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29075 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29076 comparison_operator.
29078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29080 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29081 * config/rs6000/rs6000.opt: Ditto.
29082 * config/rs6000/t-rtems: Ditto.
29084 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29086 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29087 TARGET_E500_SINGLE by 0, simplify.
29088 * config/rs6000/rs6000.c: Ditto.
29089 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29090 (spe_build_register_parallel): Delete.
29091 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29092 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29093 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29094 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29095 (E500_CONVERT): Delete.
29096 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29100 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29101 * config/rs6000/dfp.md: Ditto.
29102 (negdd2, *negdd2_fpr): Merge.
29103 (absdd2, *absdd2_fpr): Merge.
29104 (negtd2, *negtd2_fpr): Merge.
29105 (abstd2, *abstd2_fpr): Merge.
29106 * config/rs6000/e500.h: Delete file.
29107 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29108 TARGET_FPRS by 1 and simplify.
29109 * config/rs6000/rs6000-c.c: Ditto.
29110 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29111 TARGET_DF_SPE by 0.
29112 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29114 * config/rs6000/rs6000.md: Ditto.
29115 (floatdidf2, *floatdidf2_fpr): Merge.
29116 (move_from_CR_gt_bit): Delete.
29117 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29118 (E500_CR_IOR_COMPARE): Delete.
29119 (All patterns that require !TARGET_FPRS): Delete.
29120 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29124 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29128 * graphds.c (add_edge): Intitialize edge's attached data.
29129 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29130 pointer parameter. Call pointed function on each edge during
29131 graph traversing. Skip traversing the edge when the function
29133 (graphds_dfs, graphds_scc): Ditto.
29134 (for_each_edge): New parameter. Pass the new parameter to callback
29136 * graphds.h (skip_edge_callback): New function pointer type.
29137 (graphds_dfs, graphds_scc): New function pointer parameter.
29138 (graphds_edge_callback, for_each_edge): New parameter.
29140 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29142 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29143 out code checking if runtime alias check is possible to below ...
29144 Call the new function.
29145 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29146 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29148 2017-06-07 Marek Polacek <polacek@redhat.com>
29151 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29152 TYPE_OVERFLOW_WRAPS check.
29154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29156 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29157 if versioning is required.
29158 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29159 peeling with the check for versioning.
29161 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29163 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29164 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29165 Set true to new parameter if new ssa variable is defined.
29166 (vect_gen_vector_loop_niters): Refactor. Set range information
29167 for the new vector loop bound variable.
29168 (vect_do_peeling): Ditto.
29170 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29172 * tree-affine.c (ssa.h): Include header file.
29173 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29174 has wrapping overflow behavior.
29176 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29178 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29180 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29182 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29183 (tree_to_aff_combination): ... here.
29185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29187 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29188 reg_pressure model function.
29189 (ivopts_global_cost_for_size): Delete.
29190 (determine_set_costs, iv_ca_recount_cost): Call new model function
29191 ivopts_estimate_reg_pressure.
29193 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29195 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29196 expensive than udiv. Remove floating point cases from mod.
29198 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29200 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29201 Increase idiv cost.
29203 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29205 * config/aarch64/aarch64.md
29206 (copysignsf3): Fix mask generation.
29208 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29210 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29212 (class dump_manager): Add register_dumps method.
29213 * dumpfile.c: Include langhooks.h.
29214 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29215 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29216 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29217 (dump_manager::dump_register): Start with 512 entries instead of 32.
29218 (dump_manager::register_dumps): New method.
29219 * toplev.c (general_init): Instead of invoking register_dumps
29220 langhook, invoke register_dumps method on the dump manager.
29221 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29224 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29226 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29227 Say that # requires an associated define_split to exist, and that
29228 the define_split must be suitable for use after register allocation.
29230 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29232 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29233 (compute_outgoing_frequencies): Also initialize zero counts.
29234 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29235 around loops; preserve more of profile when nothing changes.
29237 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29239 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29241 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29242 * config/arm/arm-cpu-cdata.h: Regenerate.
29243 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29244 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29245 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29246 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29247 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29249 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29251 * config/arm/t-rmprofile: Likewise.
29252 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29254 2017-06-06 David S. Miller <davem@davemloft.net>
29257 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29258 blockage if function uses alloca.
29260 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29262 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29263 New "uid" fields to hold pretty-print IDs of group and ref.
29264 Memory references are now identified as <group_id>:<ref_id>
29265 instead of using [random] addresses.
29266 (dump_mem_details): Simplify, no functional change.
29267 (dump_mem_ref): Simplify and make output more concise.
29268 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29269 (find_or_create_group): Initialize group uid.
29270 (record_ref): Initialize ref uid. Improve debug output.
29271 (prune_group_by_reuse, should_issue_prefetch_p,)
29272 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29273 (mark_nontemporal_store, determine_loop_nest_reuse):
29274 Improve debug output.
29276 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29278 * dbgcnt.def (prefetch): New debug counter.
29279 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29280 (schedule_prefetches): Stop issueing prefetches if debug counter
29283 2017-06-06 Tom de Vries <tom@codesourcery.com>
29285 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29286 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29288 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29290 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29291 Use aarch64_reg_or_zero predicate for operand 4.
29292 (aarch64_compare_and_swap<mode> define_insn_and_split):
29293 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29294 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29296 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29298 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29299 (arm_compute_save_core_reg_mask): This.
29300 (thumb1_compute_save_reg_mask): Rename into ...
29301 (thumb1_compute_save_core_reg_mask): This.
29302 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29303 (arm_compute_frame_layout): Likewise.
29305 2017-06-06 Richard Biener <rguenther@suse.de>
29307 PR tree-optimization/80974
29308 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29309 keep or clear leaders SSA info.
29311 2017-06-06 Tom de Vries <tom@codesourcery.com>
29313 * config/nvptx/nvptx.c (split_mode_p): New function.
29314 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29316 2017-06-06 Tom de Vries <tom@codesourcery.com>
29318 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29320 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29323 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29326 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29328 * shrink-wrap.c (handle_simple_exit): Update profile.
29329 (try_shrink_wrapping): Upate profile.
29331 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29333 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29334 (tree_guess_outgoing_edge_probabilities): New.
29335 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29336 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29338 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29340 * ipa-split.c (split_function): Initialize return bb profile.
29342 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29344 * profile.c (compute_branch_probabilities): Also initialize
29345 EXIT_BLOCK profile.
29347 2017-06-06 Richard Biener <rguenther@suse.de>
29349 PR tree-optimization/80928
29350 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29351 (vect_analyze_loop_operations): Properly guard analysis for
29353 (vect_transform_loop): Likewise.
29354 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29355 (vect_model_induction_cost): Do not cost for pure SLP.
29356 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29357 of induction in inner loop vectorization.
29358 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29359 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29360 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29362 (vect_analyze_slp_cost_1): Cost induction.
29363 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29364 (vect_get_slp_vect_defs): Likewise.
29365 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29366 (vect_transform_stmt): Handle SLP reductions.
29367 * tree-vectorizer.h (vectorizable_induction): Adjust.
29369 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29371 * config/rs6000/rs6000.c (make_resolver_func): Update
29372 init_lowered_empty_function call.
29374 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29376 * doc/invoke.texi: Document the -fprofile-abs-path option.
29377 * common.opt (fprofile-abs-path): New option.
29378 * gcov-io.h (gcov_write_filename): Declare.
29379 * gcov-io.c (gcov_write_filename): New function.
29380 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29381 * profile.c (output_location): Likewise.
29383 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29385 * shring-wrap.c: Revert accidental commit.
29387 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29389 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29391 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29393 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29395 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29396 profile in callgraph edge.
29397 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29398 (apply_scale): Likewise.
29399 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29400 Also scale profile when inlining function with zero profile.
29401 (initialize_cfun): Update exit block profile even when it is zero.
29402 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29403 when profile is read.
29405 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29407 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29408 (CLONE_*): New constants to define the processors we can generate
29409 code for with the target_clone attribute.
29410 (rs6000_clone_map): New array to identify which clone processors
29411 the current program is running on.
29412 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29413 target_clone attribute.
29414 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29415 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29416 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29417 (cpu_expand_builtin): Add support for target_clone attribute.
29418 (rs6000_valid_attribute_p): Allow "default" attribute.
29419 (get_decl_name): New debug function to simplify printing the
29420 current function name in debugging statements.
29421 (rs6000_clone_priority): New functions to support the target_clone
29422 attribute, and be able to generate code to switch between ISA 2.05
29423 through ISA 3.0 (power6 through power9).
29424 (rs6000_compare_version_priority): Likewise.
29425 (rs6000_get_function_versions_dispatcher): Likewise.
29426 (make_resolver_func): Likewise.
29427 (add_condition_to_bb): Likewise.
29428 (dispatch_function_versions): Likewise.
29429 (rs6000_generate_version_dispatcher_body): Likewise.
29430 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29431 (fusion_gpr_load_p): Fix a spacing issue.
29432 * doc/extend.texi (Common Function Attributes): Document that the
29433 PowerPC supports the target_clone attribute.
29435 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29437 * config/arm/arm.h: explain F symbol found in description of ARM
29438 register allocation in its legend.
29440 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29442 * config/mips/frame-header-opt.c: Include profile-count.h.
29443 * config/riscv/riscv.c: Include profile-count.h
29445 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29447 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29449 (sm_set_flag_if_changed): Add bbs field.
29450 (execute_sm_if_changed_flag_set): Pass BBS.
29451 (execute_sm): Update.
29453 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29455 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29458 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29460 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29461 (peephole2): New peephole2 to emit the above.
29462 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29464 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29466 * config/aarch64/aarch64.c (define_peephole2 above
29467 *sub_<shift>_<mode>): New peephole.
29469 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29471 * config/i386/i386.c (make_resolver_func): Update.
29472 * Makefile.in: Add profile-count.h and profile-count.o
29473 * auto-profile.c (afdo_indirect_call): Update to new API.
29474 (afdo_set_bb_count): Update.
29475 (afdo_propagate_edge): Update.
29476 (afdo_propagate_circuit): Update.
29477 (afdo_calculate_branch_prob): Update.
29478 (afdo_annotate_cfg): Update.
29479 * basic-block.h: Include profile-count.h
29480 (struct edge_def): Turn count to profile_count.
29481 (struct basic_block_def): Likewie.
29482 (REG_BR_PROB_BASE): Move to profile-count.h
29483 (RDIV): Move to profile-count.h
29484 * bb-reorder.c (max_entry_count): Turn to profile_count.
29485 (find_traces): Update.
29486 (rotate_loop):Update.
29487 (connect_traces):Update.
29488 (sanitize_hot_paths):Update.
29489 * bt-load.c (migrate_btr_defs): Update.
29490 * cfg.c (RDIV): Remove.
29491 (init_flow): Use alloc_block.
29492 (alloc_block): Uninitialize count.
29493 (unchecked_make_edge): Uninitialize count.
29494 (check_bb_profile): Update.
29495 (dump_edge_info): Update.
29496 (dump_bb_info): Update.
29497 (update_bb_profile_for_threading): Update.
29498 (scale_bbs_frequencies_int): Update.
29499 (scale_bbs_frequencies_gcov_type): Update.
29500 (scale_bbs_frequencies_profile_count): New.
29501 * cfg.h (update_bb_profile_for_threading): Update.
29502 (scale_bbs_frequencies_profile_count): Declare.
29503 * cfgbuild.c (compute_outgoing_frequencies): Update.
29504 (find_many_sub_basic_blocks): Update.
29505 * cfgcleanup.c (try_forward_edges): Update.
29506 (try_crossjump_to_edge): Update.
29507 * cfgexpand.c (expand_gimple_tailcall): Update.
29508 (construct_exit_block): Update.
29509 * cfghooks.c (verify_flow_info): Update.
29510 (dump_bb_for_graph): Update.
29511 (split_edge): Update.
29512 (make_forwarder_block): Update.
29513 (duplicate_block): Update.
29514 (account_profile_record): Update.
29515 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29516 (get_estimated_loop_iterations): Update.
29517 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29518 (single_likely_exit): Update.
29519 * cfgloopmanip.c (scale_loop_profile): Update.
29521 (set_zero_probability): Update.
29522 (lv_adjust_loop_entry_edge): Update.
29523 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29524 (purge_dead_edges): Update.
29525 (rtl_account_profile_record): Update.
29526 * cgraph.c (cgraph_node::create): Uninitialize count.
29527 (symbol_table::create_edge): Uninitialize count.
29528 (cgraph_update_edges_for_call_stmt_node): Update.
29529 (cgraph_edge::dump_edge_flags): Update.
29530 (cgraph_node::dump): Update.
29531 (cgraph_edge::maybe_hot_p): Update.
29532 * cgraph.h: Include profile-count.h
29533 (create_clone), create_edge, create_indirect_edge): Update.
29534 (cgraph_node): Turn count to profile_count.
29535 (cgraph_edge0: Likewise.
29536 (make_speculative, clone): Update.
29537 (create_edge): Update.
29538 (init_lowered_empty_function): Update.
29539 * cgraphclones.c (cgraph_edge::clone): Update.
29540 (duplicate_thunk_for_node): Update.
29541 (cgraph_node::create_clone): Update.
29542 * cgraphunit.c (cgraph_node::analyze): Update.
29543 (cgraph_node::expand_thunk): Update.
29544 * final.c (dump_basic_block_info): Update.
29545 * gimple-streamer-in.c (input_bb): Update.
29546 * gimple-streamer-out.c (output_bb): Update.
29547 * graphite.c (print_global_statistics): Update.
29548 (print_graphite_scop_statistics): Update.
29549 * hsa-brig.c: Include basic-block.h.
29550 * hsa-dump.c: Include basic-block.h.
29551 * hsa-gen.c (T sum_slice): Update.
29552 (convert_switch_statements):Update.
29553 * hsa-regalloc.c: Include basic-block.h.
29554 * ipa-chkp.c (chkp_produce_thunks): Update.
29555 * ipa-cp.c (struct caller_statistics): Update.
29556 (init_caller_stats): Update.
29557 (gather_caller_stats): Update.
29558 (ipcp_cloning_candidate_p): Update.
29559 (good_cloning_opportunity_p): Update.
29560 (get_info_about_necessary_edges): Update.
29561 (dump_profile_updates): Update.
29562 (update_profiling_info): Update.
29563 (update_specialized_profile): Update.
29564 (perhaps_add_new_callers): Update.
29565 (decide_about_value): Update.
29566 (ipa_cp_c_finalize): Update.
29567 * ipa-devirt.c (struct odr_type_warn_count): Update.
29568 (struct decl_warn_count): Update.
29569 (struct final_warning_record): Update.
29570 (possible_polymorphic_call_targets): Update.
29571 (ipa_devirt): Update.
29572 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29573 * ipa-icf.c (sem_function::merge): Update.
29574 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29575 * ipa-inline.c (compute_uninlined_call_time): Update.
29576 (compute_inlined_call_time): Update.
29577 (want_inline_small_function_p): Update.
29578 (want_inline_self_recursive_call_p): Update.
29579 (edge_badness): Update.
29580 (lookup_recursive_calls): Update.
29581 (recursive_inlining): Update.
29582 (inline_small_functions): Update.
29583 (dump_overall_stats): Update.
29584 (dump_inline_stats): Update.
29585 * ipa-profile.c (ipa_profile_generate_summary): Update.
29586 (ipa_propagate_frequency): Update.
29587 (ipa_profile): Update.
29588 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29589 * ipa-utils.c (ipa_merge_profiles): Update.
29590 * loop-doloop.c (doloop_modify): Update.
29591 * loop-unroll.c (report_unroll): Update.
29592 (unroll_loop_runtime_iterations): Update.
29593 * lto-cgraph.c (lto_output_edge): Update.
29594 (lto_output_node): Update.
29595 (input_node): Update.
29596 (input_edge): Update.
29597 (merge_profile_summaries): Update.
29598 * lto-streamer-in.c (input_cfg): Update.
29599 * lto-streamer-out.c (output_cfg): Update.
29600 * mcf.c (create_fixup_graph): Update.
29601 (adjust_cfg_counts): Update.
29602 (sum_edge_counts): Update.
29603 * modulo-sched.c (sms_schedule): Update.
29604 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29605 * predict.c (maybe_hot_count_p): Update.
29606 (probably_never_executed): Update.
29607 (dump_prediction): Update.
29608 (combine_predictions_for_bb): Update.
29609 (propagate_freq): Update.
29610 (handle_missing_profiles): Update.
29611 (counts_to_freqs): Update.
29612 (rebuild_frequencies): Update.
29613 (force_edge_cold): Update.
29614 * predict.h: Include profile-count.h
29615 (maybe_hot_count_p, counts_to_freqs): UPdate.
29616 * print-rtl-function.c: Do not include cfg.h
29617 * print-rtl.c: Include basic-block.h
29618 * profile-count.c: New file.
29619 * profile-count.h: New file.
29620 * profile.c (is_edge_inconsistent): Update.
29621 (correct_negative_edge_counts): Update.
29622 (is_inconsistent): Update.
29623 (set_bb_counts): Update.
29624 (read_profile_edge_counts): Update.
29625 (compute_frequency_overlap): Update.
29626 (compute_branch_probabilities): Update; Initialize and deinitialize
29628 (branch_prob): Update.
29629 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29630 (edge_gcov_count): New.
29631 (bb_gcov_count): New.
29632 * shrink-wrap.c (try_shrink_wrapping): Update.
29633 * tracer.c (better_p): Update.
29634 * trans-mem.c (expand_transaction): Update.
29635 (ipa_tm_insert_irr_call): Update.
29636 (ipa_tm_insert_gettmclone_call): Update.
29637 * tree-call-cdce.c: Update.
29638 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29639 (gimple_duplicate_sese_tail): Update.
29640 (gimple_account_profile_record): Update.
29641 (execute_fixup_cfg): Update.
29642 * tree-inline.c (copy_bb): Update.
29643 (copy_edges_for_bb): Update.
29644 (initialize_cfun): Update.
29645 (freqs_to_counts): Update.
29646 (copy_cfg_body): Update.
29647 (expand_call_inline): Update.
29648 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29649 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29650 (try_unroll_loop_completely): Update.
29651 (try_peel_loop): Update.
29652 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29653 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29654 * tree-ssa-loop-split.c (connect_loops): Update.
29655 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29656 * tree-ssa-reassoc.c (branch_fixup): Update.
29657 * tree-ssa-tail-merge.c (replace_block_by): Update.
29658 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29659 (compute_path_counts): Update.
29660 (update_profile): Update.
29661 (recompute_probabilities): Update.
29662 (update_joiner_offpath_counts): Update.
29663 (estimated_freqs_path): Update.
29664 (freqs_to_counts_path): Update.
29665 (clear_counts_path): Update.
29666 (ssa_fix_duplicate_block_edges): Update.
29667 (duplicate_thread_path): Update.
29668 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29669 (struct switch_conv_info): Update.
29670 * tree-tailcall.c (decrease_profile): Update.
29671 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29672 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29673 * value-prof.c (check_counter): Update.
29674 (gimple_divmod_fixed_value): Update.
29675 (gimple_mod_pow2): Update.
29676 (gimple_mod_subtract): Update.
29677 (gimple_ic_transform): Update.
29678 (gimple_stringop_fixed_value): Update.
29679 * value-prof.h (gimple_ic): Update.
29681 2017-06-02 Carl Love <cel@us.ibm.com>
29683 * config/rs6000/rs6000-c: Add support for built-in functions
29684 vector double vec_doublee (vector signed int);
29685 vector double vec_doublee (vector unsigned int);
29686 vector double vec_doublee (vector float);
29687 vector double vec_doubleh (vector signed int);
29688 vector double vec_doubleh (vector unsigned int);
29689 vector double vec_doubleh (vector float);
29690 vector double vec_doublel (vector signed int);
29691 vector double vec_doublel (vector unsigned int);
29692 vector double vec_doublel (vector float);
29693 vector double vec_doubleo (vector signed int);
29694 vector double vec_doubleo (vector unsigned int);
29695 vector double vec_doubleo (vector float);.
29696 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29697 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29699 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29700 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29701 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29703 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29704 vec_doublel, vec_doubleh.
29705 * doc/extend.texi: Update the built-in documentation file for the
29706 new built-in functions.
29708 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29711 * ipa-inline-analysis.c (free_growth_caches): Set
29712 edge_removal_hook_holder to NULL after removing it.
29714 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29716 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29717 comparision with zero.
29719 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29721 for early expansion of vec_min and vec_max builtins.
29722 (builtin_function_type): Add min/max unsigned variants to those
29723 identified as having unsigned arguments.
29725 2017-06-02 Olivier Hainque <hainque@adacore.com>
29727 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29729 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29731 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29732 Use VALL_F16 iterator rather than VALL.
29734 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29736 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29737 Emit CBNZ inside loop when doing a strong exchange and comparing
29738 against zero. Generate the CC flags after the loop.
29740 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29742 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29743 (dl_section_ref): New.
29744 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29745 On AIX, append an expression to subtract the size of the
29746 section length to dl_section_ref.
29748 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29751 for early expansion of vector absolute builtins.
29753 2017-06-02 Richard Biener <rguenther@suse.de>
29755 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29756 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29758 2017-06-02 Richard Biener <rguenther@suse.de>
29760 PR tree-optimization/80948
29761 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29762 stmt order as well.
29764 2017-06-02 Richard Biener <rguenther@suse.de>
29766 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29768 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29769 for inductions as relevant.
29771 2017-06-02 Richard Biener <rguenther@suse.de>
29773 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29774 (vectorizable_induction): ... this. Remove dead code.
29776 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29778 * builtins. (expand_builtin_alloca): Remove second parameter and
29779 infer its value from the first parameter instead.
29780 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29782 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29784 PR rtl-optimization/80903
29785 * loop-doloop.c (add_test): Unshare sequence.
29787 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29789 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29791 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29793 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29795 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29796 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29797 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29798 (xlogue_layout::get_stub_rtx): Make static.
29799 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29800 (xlogue_layout::compute_stub_managed_regs): Rename to...
29801 (xlogue_layout::count_stub_managed_regs): ...this.
29802 (xlogue_layout::is_stub_managed_reg): New function.
29803 (xlogue_layout::m_stub_names): Rename to...
29804 (xlogue_layout::s_stub_names): ...this, make static.
29805 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29806 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29807 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29808 xlogue_layout::s_stub_names): Instantiate statics.
29809 (stub_managed_regs): Remove.
29810 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29811 (disable_call_ms2sysv_xlogues): Rename to...
29812 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29813 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29815 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29816 change after reload_completed.
29817 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29819 (ix86_expand_prologue): Likewise.
29820 (ix86_expand_epilogue): Likewise.
29821 (ix86_expand_split_stack_prologue): Likewise.
29822 (ix86_compute_frame_layout): Remove frame parameter ...
29823 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29824 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29826 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29827 (ix86_frame): Move from here ...
29828 * config/i386/i386.h (ix86_frame): ... to here.
29829 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29830 complete ix86_frame data structure instead. Remove some_ld_name.
29832 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29834 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29835 symbols that hold a DECL_VALUE_EXPR.
29837 2017-06-01 Martin Jambor <mjambor@suse.cz>
29839 PR tree-optimization/80898
29840 * tree-sra.c (process_subtree_disqualification): Removed.
29841 (disqualify_candidate): Do not acll
29842 process_subtree_disqualification.
29843 (subtree_mark_written_and_enqueue): New function.
29844 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29845 RHS has been disqualified and re-queue LHS if necessary. Apart
29846 from that, ignore disqualified RHS.
29848 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29850 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29851 address fetch for z10 or later.
29853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29855 * config/arc/arc.md (tst_movb): Add guard when splitting.
29857 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29859 * config/arc/arc.c (arc_can_eliminate): Test against
29860 arc_frame_pointer_needed.
29862 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29864 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29865 to prevent store reordering.
29866 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29867 (type): Add block type.
29868 (stack_tie): Define special instruction to be used in
29871 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29873 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29874 constraint. It is not valid for the pattern.
29875 (noncommutative_binary_comparison): Likewise.
29877 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29879 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29882 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29884 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29885 be used by the reg-alloc.
29887 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29889 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29890 reg-alloc when having mul64 or mul32x16 instructions.
29891 (mulsidi3): Likewise.
29892 (umulsidi3): Likewise.
29893 (mulsi32x16): New pattern.
29894 (mulsi64): Likewise.
29895 (mulsidi64): Likewise.
29896 (umulsidi64): Likewise.
29897 (MUL32x16_REG): Define.
29898 (mul64_600): Use MUL32x16_REG.
29899 (mac64_600): Likewise.
29900 (umul64_600): Likewise.
29901 (umac64_600): Likewise.
29903 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29905 * config/arc/arc.md (mulsi3_700): Make it commutative.
29907 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29909 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29911 (*sign_extendsidi2_insn): Likewise for movstosw.
29913 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29915 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29916 the type of the input discriminant value. Convert the
29917 discriminant value of signedness vary.
29919 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29921 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29922 Add to -Wall section.
29924 2017-06-01 Richard Biener <rguenther@suse.de>
29926 PR middle-end/66313
29927 * fold-const.c (fold_plusminus_mult_expr): If the factored
29928 factor may be zero use a wrapping type for the inner operation.
29929 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29930 and handle moved defs.
29931 (process_assignment): Properly guard the unary op case. Return a
29932 tri-state indicating that moving the stmt before the call may allow
29933 to continue. Pass through to_move.
29934 (find_tail_calls): Handle moving unrelated defs before
29937 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29940 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29941 splitter result in the canonical way.
29943 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29945 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29946 also for 32bit target. Update insn attributes.
29947 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29949 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29951 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29952 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29953 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29954 (_mm_maskz_min_ss): New intrinsics.
29956 2017-05-31 Martin Liska <mliska@suse.cz>
29958 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29959 Change comment style to one we normally use.
29960 (vectorizable_reduction): Likewise.
29961 (vectorizable_induction): Likewise.
29962 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29963 (vectorizable_call): Likewise.
29964 (vectorizable_simd_clone_call): Likewise.
29965 (vectorizable_conversion): Likewise.
29966 (vectorizable_assignment): Likewise.
29967 (vectorizable_shift): Likewise.
29968 (vectorizable_operation): Likewise.
29969 (vectorizable_store): Likewise.
29970 (vectorizable_load): Likewise.
29971 * tree-vectorizer.h: Likewise.
29973 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29975 * passes.c (emergency_dump_function): New.
29976 * tree-pass.h (emergency_dump_function): Declare.
29977 * plugin.c (plugins_internal_error_function): Remove.
29978 * plugin.h (plugins_internal_error_function): Remove declaration.
29979 * toplev.c (internal_error_function): New static function. Use it...
29980 (general_init): ...here.
29982 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29984 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29985 (arc_rtx_costs): Add costs for new patterns.
29986 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29987 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29989 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29991 * tree-ssa-strlen.c (get_next_strinfo): New function.
29992 (get_stridx_plus_constant): Use it.
29993 (zero_length_string): Likewise.
29994 (adjust_related_strinfos): Likewise.
29995 (adjust_last_stmt): Likewise.
29997 2017-05-31 Richard Biener <rguenther@suse.de>
30000 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30001 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30003 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30005 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30006 loop_vinfo argument and use of dependence distance vectors.
30007 Check instead whether the two references differ only in their
30008 initial value and assume that they have the same alignment if the
30009 difference is a multiple of the vector alignment.
30010 (vect_analyze_data_refs_alignment): Update call accordingly.
30012 2017-05-31 Martin Liska <mliska@suse.cz>
30015 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30017 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30019 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30020 (create_intersect_range_checks): Move from ...
30021 * tree-data-ref.c (create_intersect_range_checks_index)
30022 (create_intersect_range_checks): ... to here.
30023 (create_runtime_alias_checks): New function factored from ...
30024 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30025 here. Call above function.
30026 * tree-data-ref.h (create_runtime_alias_checks): New function.
30028 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30030 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30031 segment length for dr_b and compute it in wide_int.
30033 2017-05-31 Richard Biener <rguenther@suse.de>
30035 PR tree-optimization/80906
30036 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30037 and pass through iv_map.
30038 (copy_bb_and_scalar_dependences): Adjust.
30039 (translate_pending_phi_nodes): Likewise.
30040 (copy_loop_close_phi_args): Handle code-generating IVs instead
30043 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30045 * diagnostic-color.c (color_dict): Add "type-diff".
30046 (parse_gcc_colors): Update comment.
30047 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30048 -fdiagnostics-show-template-tree and -fno-elide-type.
30049 (GCC_COLORS): Add type-diff to example.
30051 (-fdiagnostics-show-template-tree): New.
30052 (-fno-elide-type): New.
30053 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30054 the pp_format_decoder callback. Call any m_format_postprocessor's
30056 (pretty_printer::pretty_printer): Initialize
30057 m_format_postprocessor.
30058 (pretty_printer::~pretty_printer): Delete any
30059 m_format_postprocessor.
30060 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30061 (class format_postprocessor): New class.
30062 (struct pretty_printer::format_decoder): Document the new parameters.
30063 (struct pretty_printer::m_format_postprocessor): New field.
30064 * tree-diagnostic.c (default_tree_printer): Update for new
30065 bool and const char ** params.
30066 * tree-diagnostic.h (default_tree_printer): Likewise.
30068 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30070 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30071 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30073 rs6000_gen_cell_microcode code.
30074 (rs6000_final_prescan_insn): Delete.
30075 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30076 "warn-cell-microcode" entries.
30077 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30078 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30079 throughout. Change cc_reg_not_micro_cr0_operand to
30080 cc_reg_not_cr0_operand throughout.
30081 (*extendhi<mode>2_noload): Delete.
30082 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30083 (mwarn-cell-microcode): Delete.
30084 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30085 -mgen-cell-microcode and -mwarn-cell-microcode.
30087 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30090 * config/i386/constraints.md (Yd): New constraint.
30092 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30093 and (?Yd, r) alternatives. Update insn attributes.
30094 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30095 and (?*Yd, r) alternatives. Update insn attributes.
30096 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30098 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30100 * gimplify.c (gimplify_modify_expr): Don't create a
30101 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30104 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30106 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30108 2017-05-30 Richard Biener <rguenther@suse.de>
30110 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30111 and reduc_def fields.
30112 (STMT_VINFO_REDUC_TYPE): New define.
30113 (STMT_VINFO_REDUC_DEF): Likewise.
30114 (vect_force_simple_reduction): Adjust prototype.
30115 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30116 (vect_is_simple_reduction): Remove check_reduction argument.
30117 (vect_force_simple_reduction): Adjust and set
30118 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30119 (vectorizable_reduction): Do not re-do reduction analysis
30120 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30121 * tree-parloops.c (gather_scalar_reductions): Adjust.
30123 2017-05-30 Richard Biener <rguenther@suse.de>
30125 PR middle-end/80901
30126 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30129 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30131 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30132 Introduce unknown_misalignment parameter and remove vf.
30133 (vect_peeling_hash_get_lowest_cost):
30134 Pass unknown_misalignment parameter.
30135 (vect_enhance_data_refs_alignment):
30136 Fix unsupportable data ref treatment.
30138 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30140 * tree-vect-data-refs.c (vect_get_data_access_cost):
30141 Workaround for SLP handling.
30142 (vect_enhance_data_refs_alignment):
30143 Compute costs for doing no peeling at all, compare to the best
30144 peeling costs so far and avoid peeling if cheaper.
30146 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30148 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30149 Return peeling info and set costs to zero for unlimited cost
30151 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30152 with unknown misalignment. Compute and costs for unknown
30153 misalignment, compare them to the costs for known misalignment
30154 and choose the cheapest for peeling.
30156 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30158 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30159 (vect_get_peeling_costs_all_drs): Create function.
30160 (vect_peeling_hash_get_lowest_cost):
30161 Use vect_get_peeling_costs_all_drs.
30162 (vect_peeling_supportable): Create function.
30163 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30165 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30168 DR_HAS_NEGATIVE_STEP.
30169 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30170 (vect_enhance_data_refs_alignment): Use.
30171 (vect_duplicate_ssa_name_ptr_info): Use.
30172 * tree-vectorizer.h (dr_misalignment): Use.
30173 (known_alignment_for_access_p): Use.
30175 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30178 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30180 (has_section_name): New function.
30182 2017-05-30 Martin Liska <mliska@suse.cz>
30185 * auto-profile.c (get_function_decl_from_block): Fix
30188 2017-05-30 Richard Biener <rguenther@suse.de>
30190 PR middle-end/80876
30191 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30193 2017-05-30 Martin Liska <mliska@suse.cz>
30195 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30196 * dumpfile.h (struct dump_file_info): Remove ctors.
30198 2017-05-30 Martin Liska <mliska@suse.cz>
30200 * predict.def: Fix GNU coding style.
30202 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30204 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30205 Mark 'to' argument with ATTRIBUTE_UNUSED.
30207 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30209 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30210 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30211 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30214 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30216 * doc/install.texi (Options specification): Restore entry of
30217 --enable-sjlj-exceptions.
30219 2017-05-27 Michael Eager <eager@eagercon.com>
30222 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30224 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30226 * config/microblaze/microblaze.h
30227 (FIXED_REGISTERS): Update in macro.
30228 (CALL_USED_REGISTERS): Update in macro.
30230 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30232 * doc/install.texi: Add links to macOS binary distributions.
30234 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30238 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30240 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30242 2017-05-26 Martin Liska <mliska@suse.cz>
30244 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30246 2017-05-26 Martin Liska <mliska@suse.cz>
30248 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30249 always leading ';; '.
30250 (dump_bb_info): Likewise.
30251 (brief_dump_cfg): Likewise.
30252 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30253 * dumpfile.c: Remove usage of TDF_VERBOSE.
30254 * dumpfile.h (enum dump_kind): Likewise.
30255 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30256 * print-tree.c (debug_verbose): Remove.
30257 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30258 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30259 * tree-diagnostic.c (default_tree_printer): Replace
30260 TDF_DIAGNOSTIC with TDF_SLIM.
30262 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30264 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30265 in parameter loop, rather than loop_vinfo.
30266 (create_intersect_range_checks): Ditto.
30267 (vect_create_cond_for_alias_checks): Update call to above functions.
30269 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30271 PR tree-optimization/80815
30272 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30273 for merging runtime alias checks. Handle negative DR_STEPs.
30275 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30277 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30279 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30280 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30281 out code pruning runtime alias checks.
30282 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30283 factored out from above.
30284 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30286 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30288 (prune_runtime_alias_test_list): New decalaration.
30290 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30292 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30293 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30294 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30295 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30296 (operator==, comp_dr_with_seg_len_pair): Ditto.
30297 (vect_prune_runtime_alias_test_list): Ditto.
30299 2017-05-26 Martin Liska <mliska@suse.cz>
30302 * params.def: Bound partial-inlining-entry-probability param.
30304 2017-05-26 Marek Polacek <polacek@redhat.com>
30307 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30310 2017-05-26 Richard Biener <rguenther@suse.de>
30312 PR tree-optimization/80842
30313 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30316 2017-05-26 Richard Biener <rguenther@suse.de>
30318 PR tree-optimization/80844
30319 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30321 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30323 * doc/md.texi (Machine Constraints): Update x86 family
30324 machine constraints section to match 'config/i386/constraints.md'.
30326 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30328 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30330 2017-05-25 Nathan Sidwell <nathan@acm.org>
30332 * doc/invoke.texi (--enable-languages): Update documentation.
30334 2017-05-25 Martin Liska <mliska@suse.cz>
30336 * dumpfile.c: Add TDF_FOLDING.
30337 * dumpfile.h (enum dump_kind): Likewise.
30338 * genmatch.c (dt_simplify::gen_1): Use it.
30340 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30342 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30344 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30346 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30347 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30349 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30351 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30352 * match.pd (X == C): Rewrite it here.
30353 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30354 with_certain_nonzero_bits2): New predicates.
30355 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30357 2017-05-24 Nathan Sidwell <nathan@acm.org>
30359 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30362 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30365 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30367 * config/powerpcspe: New port. Files are copied from the rs6000
30368 port, with "rs6000" in filenames replaced by "powerpcspe".
30370 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30372 PR rtl-optimization/80754
30373 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30375 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30377 * config/sparc/sparc.md (length): Return the correct value for -mflat
30378 sibcalls to match output_sibcall.
30380 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30384 * config/rs6000/rs6000.c (struct machine_function): Add new field
30386 (rs6000_get_separate_components): Init that field, use it.
30387 (rs6000_components_for_bb): Use the field.
30389 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30391 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30393 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30395 PR middle-end/80823
30396 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30398 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30401 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30402 address against address_operand predicate.
30403 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30405 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30407 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30408 parameters passed indirectly.
30410 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30412 * config/i386/i386.md (*movdi_internal): Remove SSE4
30413 alternative 18 (?r, *v). Update insn attributes.
30414 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30415 Update insn attributes.
30416 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30417 Update insn attributes.
30418 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30419 alternative 1 (r, v). Remove isa attribute.
30420 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30421 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30422 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30424 2017-05-23 Tom de Vries <tom@codesourcery.com>
30426 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30429 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30431 * cgraphunit.c (symbol_table::process_new_functions): Update.
30432 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30433 (inline_generate_summary): Rename to ...
30434 (ipa_fn_summary_generate): ... this one.
30435 (inline_read_summary): Rename to ...
30436 (ipa_fn_summary_read): ... this one.
30437 (inline_write_summary): Rename to ...
30438 (ipa_fn_summary_write): ... this one.
30439 (inline_free_summary): Rename to ...
30440 (ipa_free_fn_summary): ... this one.
30441 (pass_data_local_fn_summary, pass_local_fn_summary,
30442 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30443 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30444 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30445 make_pass_ipa_fn_summary): New.
30446 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30447 inline_write_summary, inline_free_summary): Remove.
30448 (ipa_free_fn_summary) : New.
30449 * ipa-inline.c (ipa_inline): Update.
30450 (pass_ipa_inline): Do not generate summaries.
30451 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30453 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30454 and add pass_ipa_fn_summary.
30455 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30457 (make_pass_inline_parameters): Remove.
30459 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30461 * omp-low.c (struct omp_context): Remove "default_kind" member.
30464 * omp-offload.c (execute_oacc_device_lower): Remove the
30465 parallelism dimensions function attributes for unparallelized
30466 OpenACC kernels constructs.
30468 2017-05-23 Martin Liska <mliska@suse.cz>
30470 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30472 (cgraph_edge::make_speculative): Likewise.
30473 (cgraph_edge::resolve_speculation): Likewise.
30474 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30475 (cgraph_node::dump): Likewise.
30476 * cgraph.h: Likewise.
30477 * cgraphunit.c (analyze_functions): Likewise.
30478 (symbol_table::compile): Likewise.
30479 * ipa-cp.c (print_all_lattices): Likewise.
30480 (determine_versionability): Likewise.
30481 (initialize_node_lattices): Likewise.
30482 (ipcp_verify_propagated_values): Likewise.
30483 (estimate_local_effects): Likewise.
30484 (update_profiling_info): Likewise.
30485 (create_specialized_node): Likewise.
30486 (perhaps_add_new_callers): Likewise.
30487 (decide_about_value): Likewise.
30488 (decide_whether_version_node): Likewise.
30489 (identify_dead_nodes): Likewise.
30490 (ipcp_store_bits_results): Likewise.
30491 * ipa-devirt.c (dump_targets): Likewise.
30492 (ipa_devirt): Likewise.
30493 * ipa-icf.c (sem_item::dump): Likewise.
30494 (sem_function::equals): Likewise.
30495 (sem_variable::equals): Likewise.
30496 (sem_item_optimizer::read_section): Likewise.
30497 (sem_item_optimizer::execute): Likewise.
30498 (congruence_class::dump): Likewise.
30499 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30500 (dump_inline_summary): Likewise.
30501 (estimate_node_size_and_time): Likewise.
30502 (inline_analyze_function): Likewise.
30503 * ipa-inline-transform.c (inline_call): Likewise.
30504 * ipa-inline.c (report_inline_failed_reason): Likewise.
30505 (want_early_inline_function_p): Likewise.
30506 (edge_badness): Likewise.
30507 (update_edge_key): Likewise.
30508 (inline_small_functions): Likewise.
30509 * ipa-profile.c (ipa_profile): Likewise.
30510 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30511 (ipa_make_edge_direct_to_target): Likewise.
30512 (remove_described_reference): Likewise.
30513 (ipa_impossible_devirt_target): Likewise.
30514 (propagate_controlled_uses): Likewise.
30515 (ipa_print_node_params): Likewise.
30516 (ipcp_transform_function): Likewise.
30517 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30518 (propagate_pure_const): Likewise.
30519 * ipa-reference.c (generate_summary): Likewise.
30520 (read_write_all_from_decl): Likewise.
30521 (propagate): Likewise.
30522 (ipa_reference_read_optimization_summary): Likewise.
30523 * ipa-utils.c (ipa_merge_profiles): Likewise.
30524 * ipa.c (walk_polymorphic_call_targets): Likewise.
30525 (symbol_table::remove_unreachable_nodes): Likewise.
30526 (ipa_single_use): Likewise.
30527 * passes.c (execute_todo): Likewise.
30528 * predict.c (drop_profile): Likewise.
30529 * symtab.c (symtab_node::get_dump_name): New function.
30530 (symtab_node::dump_name): Likewise.
30531 (symtab_node::dump_asm_name): Likewise.
30532 (symtab_node::dump_references): Likewise.
30533 (symtab_node::dump_referring): Likewise.
30534 (symtab_node::dump_base): Likewise.
30535 (symtab_node::debug_symtab): Likewise.
30536 * tree-sra.c (convert_callers_for_node): Likewise.
30537 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30538 * value-prof.c (init_node_map): Likewise.
30540 2017-05-23 Martin Liska <mliska@suse.cz>
30542 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30543 and symtab_node::debug_symtab to symbol_table::debug.
30544 * cgraphunit.c (analyze_functions): Use the renamed function.
30545 (symbol_table::compile): Likewise.
30546 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30547 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30548 * passes.c (execute_todo): Likewise.
30549 * symtab.c (symbol_table::dump): New function.
30550 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30552 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30554 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30555 that nonconst implies exec.
30557 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30559 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30560 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30561 (inline_edge_summary_vec): Turn into ...
30562 (ipa_call_summaries): ... this one.
30563 (redirect_to_unreachable, edge_set_predicate,
30564 evaluate_properties_for_edge, inline_summary_alloc,
30565 reset_ipa_call_summary, reset_inline_summary,
30566 inline_summary_t::duplicate): Update.
30567 (inline_edge_duplication_hook): Turn to ...
30568 (ipa_call_summary_t::duplicate): ... this one.
30569 (inline_edge_removal_hook): Turn to ...
30570 (ipa_call_summary_t::remove): ... this one.
30571 (dump_inline_edge_summary): Turn to ...
30572 (dump_ipa_call_summary): ... this one.
30573 (estimate_function_body_sizes): Update.
30574 (inline_update_callee_summaries): Update.
30575 (remap_edge_change_prob): Update.
30576 (remap_edge_summaries): Update.
30577 (inline_merge_summary): Update.
30578 (do_estimate_edge_time): Update.
30579 (inline_generate_summary): Update.
30580 (inline_read_section): Update.
30581 (inline_read_summary): Update.
30582 (inline_free_summary): Update.
30583 * ipa-inline.c (can_inline_edge_p): Update.
30584 (compute_inlined_call_time): Update.
30585 (want_inline_small_function_p): Update.
30586 (edge_badness): Update.
30587 (early_inliner): Update.
30588 * ipa-inline.h (inline_edge_summary): Turn to ...
30589 (ipa_call_summary): ... this one.
30590 (ipa_call_summary_t): New class.
30591 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30592 (ipa_call_summaries): New.
30593 (inline_edge_summary): Remove.
30594 (estimate_edge_growth): Update.
30595 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30596 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30597 * ipa-split.c (execute_split_functions): Update.
30598 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30600 2017-05-23 Tom de Vries <tom@codesourcery.com>
30602 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30603 attributes): Document rdrand effective target.
30605 2017-05-23 Tom de Vries <tom@codesourcery.com>
30607 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30608 attributes): Sort alphabetically.
30610 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30612 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30614 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30617 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30618 V2DF/V2DI splat into two separate patterns, one that handles
30619 registers, and the other that only handles memory. Drop support
30620 for splatting from a GPR on ISA 2.07 and then splitting the
30621 splat into direct move and splat.
30622 (vsx_splat_<mode>_reg): Likewise.
30623 (vsx_splat_<mode>_mem): Likewise.
30625 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30627 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30629 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30631 PR middle-end/80809
30632 * omp-low.c (finish_taskreg_remap): New function.
30633 (finish_taskreg_scan): If unit size of ctx->record_type
30634 is non-constant, unshare the size expression and replace
30635 decls in it with possible outer var refs.
30637 PR middle-end/80809
30638 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30639 GOVD_SHARED rather than GOVD_PRIVATE with it.
30640 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30641 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30643 PR middle-end/80853
30644 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30645 as last argument to build_outer_var_ref for pointer bases of array
30646 section reductions.
30648 2017-05-19 Martin Sebor <msebor@redhat.com>
30650 * print-tree.c (print_node): Print DECL_READ_P flag.
30652 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30654 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30655 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30656 * cgraph.c: Likewise.
30657 * cgraphunit.c: Likewise.
30658 * gengtype.c: Likewise.
30659 * ipa-cp.c: Likewise.
30660 * ipa-devirt.c: Likewise.
30661 * ipa-icf.c: Likewise.
30662 * ipa-predicate.c: Likewise.
30663 * ipa-profile.c: Likewise.
30664 * ipa-prop.c: Likewise.
30665 * ipa-split.c: Likewise.
30667 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30668 edge_predicate_pool, dump_inline_hints,
30669 inline_summary::account_size_time, redirect_to_unreachable,
30670 edge_set_predicate, set_hint_predicate,
30671 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30672 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30673 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30674 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30675 ipa_call_summary_t::remove, initialize_growth_caches,
30676 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30677 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30678 mark_modified, unmodified_parm_1, unmodified_parm,
30679 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30680 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30681 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30682 will_be_nonconstant_predicate, record_modified_bb_info,
30683 get_minimal_bb, record_modified, param_change_prob,
30684 phi_result_unknown_predicate, predicate_for_phi_result,
30685 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30686 estimate_function_body_sizes, compute_inline_parameters,
30687 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30688 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30689 inline_update_callee_summaries, remap_edge_change_prob,
30690 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30691 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30692 inline_analyze_function, inline_summary_t::insert,
30693 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30694 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30695 inline_free_summary): Move to ipa-fnsummary.h
30696 (predicate_t): Remove.
30697 * ipa-fnsummary.c: New file.
30698 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30699 (enum inline_hints_vals, inline_hints, agg_position_info,
30700 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30701 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30702 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30703 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30704 inline_read_summary, inline_write_summary, inline_free_summary,
30705 inline_analyze_function, initialize_inline_failed,
30706 inline_merge_summary, inline_update_overall_summary,
30707 compute_inline_parameters): Move to ipa-fnsummary.h
30708 * ipa-fnsummary.h: New file.
30709 * ipa-inline-transform.h: Include ipa-inline.h.
30710 * ipa-inline.c: LIkewise.
30712 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30714 * ipa-inline.c (edge_badness): Use inlined_time instead of
30715 inline_summaries->get.
30717 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30719 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30721 2017-05-22 Nathan Sidwell <nathan@acm.org>
30723 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30724 (fdump-lang): Document 'raw' option.
30725 * dumpfile.h (TDI_tu): Delete.
30726 * dumpfile.c (dump_files): Remove translation-unit.
30727 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30729 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30731 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30732 command option from $(AWK) call.
30733 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30734 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30735 [FORMAT]: Remove handling of variable.
30736 * config/avr/t-multilib: Regenerate.
30738 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30740 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30742 (dump_inline_summary): Do not print self_time.
30743 (estimate_function_body_sizes): Do not set self_time.
30744 (compute_inline_parameters): Likewise.
30745 (inline_read_section, inline_write_summary): Do not stream self_time.
30746 * ipa-inline.h (inline_summary): Drop self_time.
30748 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30750 * ipa-inline-analysis.c (account_size_time): Rename to ...
30751 (inline_summary::account_size_time): ... this one.
30752 (reset_ipa_call_summary): Turn to ...
30753 (ipa_call_summary::reset): ... this one.
30754 (reset_inline_summary): Turn to ...
30755 (inline_summary::reset): ... this one.
30756 (inline_summary_t::remove): Update.
30757 (inline_summary_t::duplicate): Update.
30758 (ipa_call_summary_t::remove): Update.
30759 (dump_inline_summary): Update.
30760 (estimate_function_body_sizes): Update.
30761 (compute_inline_parameters): Update.
30762 (estimate_node_size_and_time): Update.
30763 (inline_merge_summary): Update.
30764 (inline_update_overall_summary): Update.
30765 (inline_read_section): Update.
30766 (inline_write_summary): Update.
30767 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30768 add account_size_time and reset member functions.
30769 (ipa_call_summary): Add reset function.
30770 * ipa-predicate.h (predicate::operator &): Constify.
30772 2017-05-22 Richard Biener <rguenther@suse.de>
30774 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30776 2017-05-19 Jason Merrill <jason@redhat.com>
30778 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30780 2017-05-19 Marek Polacek <polacek@redhat.com>
30783 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30784 TYPE_OVERFLOW_WRAPS checks.
30786 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30788 * tree-core.h (enum omp_clause_default_kind): Add
30789 "OMP_CLAUSE_DEFAULT_PRESENT".
30790 * tree-pretty-print.c (dump_omp_clause): Handle it.
30791 * gimplify.c (enum gimplify_omp_var_data): Add
30792 "GOVD_MAP_FORCE_PRESENT".
30793 (gimplify_adjust_omp_clauses_1): Map it to
30794 "GOMP_MAP_FORCE_PRESENT".
30795 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30797 * gimplify.c (oacc_default_clause): Clarify.
30799 2017-05-19 Nathan Sidwell <nathan@acm.org>
30801 LANG_HOOK_REGISTER_DUMPS
30802 * toplev.c (general_init): Call register dump lang hook.
30803 * doc/invoke.texi: Document -fdump-lang option family.
30804 * dumpfile.c (dump_files): Remove class dump here.
30805 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30806 * dumpfile.h (tree_dump_index): Remove TDI_class.
30807 * langhooks-def.h (lhd_register_dumps): Declare.
30808 (LANG_HOOKS_REGISTER_DUMPS): Define.
30809 (LANG_HOOKS_INITIALIZER): Add it.
30810 * langhooks.c (lhd_register_dumps): Define.
30811 * langhooks.h (struct lang_hooks): Add register_dumps.
30813 2017-05-19 Nathan Sidwell <nathan@acm.org>
30815 * context.h (context::set_passes): New.
30816 * context.c (context::context): Do not create pass manager.
30817 * toplev.c (general_init): Create pass manager here.
30819 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30821 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30822 use this splitter if two add or or instructions would also work for
30823 the constant we want to generate.
30825 2017-05-19 Richard Biener <rguenther@suse.de>
30828 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30829 predicate evaluation.
30831 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30833 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30835 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30838 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30840 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30841 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30842 (inline_edge_summary_vec): Turn into ...
30843 (ipa_call_summaries): ... this one.
30844 (redirect_to_unreachable, edge_set_predicate,
30845 evaluate_properties_for_edge, inline_summary_alloc,
30846 reset_ipa_call_summary, reset_inline_summary,
30847 inline_summary_t::duplicate): Update.
30848 (inline_edge_duplication_hook): Turn to ...
30849 (ipa_call_summary_t::duplicate): ... this one.
30850 (inline_edge_removal_hook): Turn to ...
30851 (ipa_call_summary_t::remove): ... this one.
30852 (dump_inline_edge_summary): Turn to ...
30853 (dump_ipa_call_summary): ... this one.
30854 (estimate_function_body_sizes): Update.
30855 (inline_update_callee_summaries): Update.
30856 (remap_edge_change_prob): Update.
30857 (remap_edge_summaries): Update.
30858 (inline_merge_summary): Update.
30859 (do_estimate_edge_time): Update.
30860 (inline_generate_summary): Update.
30861 (inline_read_section): Update.
30862 (inline_read_summary): Update.
30863 (inline_free_summary): Update.
30864 * ipa-inline.c (can_inline_edge_p): Update.
30865 (compute_inlined_call_time): Update.
30866 (want_inline_small_function_p): Update.
30867 (edge_badness): Update.
30868 (early_inliner): Update.
30869 * ipa-inline.h (inline_edge_summary): Turn to ...
30870 (ipa_call_summary): ... this one.
30871 (ipa_call_summary_t): New class.
30872 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30873 (ipa_call_summaries): New.
30874 (inline_edge_summary): Remove.
30875 (estimate_edge_growth): Update.
30876 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30877 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30878 * ipa-split.c (execute_split_functions): Update.
30879 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30881 2017-05-19 Richard Biener <rguenther@suse.de>
30883 PR middle-end/80764
30884 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30886 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30888 * config/rs6000/rs6000.c (struct machine_function): Add field
30889 fpr_is_wrapped_separately.
30890 (rs6000_get_separate_components): Use 64 components. Handle the
30891 new FPR components.
30892 (rs6000_components_for_bb): Handle the FPR components.
30893 (rs6000_emit_prologue_components): Handle the FPR components.
30894 (rs6000_emit_epilogue_components): Handle the FPR components.
30895 (rs6000_set_handled_components): Handle the FPR components.
30896 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30897 that are already separately shrink-wrapped.
30898 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30899 that are already separately shrink-wrapped.
30901 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30904 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30907 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30908 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30909 cases where the register allocator uses a move and an offsettable
30910 memory operation to/from a FPR register on ISA 2.06/2.07.
30911 (define_peephole2 for Altivec d-form store): Likewise.
30913 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30916 * config/i386/mmx.md (*mov<mode>_internal): Enable
30917 alternatives 11, 12, 13 and 14 also for 32bit targets.
30918 Remove alternatives 15, 16, 17 and 18.
30919 * config/i386/sse.md (vec_concatv2di): Change
30920 alternative (!x, *y) to (x, ?!*Yn).
30922 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30924 * dumpfile.h (enum dump_kind): Remove stray comma.
30926 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30928 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30929 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30930 predicate::num_conditions
30931 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30932 (CHANGED): turn into predicate::changed.
30933 (agg_position_info): Move to ipa-predicate.h
30934 (add_condition, predicate::add_clause, predicate::operator &=,
30935 predicate::or_with, predicate::evaluate, predicate::probability,
30936 dump_condition, dump_clause, predicate::dump,
30937 predicate::remap_after_duplication, predicate::remap_after_inlining,
30938 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30939 (evaluate_conditions_for_known_args): Update.
30940 (set_cond_stmt_execution_predicate): Update.
30941 * ipa-inline.h: Include ipa-predicate.h
30942 (condition, inline_param_summary, conditions, agg_position_info,
30943 predicate): Move to ipa-predicate.h
30944 * ipa-predicate.c: New file.
30945 * ipa-predicate.h: New file.
30947 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30949 * final.c (leaf_function_p): Check we are not in a sequence.
30951 2017-05-18 Martin Liska <mliska@suse.cz>
30953 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30954 * dumpfile.c (dump_register): Use new enum dump_kind.
30955 (get_dump_file_name): Likewise.
30956 (dump_enable_all): Likewise.
30957 (dump_switch_p_1): Likewise.
30958 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30959 * dumpfile.h (enum dump_kind): New enum type.
30960 (struct dump_file_info): Create constructor and
30961 format fields and comments.
30962 * passes.c (pass_manager::register_one_dump_file):
30964 * statistics.c (statistics_early_init): Likewise.
30965 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30966 TDF_TREE with TDF_SLIM.
30967 (gather_memory_references_ref): Likewise.
30969 2017-05-18 Martin Liska <mliska@suse.cz>
30971 * vec.h (struct vnull): Use it.
30973 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30975 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30976 (true_predicate, false_predicate, true_predicate_p,
30977 false_predicate_p): Remove.
30978 (single_cond_predicate, not_inlined_predicate): Turn to member function
30980 (add_condition): Update.
30981 (add_clause): Turn to...
30982 (predicate::add_clause): ... this one; update; allow passing NULL
30984 (and_predicates): Turn to ...
30985 (predicate::operator &=): ... this one.
30986 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30987 (or_predicates): Turn to ...
30988 (predicate::or_with): ... this one.
30989 (evaluate_predicate): Turn to ...
30990 (predicate::evaluate): ... this one.
30991 (predicate_probability): Turn to ...
30992 (predicate::probability): ... this one.
30993 (dump_condition): Update.
30994 (dump_predicate): Turn to ...
30995 (predicate::dump): ... this one.
30996 (account_size_time): Update.
30997 (edge_set_predicate): Update.
30998 (set_hint_predicate): UPdate.
30999 (evaluate_conditions_for_known_args): Update.
31000 (evaluate_properties_for_edge): Update.
31001 (remap_predicate_after_duplication): Turn to...
31002 (predicate::remap_after_duplication): ... this one.
31003 (remap_hint_predicate_after_duplication): Update.
31004 (inline_summary_t::duplicate): UPdate.
31005 (dump_inline_edge_summary): Update.
31006 (dump_inline_summary): Update.
31007 (set_cond_stmt_execution_predicate): Update.
31008 (set_switch_stmt_execution_predicate): Update.
31009 (compute_bb_predicates): Update.
31010 (will_be_nonconstant_expr_predicate): Update.
31011 (will_be_nonconstant_predicate): Update.
31012 (phi_result_unknown_predicate): Update.
31013 (predicate_for_phi_result): Update.
31014 (array_index_predicate): Update.
31015 (estimate_function_body_sizes): Update.
31016 (estimate_node_size_and_time): Update.
31017 (estimate_ipcp_clone_size_and_time): Update.
31018 (remap_predicate): Rename to ...
31019 (predicate::remap_after_inlining): ... this one.
31020 (remap_hint_predicate): Update.
31021 (inline_merge_summary): Update.
31022 (inline_update_overall_summary): Update.
31023 (estimate_size_after_inlining): Update.
31024 (read_predicate): Rename to ...
31025 (predicate::stream_in): ... this one.
31026 (read_inline_edge_summary): Update.
31027 (write_predicate): Rename to ...
31028 (predicate::stream_out): ... this one.
31029 (write_inline_edge_summary): Update.
31030 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31031 (clause_t): Turn to uint32_t
31032 (predicate): Turn to class; implement constructor and operators
31034 (size_time_entry): Update.
31035 (inline_summary): Update.
31036 (inline_edge_summary): Update.
31038 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31040 * fold-const.c (fold_binary_loc): Move transformation...
31041 * match.pd (C - X CMP X): ... here.
31043 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31045 * config/sparc/sparc.c (sparc_option_override): Set function
31046 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31047 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31049 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31051 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31053 2017-05-18 Marek Polacek <polacek@redhat.com>
31056 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31057 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31058 gimple_assign_load_p.
31060 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31062 PR middle-end/80692
31063 * real.c (do_compare): Give decimal_do_compare preference over
31064 comparing just the signs.
31066 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31068 * doc/md.texi (Canonicalization of Instructions): Describe the
31069 canonical form of instructions that inherently set a condition
31072 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31074 PR middle-end/80775
31075 * tree-cfg.c: Move deletion of unreachable case statements to after
31076 the merging of consecutive case labels.
31078 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31080 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31081 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31082 restoring of callee-saved registers.
31084 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31086 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31087 * config/visium/visium.c (single_set_and_flags): Likewise.
31088 * config/visium/visium.md (Substitutions): Likewise.
31090 2017-05-17 Martin Liska <mliska@suse.cz>
31092 * cfg.c: Introduce dump_flags_t type and
31093 use it instead of int type.
31095 * cfghooks.c: Likewise.
31096 * cfghooks.h (struct cfg_hooks): Likewise.
31097 * cfgrtl.c: Likewise.
31098 * cfgrtl.h: Likewise.
31099 * cgraph.c (cgraph_node::get_body): Likewise.
31100 * coretypes.h: Likewise.
31101 * domwalk.c: Likewise.
31102 * domwalk.h: Likewise.
31103 * dumpfile.c (struct dump_option_value_info): Likewise.
31104 (dump_enable_all): Likewise.
31105 (dump_switch_p_1): Likewise.
31106 (opt_info_switch_p): Likewise.
31107 * dumpfile.h (enum tree_dump_index): Likewise.
31108 (struct dump_file_info): Likewise.
31109 * genemit.c: Likewise.
31110 * generic-match-head.c: Likewise.
31111 * gengtype.c (open_base_files): Likewise.
31112 * gimple-pretty-print.c: Likewise.
31113 * gimple-pretty-print.h: Likewise.
31114 * graph.c (print_graph_cfg): Likewise.
31115 * graphite-scop-detection.c (dot_all_sese): Likewise.
31116 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31117 * loop-unroll.c (report_unroll): Likewise.
31118 * passes.c (pass_manager::register_one_dump_file): Likewise.
31119 * print-tree.c: Likewise.
31120 * statistics.c: Likewise.
31121 * tree-cfg.c: Likewise.
31122 * tree-cfg.h: Likewise.
31123 * tree-dfa.c: Likewise.
31124 * tree-dfa.h: Likewise.
31125 * tree-dump.c (dump_function): Likewise.
31126 * tree-dump.h (struct dump_info): Likewise.
31127 * tree-pretty-print.c: Likewise.
31128 * tree-pretty-print.h: Likewise.
31129 * tree-ssa-live.c: Likewise.
31130 * tree-ssa-live.h: Likewise.
31131 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31132 * tree-vect-loop.c: Likewise.
31133 * tree-vect-slp.c: Likewise.
31135 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31136 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31138 PR tree-optimization/80457
31139 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31140 of all arguments to a statement as scalar_to_vec operations.
31141 (vectorizable_call): Adjust call to vect_model_simple_cost for
31143 (vectorizable_conversion): Likewise.
31144 (vectorizable_assignment): Likewise.
31145 (vectorizable_shift): Likewise.
31146 (vectorizable_operation): Likewise.
31147 (vectorizable_comparison): Likewise.
31148 (vect_is_simple_cond): Record the def types for operands.
31149 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31150 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31151 for statement argument count.
31153 2017-05-16 Carl Love <cel@us.ibm.com>
31155 * config/rs6000/rs6000-c: Add support for built-in functions
31156 vector unsigned long long vec_bperm (vector unsigned long long,
31157 vector unsigned char)
31158 vector signed long long vec_mule (vector signed int,
31160 vector unsigned long long vec_mule (vector unsigned int,
31161 vector unsigned int)
31162 vector signed long long vec_mulo (vector signed int,
31164 vector unsigned long long vec_mulo (vector unsigned int,
31165 vector unsigned int)
31166 vector signed char vec_sldw (vector signed char,
31167 vector signed char,
31169 vector unsigned char vec_sldw (vector unsigned char,
31170 vector unsigned char,
31172 vector signed short vec_sldw (vector signed short,
31173 vector signed short,
31175 vector unsigned short vec_sldw (vector unsigned short,
31176 vector unsigned short,
31178 vector signed int vec_sldw (vector signed int,
31181 vector unsigned int vec_sldw (vector unsigned int,
31182 vector unsigned int,
31184 vector signed long long vec_sldw (vector signed long long,
31185 vector signed long long,
31187 vector unsigned long long vec_sldw (vector unsigned long long,
31188 vector unsigned long long,
31190 * config/rs6000/rs6000-c: Add support for built-in functions
31191 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31192 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31193 * doc/extend.texi: Update the built-in documentation for the
31194 new built-in functions.
31196 2017-05-16 Marek Polacek <polacek@redhat.com>
31200 * tree.c (save_expr): Don't fold the expression.
31202 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31204 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31205 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31206 and (?*y,m). Update insn attributes.
31208 2017-05-16 Martin Liska <mliska@suse.cz>
31210 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31211 flags argument of print_gimple_stmt, print_gimple_expr,
31212 print_generic_stmt and print_generic_expr.
31213 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31214 * coretypes.h: Likewise.
31215 * except.c (dump_eh_tree): Likewise.
31216 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31217 * gimple-pretty-print.h: Likewise.
31218 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31219 (backprop::push_to_worklist): Likewise.
31220 (backprop::pop_from_worklist): Likewise.
31221 (backprop::process_use): Likewise.
31222 (backprop::intersect_uses): Likewise.
31223 (note_replacement): Likewise.
31224 * gimple-ssa-store-merging.c
31225 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31226 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31227 (pass_store_merging::execute): Likewise.
31228 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31229 (ssa_base_cand_dump_callback): Likewise.
31230 (dump_incr_vec): Likewise.
31231 (replace_refs): Likewise.
31232 (replace_mult_candidate): Likewise.
31233 (create_add_on_incoming_edge): Likewise.
31234 (create_phi_basis): Likewise.
31235 (insert_initializers): Likewise.
31236 (all_phi_incrs_profitable): Likewise.
31237 (introduce_cast_before_cand): Likewise.
31238 (replace_one_candidate): Likewise.
31239 * gimplify.c (gimplify_expr): Likewise.
31240 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31241 (set_rename): Likewise.
31242 (rename_uses): Likewise.
31243 (copy_loop_phi_nodes): Likewise.
31244 (add_close_phis_to_merge_points): Likewise.
31245 (copy_loop_close_phi_args): Likewise.
31246 (copy_cond_phi_args): Likewise.
31247 (graphite_copy_stmts_from_block): Likewise.
31248 (translate_pending_phi_nodes): Likewise.
31249 * graphite-poly.c (print_pdr): Likewise.
31250 (dump_gbb_cases): Likewise.
31251 (dump_gbb_conditions): Likewise.
31252 (print_scop_params): Likewise.
31253 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31254 (build_cross_bb_scalars_use): Likewise.
31255 (gather_bbs::before_dom_children): Likewise.
31256 * hsa-dump.c (dump_hsa_immed): Likewise.
31257 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31258 (get_replacement_map): Likewise.
31259 * ipa-inline-analysis.c (dump_condition): Likewise.
31260 (estimate_function_body_sizes): Likewise.
31261 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31262 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31263 * ipa-prop.c (ipa_dump_param): Likewise.
31264 (ipa_print_node_jump_functions_for_edge): Likewise.
31265 (ipa_modify_call_arguments): Likewise.
31266 (ipa_modify_expr): Likewise.
31267 (ipa_dump_param_adjustments): Likewise.
31268 (ipa_dump_agg_replacement_values): Likewise.
31269 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31270 * ipa-pure-const.c (check_stmt): Likewise.
31271 (pass_nothrow::execute): Likewise.
31272 * ipa-split.c (execute_split_functions): Likewise.
31273 * omp-offload.c (dump_oacc_loop_part): Likewise.
31274 (dump_oacc_loop): Likewise.
31275 * trans-mem.c (tm_log_emit): Likewise.
31276 (tm_memopt_accumulate_memops): Likewise.
31277 (dump_tm_memopt_set): Likewise.
31278 (dump_tm_memopt_transform): Likewise.
31279 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31280 (print_loop): Likewise.
31281 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31282 (chkp_gather_checks_info): Likewise.
31283 (chkp_get_check_result): Likewise.
31284 (chkp_remove_check_if_pass): Likewise.
31285 (chkp_use_outer_bounds_if_possible): Likewise.
31286 (chkp_reduce_bounds_lifetime): Likewise.
31287 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31288 (chkp_mark_completed_bounds): Likewise.
31289 (chkp_register_incomplete_bounds): Likewise.
31290 (chkp_mark_invalid_bounds): Likewise.
31291 (chkp_maybe_copy_and_register_bounds): Likewise.
31292 (chkp_build_returned_bound): Likewise.
31293 (chkp_get_bound_for_parm): Likewise.
31294 (chkp_build_bndldx): Likewise.
31295 (chkp_get_bounds_by_definition): Likewise.
31296 (chkp_generate_extern_var_bounds): Likewise.
31297 (chkp_get_bounds_for_decl_addr): Likewise.
31298 * tree-chrec.c (chrec_apply): Likewise.
31299 * tree-data-ref.c (dump_data_reference): Likewise.
31300 (dump_subscript): Likewise.
31301 (dump_data_dependence_relation): Likewise.
31302 (analyze_overlapping_iterations): Likewise.
31303 * tree-inline.c (expand_call_inline): Likewise.
31304 (tree_function_versioning): Likewise.
31305 * tree-into-ssa.c (dump_defs_stack): Likewise.
31306 (dump_currdefs): Likewise.
31307 (dump_names_replaced_by): Likewise.
31308 (dump_update_ssa): Likewise.
31309 (update_ssa): Likewise.
31310 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31311 * tree-parloops.c (build_new_reduction): Likewise.
31312 (try_create_reduction_list): Likewise.
31313 (ref_conflicts_with_region): Likewise.
31314 (oacc_entry_exit_ok_1): Likewise.
31315 (oacc_entry_exit_single_gang): Likewise.
31316 * tree-pretty-print.h: Likewise.
31317 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31318 (get_scalar_evolution): Likewise.
31319 (add_to_evolution): Likewise.
31320 (get_loop_exit_condition): Likewise.
31321 (analyze_evolution_in_loop): Likewise.
31322 (analyze_initial_condition): Likewise.
31323 (analyze_scalar_evolution): Likewise.
31324 (instantiate_scev): Likewise.
31325 (number_of_latch_executions): Likewise.
31326 (gather_chrec_stats): Likewise.
31327 (final_value_replacement_loop): Likewise.
31328 (scev_const_prop): Likewise.
31329 * tree-sra.c (dump_access): Likewise.
31330 (disqualify_candidate): Likewise.
31331 (create_access): Likewise.
31332 (reject): Likewise.
31333 (maybe_add_sra_candidate): Likewise.
31334 (create_access_replacement): Likewise.
31335 (analyze_access_subtree): Likewise.
31336 (analyze_all_variable_accesses): Likewise.
31337 (sra_modify_assign): Likewise.
31338 (initialize_constant_pool_replacements): Likewise.
31339 (find_param_candidates): Likewise.
31340 (decide_one_param_reduction): Likewise.
31341 (replace_removed_params_ssa_names): Likewise.
31342 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31343 * tree-ssa-copy.c (dump_copy_of): Likewise.
31344 (copy_prop_visit_cond_stmt): Likewise.
31345 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31346 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31347 (record_equivalences_from_stmt): Likewise.
31348 * tree-ssa-dse.c (compute_trims): Likewise.
31349 (delete_dead_call): Likewise.
31350 (delete_dead_assignment): Likewise.
31351 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31352 (forward_propagate_into_cond): Likewise.
31353 (pass_forwprop::execute): Likewise.
31354 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31355 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31357 (move_computations_worker): Likewise.
31358 (execute_sm): Likewise.
31359 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31360 (remove_exits_and_undefined_stmts): Likewise.
31361 (remove_redundant_iv_tests): Likewise.
31362 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31363 (adjust_iv_update_pos): Likewise.
31364 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31365 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31366 (value_replacement): Likewise.
31367 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31368 * tree-ssa-pre.c (print_pre_expr): Likewise.
31369 (get_representative_for): Likewise.
31370 (create_expression_by_pieces): Likewise.
31371 (insert_into_preds_of_block): Likewise.
31372 (eliminate_insert): Likewise.
31373 (eliminate_dom_walker::before_dom_children): Likewise.
31374 (eliminate): Likewise.
31375 (remove_dead_inserted_code): Likewise.
31376 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31377 * tree-ssa-reassoc.c (get_rank): Likewise.
31378 (eliminate_duplicate_pair): Likewise.
31379 (eliminate_plus_minus_pair): Likewise.
31380 (eliminate_not_pairs): Likewise.
31381 (undistribute_ops_list): Likewise.
31382 (eliminate_redundant_comparison): Likewise.
31383 (update_range_test): Likewise.
31384 (optimize_range_tests_var_bound): Likewise.
31385 (optimize_vec_cond_expr): Likewise.
31386 (rewrite_expr_tree): Likewise.
31387 (rewrite_expr_tree_parallel): Likewise.
31388 (linearize_expr): Likewise.
31389 (break_up_subtract): Likewise.
31390 (linearize_expr_tree): Likewise.
31391 (attempt_builtin_powi): Likewise.
31392 (attempt_builtin_copysign): Likewise.
31393 (transform_stmt_to_copy): Likewise.
31394 (transform_stmt_to_multiply): Likewise.
31395 (dump_ops_vector): Likewise.
31396 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31397 (print_scc): Likewise.
31398 (set_ssa_val_to): Likewise.
31399 (visit_reference_op_store): Likewise.
31400 (visit_use): Likewise.
31401 (sccvn_dom_walker::before_dom_children): Likewise.
31402 (run_scc_vn): Likewise.
31403 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31405 (expr_hash_elt::print): Likewise.
31406 (const_and_copies::pop_to_marker): Likewise.
31407 (const_and_copies::record_const_or_copy_raw): Likewise.
31408 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31409 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31410 (dump_predicates): Likewise.
31411 (find_uninit_use): Likewise.
31412 (warn_uninitialized_phi): Likewise.
31413 (pass_late_warn_uninitialized::execute): Likewise.
31414 * tree-ssa.c (verify_vssa): Likewise.
31415 (verify_ssa): Likewise.
31416 (maybe_optimize_var): Likewise.
31417 * tree-vrp.c (dump_value_range): Likewise.
31418 (dump_all_value_ranges): Likewise.
31419 (dump_asserts_for): Likewise.
31420 (register_edge_assert_for_2): Likewise.
31421 (vrp_visit_cond_stmt): Likewise.
31422 (vrp_visit_switch_stmt): Likewise.
31423 (vrp_visit_stmt): Likewise.
31424 (vrp_visit_phi_node): Likewise.
31425 (simplify_cond_using_ranges_1): Likewise.
31426 (fold_predicate_in): Likewise.
31427 (evrp_dom_walker::before_dom_children): Likewise.
31428 (evrp_dom_walker::push_value_range): Likewise.
31429 (evrp_dom_walker::pop_value_range): Likewise.
31430 (execute_early_vrp): Likewise.
31432 2017-05-16 Richard Biener <rguenther@suse.de>
31434 * dwarf2out.c (loc_list_from_tree_1): Do not create
31435 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31437 2017-05-16 Richard Biener <rguenther@suse.de>
31439 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31441 (note_variable_value_in_expr): If we resolved the decl ref
31442 do not push to the stack.
31444 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31446 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31447 operations in fast-math mode.
31448 (vaddq_f16): Likewise.
31449 (vmul_f16): Likewise.
31450 (vmulq_f16): Likewise.
31451 (vsub_f16): Likewise.
31452 (vsubq_f16): Likewise.
31453 * config/arm/neon.md (add<mode>3): New.
31455 (fma:<VH:mode>3): New. Also remove outdated comment.
31458 2017-05-16 Martin Liska <mliska@suse.cz>
31462 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31463 (odr_types_equivalent_p): Likewise.
31465 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31467 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31469 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31472 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31473 non-interunit SSE move alternatives with '?'.
31474 (zero-extendsidi peephole2): New peephole to skip intermediate
31475 general register in SSE zero-extend sequence.
31477 2017-05-15 Jeff Law <law@redhat.com>
31479 * reorg.c (relax_delay_slots): Create a new variable to hold
31480 the temporary target rather than clobbering TARGET_LABEL.
31482 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31483 missing argument to extract_bit_field call.
31484 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31486 2017-05-15 Martin Liska <mliska@suse.cz>
31489 * gcc.c (process_command): Do not allow empty argument of -o option.
31491 2017-05-15 Renlin Li <renlin.li@arm.com>
31493 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31494 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31495 * config/aarch64/constraints.md (Usf): Add long call check.
31496 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31497 (call_value): Likewise.
31498 (sibcall): Likewise.
31499 (sibcall_value): Likewise.
31501 (call_value_insn): New.
31502 (sibcall_insn): Update rtx pattern.
31503 (sibcall_value_insn): Likewise.
31504 (call_internal): Remove.
31505 (call_value_internal): Likewise.
31506 (sibcall_internal): Likewise.
31507 (sibcall_value_internal): Likewise.
31508 (call_reg): Likewise.
31509 (call_symbol): Likewise.
31510 (call_value_reg): Likewise.
31511 (call_value_symbol): Likewise.
31513 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31516 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31518 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31520 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31521 compatible with CCGOCmode and with CCZmode.
31523 2017-05-14 Martin Sebor <msebor@redhat.com>
31525 PR middle-end/77671
31526 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31527 (gimple_fold_builtin_snprintf): Same.
31528 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31529 (gimple_fold_builtin_snprintf): Same.
31530 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31531 of character types.
31532 (is_call_safe): New function.
31533 (try_substitute_return_value): Call it.
31534 (try_simplify_call): New function.
31535 (pass_sprintf_length::handle_gimple_call): Call it.
31537 2017-05-14 Martin Sebor <msebor@redhat.com>
31539 PR middle-end/80669
31540 * builtins.c (expand_builtin_stpncpy): Simplify.
31542 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31544 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31545 * config/i386/i386.h
31546 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31547 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31548 (struct machine_function): Add new members call_ms2sysv,
31549 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31550 (struct machine_frame_state): New fields sp_realigned and
31551 sp_realigned_offset.
31552 * config/i386/i386.c
31553 (enum xlogue_stub): New enum.
31554 (enum xlogue_stub_sets): New enum.
31555 (class xlogue_layout): New class.
31556 (struct ix86_frame): New fields stack_realign_allocate_offset,
31557 stack_realign_offset and outlined_save_offset. Modify comments to
31558 detail stack layout when using out-of-line stubs.
31559 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31560 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31561 -mcall-ms2sysv-xlogues.
31562 (stub_managed_regs): New static variable.
31563 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31564 registers managed by out-of-line stub.
31565 (disable_call_ms2sysv_xlogues): New function.
31566 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31567 m->call_ms2sysv when appropriate and compute frame layout for
31569 (sp_valid_at, fp_valid_at): New inline functions.
31570 (choose_basereg): New function.
31571 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31573 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31574 Use align parameter of choose_baseaddr to generated aligned SSE movs
31576 (pro_epilogue_adjust_stack): Modify to track
31577 machine_frame_state::sp_realigned.
31578 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31579 (ix86_nsaved_sseregs): Likewise.
31580 (ix86_emit_save_regs): Likewise.
31581 (ix86_emit_save_regs_using_mov): Likewise.
31582 (ix86_emit_save_sse_regs_using_mov): Likewise.
31583 (get_scratch_register_on_entry): Likewise.
31584 (gen_frame_set): New function.
31585 (gen_frame_load): Likewise.
31586 (gen_frame_store): Likewise.
31587 (emit_outlined_ms2sysv_save): Likewise.
31588 (emit_outlined_ms2sysv_restore): Likewise.
31589 (ix86_expand_prologue): Modify stack re-alignment code and call
31590 emit_outlined_ms2sysv_save when appropriate.
31591 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31592 parameter rtx_insn *insn, which allows the function to be used to only
31593 generate the notes.
31594 (ix86_expand_epilogue): Modify validity checks of frame and stack
31595 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31596 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31597 * config/i386/predicates.md
31598 (save_multiple): New predicate.
31599 (restore_multiple): Likewise.
31600 * config/i386/sse.md
31601 (save_multiple<mode>): New pattern.
31602 (save_multiple_realign<mode>): Likewise.
31603 (restore_multiple<mode>): Likewise.
31604 (restore_multiple_and_return<mode>): Likewise.
31605 (restore_multiple_leave_return<mode>): Likewise.
31606 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31608 2017-05-14 Julia Koval <julia.koval@intel.com>
31610 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31611 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31612 (__builtin_ia32_xsetbv): New builtins.
31613 * config/i386/i386.c (ix86_expand_special_args_builtin):
31615 (ix86_expand_builtin): Special expand for new intrinsics.
31616 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31617 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31618 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31620 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31622 * cfganal.c (inverted_post_order_compute): Change argument type
31624 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31625 * df-core.c (rest_of_handle_df_initialize): Adjust.
31626 (rest_of_handle_df_finish): Likewise.
31627 (df_analyze_1): Likewise.
31628 (df_analyze): Likewise.
31629 (loop_inverted_post_order_compute): Change argument to be a vec *.
31630 (df_analyze_loop): Adjust.
31631 (df_get_n_blocks): Likewise.
31632 (df_get_postorder): Likewise.
31633 * df.h (struct df_d): Change field to be a vec.
31634 * lcm.c (compute_laterin): Adjust.
31635 (compute_available): Likewise.
31636 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31637 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31638 * tree-ssa-pre.c (compute_antic): Likewise.
31640 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31642 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31643 (depth_first_search::depth_first_search): Change structure init
31644 function to this constructor.
31645 (depth_first_search::add_bb): Rename function to this member.
31646 (depth_first_search::execute): Likewise.
31647 (flow_dfs_compute_reverse_finish): Adjust.
31649 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31651 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31652 (longest_simple_path): Likewise.
31653 * shrink-wrap.c (spread_components): Likewise.
31654 (disqualify_problematic_components): Likewise.
31655 (emit_common_heads_for_components): Likewise.
31656 (emit_common_tails_for_components): Likewise.
31657 (insert_prologue_epilogue_for_components): Likewise.
31659 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31661 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31664 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31666 * df-core.c (df_set_blocks): Start using auto_bitmap.
31667 (df_compact_blocks): Likewise.
31668 * df-problems.c (df_rd_confluence_n): Likewise.
31669 * df-scan.c (df_insn_rescan_all): Likewise.
31670 (df_process_deferred_rescans): Likewise.
31671 (df_update_entry_block_defs): Likewise.
31672 (df_update_exit_block_uses): Likewise.
31673 (df_entry_block_bitmap_verify): Likewise.
31674 (df_exit_block_bitmap_verify): Likewise.
31675 (df_scan_verify): Likewise.
31676 * lra-constraints.c (lra_constraints): Likewise.
31677 (undo_optional_reloads): Likewise.
31678 (lra_undo_inheritance): Likewise.
31679 * lra-remat.c (calculate_gen_cands): Likewise.
31680 (do_remat): Likewise.
31681 * lra-spills.c (assign_spill_hard_regs): Likewise.
31682 (spill_pseudos): Likewise.
31683 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31684 (bitmap_set_subtract_values): Likewise.
31686 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31688 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31689 management with auto_bitmap.
31690 (fix_inter_tick): Likewise.
31691 (fix_recovery_deps): Likewise.
31692 * ira.c (add_store_equivs): Likewise.
31693 (find_moveable_pseudos): Likewise.
31694 (split_live_ranges_for_shrink_wrap): Likewise.
31695 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31696 (rtx_reuse_manager::seen_def_p): Likewise.
31697 (rtx_reuse_manager::set_seen_def): Likewise.
31698 * print-rtl.h (class rtx_reuse_manager): Likewise.
31700 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31702 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31704 (migrate_btr_def): Likewise.
31705 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31706 * df-core.c (loop_post_order_compute): Likewise.
31707 (loop_inverted_post_order_compute): Likewise.
31708 * hsa-common.h: Likewise.
31709 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31710 * init-regs.c (initialize_uninitialized_regs): Likewise.
31711 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31712 (inline_small_functions): Likewise.
31713 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31714 * ira.c (combine_and_move_insns): Likewise.
31715 (build_insn_chain): Likewise.
31716 * loop-invariant.c (find_invariants): Likewise.
31717 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31718 * predict.c (tree_predict_by_opcode): Likewise.
31719 (predict_paths_leading_to): Likewise.
31720 (predict_paths_leading_to_edge): Likewise.
31721 (estimate_loops_at_level): Likewise.
31722 (estimate_loops): Likewise.
31723 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31724 (spread_components): Likewise.
31725 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31726 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31727 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31728 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31729 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31730 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31731 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31732 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31733 (mark_threaded_blocks): Likewise.
31734 (thread_through_all_blocks): Likewise.
31735 * tree-ssa.c (verify_ssa): Likewise.
31736 (execute_update_addresses_taken): Likewise.
31737 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31739 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31741 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31743 (post_order_compute): Likewise.
31744 (inverted_post_order_compute): Likewise.
31745 (pre_and_rev_post_order_compute_fn): Likewise.
31747 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31749 * genrecog.c (int_set::int_set): Explicitly construct our
31750 auto_vec base class.
31751 * vec.h (auto_vec::auto_vec): New constructor.
31753 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31755 * bitmap.h (class auto_bitmap): New constructor taking
31756 bitmap_obstack * argument.
31758 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31760 * bitmap.h (class auto_bitmap): Change type of m_bits to
31761 bitmap_head, and adjust ctor / dtor and member operators.
31763 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31765 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31766 when returned register mode doesn't match original mode.
31768 2017-05-12 Jeff Law <law@redhat.com>
31769 Jakub Jelinek <jakub@redhat.com>
31771 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31772 we look for cc setter after the compare-elim changes.
31773 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31774 within the vector to match what compare-elim now expects.
31775 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31776 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31778 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31779 after the compare-elim changes.
31780 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31781 the vector to match what compare-elim now expects.
31782 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31783 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31784 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31785 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31786 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31788 * config/visium/visium.c (single_set_and_flags): Fix where
31789 we look for cc setter after the compare-elim changes.
31790 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31791 with the vector to match what compare-elim now expects.
31792 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31793 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31794 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31795 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31796 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31797 (neg<mode>2_insn_set_overflow): Likewise.
31799 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31801 PR middle-end/79794
31802 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31803 maybe_expand_insn call, set ops[0].target. If still set after call,
31804 set alt_rtl. Add extra arg to recursive calls.
31805 (extract_bit_field): Add alt_rtl argument. Pass to
31807 * expmed.h (extract_bit_field): Fix prototype.
31808 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31809 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31810 to extract_bit_field_calls.
31811 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31812 Pass alt_rtl to extract_bit_field calls.
31813 * calls.c (store_unaligned_arguments_into_psuedos)
31814 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31815 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31817 * optabs.h (struct expand_operand): Add target bitfield.
31819 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31821 * compare-elim.c (try_eliminate_compare): Canonicalize
31822 operation with embedded compare to
31823 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31824 (set (reg) (operation)].
31826 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31828 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31831 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31832 cost of adding a carry flag for ADC instruction.
31833 [case MINUS]: Ignore the cost of subtracting a carry flag
31834 for SBB instruction.
31836 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31838 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31840 * config/rs6000/bmiintrin.h: New file.
31841 * config/rs6000/bmi2intrin.h: New file.
31842 * config/rs6000/x86intrin.h: New file.
31844 2017-05-12 Jeff Law <law@redhat.com>
31846 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31849 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31851 PR middle-end/80707
31852 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31854 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31857 early expansion of vector divide builtins.
31858 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31859 builtins identified as having unsigned arguments.
31861 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31863 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31864 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31865 expansion of vector logical operations (and, andc, or, xor,
31868 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31870 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31871 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31873 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31875 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31876 early GIMPLE expansion of vector multiplies.
31878 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31880 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31881 TARGET_HAVE_MOVT conditional.
31882 (movt splitter): Likewise.
31884 2017-05-12 Richard Biener <rguenther@suse.de>
31886 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31887 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31888 Fold all stmts not inplace.
31890 2017-05-12 Richard Biener <rguenther@suse.de>
31892 PR tree-optimization/80713
31893 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31894 inserted_exprs bit for not removed stmts.
31896 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31898 PR middle-end/69921
31899 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31900 parallelized" attribute for parallelized OpenACC kernels.
31901 * omp-offload.c (execute_oacc_device_lower): Use it.
31903 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31904 Set "oacc kernels" attribute.
31905 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31906 parameter. Adjust all users.
31907 (oacc_fn_attrib_kernels_p): Remove function.
31908 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31909 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31910 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31911 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31912 assert "oacc kernels" attribute is set.
31914 2017-05-11 Carl Love <cel@us.ibm.com>
31916 * config/rs6000/rs6000-c: Add support for built-in functions
31917 vector unsigned char vec_popcnt (vector signed char)
31918 vector unsigned char vec_popcnt (vector unsigned char)
31919 vector unsigned short vec_popcnt (vector signed short)
31920 vector unsigned short vec_popcnt (vector unsigned short)
31921 vector unsigned int vec_popcnt (vector signed int)
31922 vector unsigned int vec_popcnt (vector unsigned int)
31923 vector unsigned long long vec_popcnt (vector signed long long)
31924 vector unsigned long long vec_popcnt (vector unsigned long long)
31925 vector signed long long vec_slo (vector signed long long,
31926 vector signed char)
31927 vector signed long long vec_slo (vector signed long long,
31928 vector unsigned char)
31929 vector unsigned long long vec_slo (vector unsigned long long,
31930 vector signed char)
31931 vector unsigned long long vec_slo (vector unsigned long long,
31932 vector unsigned char)
31933 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31934 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31935 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31936 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31937 * doc/extend.texi: Update the built-in documentation file for the
31938 new built-in functions.
31940 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31942 * attribs.h (sorted_attr_string): Move machine independent
31943 functions for target clone support from the i386 port to common
31944 code. Rename ix86_function_versions to common_function_versions.
31945 Rename make_name to make_unique_name.
31946 (common_function_versions): Likewise.
31947 (make_unique_name): Likewise.
31948 (make_dispatcher_decl): Likewise.
31949 (is_function_default_version): Likewise.
31950 * attribs.c (attr_strcmp): Likewise.
31951 (sorted_attr_string): Likewise.
31952 (common_function_versions): Likewise.
31953 (make_unique_name): Likewise.
31954 (make_dispatcher_decl): Likewise.
31955 (is_function_default_version): Likewise.
31956 * config/i386/i386.c (attr_strcmp): Likewise.
31957 (sorted_attr_string): Likewise.
31958 (ix86_function_versions): Likewise.
31959 (make_name): Likewise.
31960 (make_dispatcher_decl): Likewise.
31961 (is_function_default_version): Likewise.
31962 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31964 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31967 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31968 Account for direct move costs for vec_construct of integer
31971 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31974 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31975 (UNSPEC_STX_ATOMIC): Ditto.
31976 (loaddi_via_sse): New insn.
31977 (storedi_via_sse): Ditto.
31978 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31979 Update corresponding peephole2 patterns.
31980 (atomic_storedi_fpu): Ditto.
31982 2017-05-11 Julia Koval <julia.koval@intel.com>
31984 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31985 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31987 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31988 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31989 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31991 2017-05-11 Nathan Sidwell <nathan@acm.org>
31993 * graphite-poly.c: Include dumpfile.h.
31995 * dumpfle.h (dump_function): Declare here ...
31996 * tree-dump.h (dump_function): ... not here.
31997 * dumpfile.c: #include tree-cfg.h.
31998 (dump_function): Move here from ...
31999 * tree-dump.c (dump_function): ... here.
32000 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32001 * graphite-poly.c: Don't include tree-dump.h.
32002 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32003 * print-tree.c: Likewise.
32004 * stor-layout.c: Likewise.
32005 * tree-nested.c: Likewise.
32007 * dumpfile.c (dump_start): Use TDF_FLAGS.
32008 (dump_enable_all): Fix TDF_KIND check thinko.
32010 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32012 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32013 array entries to represent two legal parameterizations of the
32014 overloaded __builtin_cmpb function, as represented by the
32015 P6_OV_BUILTIN_CMPB constant.
32016 (altivec_resolve_overloaded_builtin): Add special case handling
32017 for the __builtin_cmpb function, as represented by the
32018 P6_OV_BUILTIN_CMPB constant.
32019 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32020 (BU_P6_64BIT_2): New macro.
32021 (BU_P6_OVERLOAD_2): New macro
32022 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32023 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32024 (CMPB): Add overload support to represent both 32-bit and 64-bit
32025 compare-bytes function.
32026 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32027 support for TARGET_CMPB.
32028 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32029 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32030 documentation of the __builtin_cmpb overloaded built-in function.
32032 2017-05-11 Richard Biener <rguenther@suse.de>
32034 PR tree-optimization/80705
32035 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32036 bases are not vectorizable.
32038 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32040 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32041 when counting register pressure.
32043 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32045 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32046 (iv_ca_more_deps): Renamed to ...
32047 (iv_ca_compare_deps): ... this.
32048 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32052 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32054 (determine_group_iv_costs): ... here.
32055 (find_inv_vars_cb): Record inv var if it's not recorded before.
32057 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32059 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32060 (get_shiftadd_cost): Ditto.
32062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32064 * tree-ssa-address.c: Include header file.
32065 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32067 (add_to_parts): Refactor.
32068 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32069 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32072 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32074 PR tree-optimization/53090
32075 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32077 (extract_cond_operands): Detect condition with IV on both sides
32078 and return COMP_IV_EXPR_2.
32079 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32080 (rewrite_use_compare): Simplify by removing call to function
32081 extract_cond_operands.
32083 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32085 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32086 (extract_cond_operands): Detect condition comparing against non-
32087 invariant bound and return appropriate enum value.
32088 (find_interesting_uses_cond): Update use of extract_cond_operands.
32089 Handle its return value accordingly.
32090 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32094 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32095 nonlinear iv_use computation in loop invariant sensitive way.
32097 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32099 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32100 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32102 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32104 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32105 (dump_cand): Support iv_cand.inv_exprs.
32106 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32108 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32113 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32115 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32116 as local function. Include necessary header files.
32117 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32119 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32121 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32123 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32125 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32126 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32127 RSHIFT_EXPR and BIT_NOT_EXPR.
32129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32131 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32132 (adjust_setup_cost): New parameter supporting round up adjustment.
32133 (struct address_cost_data): Delete.
32134 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32135 (split_address_cost, ptr_difference_cost): Delete.
32136 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32137 (struct ainc_cost_data): New struct.
32138 (get_address_cost_ainc): New function.
32139 (get_address_cost, get_computation_cost): Reimplement.
32140 (determine_group_iv_cost_address): Record inv_expr for all uses of
32142 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32143 (iv_ca_has_deps): Reimplemented to ...
32144 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32146 (iv_ca_extend): Call iv_ca_more_deps.
32148 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32150 * tree-ssa-address.c (struct mem_address): Move to header file.
32151 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32152 * tree-ssa-address.h (struct mem_address): Move from C file.
32153 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32157 * tree-affine.h (aff_combination_type): New interface.
32158 (aff_combination_zero_p): Remove static.
32159 (aff_combination_const_p): New interface.
32160 (aff_combination_singleton_var_p): New interfaces.
32162 2017-05-11 Richard Biener <rguenther@suse.de>
32164 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32165 Skip unreachable blocks and destinations.
32166 (eliminate): Move stmt removal and fixup ...
32167 (fini_eliminate): ... here. Skip inserted exprs.
32168 (pass_pre::execute): Move fini_pre after fini_eliminate.
32169 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32170 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32171 PRE to get rid of dead code that has invalid SSA form and
32172 split critical edges again.
32174 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32176 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32178 2017-05-11 Richard Biener <rguenther@suse.de>
32180 * passes.c (execute_function_todo): Verify loops if they are
32181 said to be up-to-date.
32182 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32183 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32185 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32188 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32189 handle calling assemble_external ourself.
32192 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32193 modes with zero size. Enhance comment.
32195 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32197 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32198 built-ins for vec_xl and vec_xst with short and char pointer
32201 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32203 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32204 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32205 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32206 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32207 (_mm_maskz_min_round_ss): New intrinsics.
32208 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32209 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32210 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32211 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32212 (__builtin_ia32_minss_mask_round): New builtins.
32213 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32214 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32215 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32217 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32218 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32220 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32223 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32225 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32226 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32227 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32228 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32229 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32230 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32231 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32232 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32234 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32235 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32236 (__builtin_ia32_mulss_mask_round): New builtins.
32237 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32238 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32239 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32241 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32242 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32244 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32247 2017-05-10 Julia Koval <julia.koval@intel.com>
32249 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32250 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32251 (_mm256_setr_m128i): New intrinsics.
32253 2017-05-10 Julia Koval <julia.koval@intel.com>
32255 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32256 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32257 (_mm_maskz_rcp14_ss): New intrinsics.
32258 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32259 (__builtin_ia32_rcp14ss_mask): New builtins.
32260 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32262 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32264 PR tree-optimization/51513
32265 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32266 (assert_unreachable_fallthru_edge_p): Use it.
32267 (group_case_labels_stmt): Likewise.
32268 * tree-cfg.h: Prototype it.
32269 * stmt.c: Include cfghooks.h and tree-cfg.h.
32270 (emit_case_dispatch_table) <gap_label>: New local variable.
32271 Use it to fill dispatch table gaps.
32272 Test for default_label before updating probabilities.
32273 (expand_case) <default_label>: Remove unneeded initialization.
32274 Test for unreachable default case statement and remove its edge.
32275 Set default_label accordingly.
32276 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32278 2017-05-10 Carl Love <cel@us.ibm.com>
32280 * config/rs6000/rs6000-c: Add support for built-in functions
32281 vector signed char vec_neg (vector signed char)
32282 vector signed short int vec_neg (vector short int)
32283 vector signed int vec_neg (vector signed int)
32284 vector signed long long vec_neg (vector signed long long)
32285 vector float vec_neg (vector float)
32286 vector double vec_neg (vector double)
32287 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32289 * config/rs6000/altivec.h: Add define for vec_neg
32290 * doc/extend.texi: Update the built-in documentation for the
32291 new built-in functions.
32293 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32295 PR tree-optimization/77644
32296 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32298 2017-05-10 Nathan Sidwell <nathan@acm.org>
32300 * dumpfile.h (TDI_lang_all): New.
32301 (TDF_KIND): New. Renumber others
32302 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32304 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32306 (get_dump_file_name): Adjust suffix generation.
32307 (dump_enable_all): Use TDF_KIND.
32308 * doc/invoke.texi (-fdump-lang-all): Document.
32310 * dumpfile.h: Tabify.
32312 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32315 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32316 Move member access before delete.
32318 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32320 * tree-inline.c (expand_call_inline): Split block at stmt
32323 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32326 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32327 are now unused after splitting mov{sf,sd}_hardfloat.
32328 (f32_lr2): Likewise.
32329 (f32_lm): Likewise.
32330 (f32_lm2): Likewise.
32331 (f32_li): Likewise.
32332 (f32_li2): Likewise.
32333 (f32_lv): Likewise.
32334 (f32_sr): Likewise.
32335 (f32_sr2): Likewise.
32336 (f32_sm): Likewise.
32337 (f32_sm2): Likewise.
32338 (f32_si): Likewise.
32339 (f32_si2): Likewise.
32340 (f32_sv): Likewise.
32341 (f32_dm): Likewise.
32342 (f32_vsx): Likewise.
32343 (f32_av): Likewise.
32344 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32345 For movsf, order stores so the VSX stores occur before the GPR
32346 store which encourages the register allocator to use a traditional
32347 FPR instead of a GPR. For movsd, order the stores so that the GPR
32348 store comes before the VSX stores to allow the power6 to work.
32349 This is due to the power6 not having a 32-bit integer store
32350 instruction from a FPR.
32351 (movsf_hardfloat): Likewise.
32352 (movsd_hardfloat): Likewise.
32354 2017-05-09 Martin Sebor <msebor@redhat.com>
32356 PR translation/80280
32357 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32360 PR translation/80280
32361 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32362 data member added in r247778.
32363 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32365 2017-05-09 Nathan Sidwell <nathan@acm.org>
32367 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32369 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32372 2017-05-09 Marek Polacek <polacek@redhat.com>
32374 * doc/invoke.texi: Fix typo.
32376 2017-05-09 Richard Biener <rguenther@suse.de>
32378 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32379 (vrp_val_is_min): Likewise.
32380 (set_value_range_to_value): Likewise.
32381 (set_value_range_to_nonnegative): Likewise.
32382 (gimple_assign_nonzero_p): Likewise.
32383 (gimple_stmt_nonzero_p): Likewise.
32384 (vrp_int_const_binop): Likewise. Remove unreachable case.
32385 (adjust_range_with_scev): Adjust comments.
32386 (compare_range_with_value): Likewise.
32387 (extract_range_from_phi_node): Likewise.
32388 (test_for_singularity): Likewise.
32390 2017-05-09 Richard Biener <rguenther@suse.de>
32392 * tree-vrp.c (get_single_symbol): Add assert that we don't
32393 get overflowed constants as invariant part.
32394 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32395 checks. Use wi::cmp instead of recursing for integer constants.
32396 (compare_values): Just ignore whether we assumed undefined
32397 overflow instead of failing the compare.
32398 (extract_range_for_var_from_comparison_expr): Add comment before the
32399 TREE_NO_WARNING sets.
32400 (test_for_singularity): Likewise.
32401 (extract_range_from_comparison): Do not disable optimization
32402 when we assumed undefined overflow.
32403 (extract_range_basic): Remove init of unused var.
32405 2017-05-09 Richard Biener <rguenther@suse.de>
32407 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32408 (extract_range_from_multiplicative_op_1): Adjust.
32409 (extract_range_from_binary_expr_1): Use int_const_binop.
32411 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32414 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32415 rs6000_store_data_bypass_p in seven define_bypass directives and
32416 in several comments.
32417 * config/rs6000/rs6000-protos.h: Add prototype for
32418 rs6000_store_data_bypass_p function.
32419 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32420 function implements slightly different (rs6000-specific) semantics
32421 than store_data_bypass_p, returning false rather than aborting
32422 with assertion error when arguments do not satisfy the
32423 requirements of store data bypass.
32424 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32425 rs6000_store_data_bypass_p.
32427 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32429 * config/xtensa/xtensa-protos.h
32430 (xtensa_initial_elimination_offset): New declaration.
32431 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32432 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32433 macro definition, add case for FRAME_POINTER_REGNUM when
32434 FRAME_GROWS_DOWNWARD.
32435 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32436 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32437 xtensa_initial_elimination_offset.
32439 2017-05-08 Nathan Sidwell <nathan@acm.org>
32441 * doc/invoke.texi: Alphabetize -fdump options.
32443 2017-05-08 Martin Sebor <msebor@redhat.com>
32445 PR translation/80280
32446 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32448 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32450 * target.def (compute_frame_layout): New optional target hook.
32451 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32452 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32453 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32455 * reload1.c (verify_initial_elim_offsets): Likewise.
32456 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32457 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32458 (arm_get_frame_offsets): Split up into this ...
32459 (arm_compute_frame_layout): ... and this function.
32461 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32463 * config/aarch64/constraints.md (Usa): New constraint.
32464 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32466 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32468 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32469 with_multilib_list after it has been checked.
32471 2017-05-08 Richard Biener <rguenther@suse.de>
32473 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32474 (bitmap_set_subtract_values): Likewise.
32476 2017-05-08 Richard Biener <rguenther@suse.de>
32478 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32479 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32481 (gimple_stmt_nonzero_warnv_p): Rename to ...
32482 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32484 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32485 (extract_range_basic): Adjust, do not disable propagation on
32486 strict overflow sensitive simplification.
32487 (vrp_visit_cond_stmt): Likewise.
32489 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32491 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32492 body size unconditionally.
32494 2017-05-07 Jeff Law <law@redhat.com>
32497 2017-05-06 Jeff Law <law@redhat.com>
32498 PR tree-optimization/78496
32499 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32502 PR tree-optimization/78496
32503 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32504 (simplify_stmt_using_ranges): Call it.
32505 (vrp_dom_walker::before_dom_children): Extract equivalences
32506 from an ASSERT_EXPR with an equality comparison against a
32509 2017-05-06 Jeff Law <law@redhat.com>
32511 PR tree-optimization/78496
32512 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32515 PR tree-optimization/78496
32516 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32517 (simplify_stmt_using_ranges): Call it.
32518 (vrp_dom_walker::before_dom_children): Extract equivalences
32519 from an ASSERT_EXPR with an equality comparison against a
32522 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32524 * lra-constraints.c (lra_copy_reg_equiv): New function.
32525 (split_reg): Use it to copy equivalence information from the
32526 original register to the spill register.
32528 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32530 PR rtl-optimization/75964
32531 * simplify-rtx.c (simplify_const_relational_operation): Remove
32532 invalid handling of comparisons of integer ABS.
32534 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32536 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32537 initialize to zero.
32538 (init_regs): Remove declaration.
32539 (function_arg_advance_32): Initialize error_p as boolean variable.
32541 2017-05-05 Nathan Sidwell <nathan@acm.org>
32543 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32544 lines. Use for (;;).
32546 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32548 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32549 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32550 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32551 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32552 VF=2 that require versioning.
32554 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32556 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32559 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32561 * diagnostic.h (diagnostic_override_option_index): Convert from
32562 macro to inline function.
32564 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32566 * diagnostic.c (last_module_changed_p): New function.
32567 (set_last_module): New function.
32568 (diagnostic_report_current_module): Convert macro usage to
32569 the above functions.
32570 * diagnostic.h (diagnostic_context::last_module): Strengthen
32571 from const line_map * to const line_map_ordinary *.
32572 (diagnostic_last_module_changed): Delete macro.
32573 (diagnostic_set_last_module): Delete macro.
32575 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32577 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32578 with diagnostic_report_diagnostic.
32579 (diagnostic_n_impl_richloc): Likewise.
32580 * diagnostic.h (report_diagnostic): Delete macro.
32581 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32582 with diagnostic_report_diagnostic.
32583 * substring-locations.c (format_warning_va): Likewise.
32585 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32587 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32588 save/restor of format_spec. Move option-printing code to...
32589 (print_option_information): ...this new function, and
32590 reimplement by simply printing to the pretty_printer,
32591 rather than appending to the format string.
32593 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32595 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32596 handling logic into...
32597 (update_effective_level_from_pragmas): ...this new function.
32599 2017-05-04 Andrew Waterman <andrew@sifive.com>
32601 * config/riscv/riscv.opt (mstrict-align): New option.
32602 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32603 (SLOW_UNALIGNED_ACCESS): Define.
32604 (riscv_slow_unaligned_access): Declare.
32605 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32607 (riscv_slow_unaligned_access): New variable.
32608 (rocket_tune_info): Set slow_unaligned_access to true.
32609 (optimize_size_tune_info): Set slow_unaligned_access to false.
32610 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32611 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32612 (riscv_option_override): Set riscv_slow_unaligned_access.
32613 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32615 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32617 * config/riscv/riscv.md: Unify indentation.
32619 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32624 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32626 (extendsi<mode>2): Add support for doing sign extension via
32627 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32628 don't have ISA 3.0 instructions.
32629 (extendsi<mode>2 splitter): Likewise.
32630 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32631 generate the normal insns since SImode can now go in vector
32632 registers. Disallow the special UNSPECs needed for previous
32633 machines to hide SImode being used. Add new insns
32634 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32635 (fix_trunc<mode>si2_stfiwx): Likewise.
32636 (fix_trunc<mode>si2_internal): Likewise.
32637 (fixuns_trunc<mode>si2): Likewise.
32638 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32639 (fctiw<u>z_<mode>_smallint): Likewise.
32640 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32641 of floating point to 32-bit integer from doing a direct move to
32642 the GPR registers to do a store.
32643 (fctiwz_<mode>): Break long line.
32645 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32647 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32648 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32649 (addr_list, addr_offset_valid_p): New.
32650 (split_address_groups): Check offset validity with above function.
32651 (gt-tree-ssa-loop-ivopts.h): Include header file.
32653 2017-05-05 Nathan Sidwell <nathan@acm.org>
32655 * config.gcc (arm*-*-*): Add missing 'fi'.
32657 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32659 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32660 included in -fopt-info does not matter.
32661 * doc/optinfo.texi (-fopt-info): Fix description of default
32662 behavour. Explicitly say order of options included in -fopt-info
32665 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32667 * config.gcc: Allow combinations of aprofile and rmprofile values for
32668 --with-multilib-list.
32669 * config/arm/t-multilib: New file.
32670 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32671 variables. Remove setting of ISA and floating-point ABI in
32672 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32673 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32674 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32675 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32677 * config/arm/t-rmprofile: Likewise except for the matches changes.
32678 * doc/install.texi (--with-multilib-list): Document the combination of
32679 aprofile and rmprofile values and warn about pitfalls in doing that.
32681 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32683 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32684 (movdi_aarch64): Likewise.
32686 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32688 PR tree-optimization/80632
32689 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32691 (build_arrays): Initialize it for virtual phis.
32692 (fix_phi_nodes): Use it for virtual phis.
32694 PR tree-optimization/80558
32695 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32696 [x, y] op z into [x op, y op z] for op & or | if conditions
32699 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32700 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32703 * config/arm/arm.md (use_literal_pool): Remove.
32704 (64-bit immediate split): No longer takes cost into consideration
32705 if arm_disable_literal_pool is enabled.
32706 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32707 used when arm_disable_literal_pool is enabled.
32708 (arm_max_const_double_inline_cost): Remove use of
32709 arm_disable_literal_pool.
32710 (push_minipool_fix): Add assert.
32711 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32712 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32713 (no_literal_pool_sf_immediate): New.
32715 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32717 PR tree-optimization/80613
32718 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32719 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32721 2017-05-05 Richard Biener <rguenther@suse.de>
32723 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32725 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32727 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32728 of this flag from insn conditions due to removal from r247495.
32730 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32732 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32734 (arm_early_store_addr_dep_ptr): Likewise.
32735 * config/arm/aarch-common-protos.h
32736 (arm_early_load_addr_dep_ptr): Add prototype.
32737 (arm_early_store_addr_dep_ptr): Likewise.
32738 * config/arm/cortex-a53.md: Add new bypasses.
32740 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32742 * tree.c (next_type_uid): Change type to unsigned.
32743 (type_hash_canon): Decrement back next_type_uid if
32744 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32745 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32748 2017-05-04 Martin Sebor <msebor@redhat.com>
32750 * builtins.c: Fix a trivial typo in a comment.
32752 PR middle-end/79234
32753 * builtins.c (check_sizes): Adjust to handle reading past the end.
32754 Avoid printing excessive upper bound of ranges. Use %E to print
32755 tree nodes instead of converting them to %wu.
32756 (expand_builtin_memchr): New function.
32757 (compute_dest_size): Rename...
32758 (compute_objsize): ...to this.
32759 (expand_builtin_memcpy): Adjust.
32760 (expand_builtin_mempcpy): Adjust.
32761 (expand_builtin_strcat): Adjust.
32762 (expand_builtin_strcpy): Adjust.
32763 (check_strncat_sizes): Adjust.
32764 (expand_builtin_strncat): Adjust.
32765 (expand_builtin_strncpy): Adjust and simplify.
32766 (expand_builtin_memset): Adjust.
32767 (expand_builtin_bzero): Adjust.
32768 (expand_builtin_memcmp): Adjust.
32769 (expand_builtin): Handle memcmp.
32770 (maybe_emit_chk_warning): Check strncat just once.
32772 2017-05-04 Martin Sebor <msebor@redhat.com>
32774 PR preprocessor/79214
32775 PR middle-end/79222
32776 PR middle-end/79223
32777 * builtins.c (check_sizes): Add inlining context and issue
32778 warnings even when -Wno-system-headers is set.
32779 (check_strncat_sizes): Same.
32780 (expand_builtin_strncat): Same.
32781 (expand_builtin_memmove): New function.
32782 (expand_builtin_stpncpy): Same.
32783 (expand_builtin): Handle memmove and stpncpy.
32785 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32787 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32788 which is not used any more.
32790 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32792 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32794 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32796 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32797 (cortexa53_tunings): Likewise.
32798 (cortexa57_tunings): Likewise.
32799 (cortexa72_tunings): Likewise.
32800 (cortexa73_tunings): Likewise.
32802 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32804 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32805 Set loop alignment to 8.
32807 2017-05-04 Martin Sebor <msebor@redhat.com>
32809 PR translation/80280
32810 * builtins.c (expand_builtin_object_size): Add missing quoting to
32811 %D and like directives.
32812 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32813 (hsa_type_for_tree_type): Same.
32814 (verify_function_arguments): Same.
32815 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32816 * varasm.c (get_section): Same.
32819 2017-05-04 Martin Sebor <msebor@redhat.com>
32821 PR translation/80280
32822 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32824 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32826 * config/aarch64/aarch64.c (generic_addrcost_table):
32827 Change HI/TI mode setting.
32829 2017-05-04 Martin Jambor <mjambor@suse.cz>
32831 PR tree-optimization/80622
32832 * tree-sra.c (comes_initialized_p): New function.
32833 (build_accesses_from_assign): Only set write lazily when
32834 comes_initialized_p is false.
32835 (analyze_access_subtree): Use comes_initialized_p.
32836 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32837 instead of testing for PARM_DECL.
32839 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32841 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32842 constraint on operand 0 to allow more general addressing modes.
32843 Adjust output template.
32844 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32846 * config/aarch64/aarch64-protos.h
32847 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32848 * config/aarch64/constraints.md (Dp): New address constraint.
32849 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32852 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32854 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32855 update use of estimate_ipcp_clone_size_and_time.
32856 (estimate_local_effects): Update use of
32857 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32858 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32859 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32860 Return nonspecialized time.
32862 2017-05-04 Richard Biener <rguenther@suse.de>
32864 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32865 for the last VUSE which def dominates the PHI. Directly call
32867 (get_continuation_for_phi_1): Remove.
32869 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32871 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32872 to explain the use of truncating division. Cap the number of
32873 iterations to the maximum given by nb_iterations_upper_bound,
32876 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32878 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32879 * configure: Regenerate.
32880 * config.in: Regenerate.
32881 * config/i386/driver-mingw32.c: new file.
32882 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32883 * config.host: Link driver-mingw32.o on MinGW host.
32884 * doc/install.texi: Document new --enable-mingw-wildcard configure
32887 2017-05-04 Marek Polacek <polacek@redhat.com>
32889 PR tree-optimization/80612
32890 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32892 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32893 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32895 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32896 (movt splitter): Likewise.
32897 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32898 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32899 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32900 block for Thumb-1 with MOVT.
32901 (thumb2_legitimate_address_p): Move code block ...
32902 (can_avoid_literal_pool_for_label_p): ... into this new function.
32903 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32905 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32906 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32907 "M-profile targets with the MOVT instruction".
32909 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32911 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32912 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32913 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32915 2017-05-04 Martin Liska <mliska@suse.cz>
32917 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32918 variable cond_code.
32920 2017-05-04 Richard Biener <rguenther@suse.de>
32922 * tree.c (array_at_struct_end_p): Handle arrays at struct
32923 end with flexarrays more conservatively. Refactor and treat
32924 arrays of arrays or aggregates more strict. Fix
32925 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32926 * tree.c (array_at_struct_end_p): Adjust prototype.
32927 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32928 * gimple-fold.c (get_range_strlen): Likewise.
32929 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32931 2017-05-04 Richard Biener <rguenther@suse.de>
32933 PR tree-optimization/31130
32934 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32936 (supports_overflow_infinity): Likewise.
32937 (is_negative_overflow_infinity): Likewise.
32938 (is_positive_overflow_infinity): Likewise.
32939 (is_overflow_infinity): Likewise.
32940 (stmt_overflow_infinity): Likewise.
32941 (overflow_infinity_range_p): Likewise.
32942 (usable_range_p): Remove as always returning true.
32943 (make_overflow_infinity): Remove.
32944 (negative_overflow_infinity): Likewise.
32945 (positive_overflow_infinity): Likewise.
32946 (avoid_overflow_infinity): Likewise.
32947 (set_value_range): Adjust accordingly.
32948 (set_value_range_to_nonnegative): Likewise, remove now unused
32949 overflow_infinity arg.
32950 (vrp_operand_equal_p): Adjust.
32951 (update_value_range): Likewise.
32952 (range_int_cst_singleton_p): Likewise.
32953 (operand_less_p): Likewise.
32954 (compare_values_warnv): Likewise.
32955 (extract_range_for_var_from_comparison_expr): Likewise.
32956 (vrp_int_const_binop): Likewise.
32957 (zero_nonzero_bits_from_vr): Likewise.
32958 (extract_range_from_multiplicative_op_1): Likewise.
32959 (extract_range_from_binary_expr_1): Likewise.
32960 (extract_range_from_unary_expr): Likewise.
32961 (extract_range_from_comparison): Likewise.
32962 (extract_range_basic): Likewise.
32963 (adjust_range_with_scev): Likewise.
32964 (compare_ranges): Likewise.
32965 (compare_range_with_value): Likewise.
32966 (dump_value_range): Likewise.
32967 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32969 (simplify_cond_using_ranges): Adjust.
32971 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32973 * brig-builtins.def: Added a builtin for class_f64.
32974 * builtin-types.def: Added a builtin type needed by class_f64.
32976 2017-05-03 Jason Merrill <jason@redhat.com>
32978 * timevar.def: Add TV_CONSTEXPR.
32980 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32982 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32984 2017-05-03 Martin Jambor <mjambor@suse.cz>
32986 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32987 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32988 * ipa-cp.c (ipcp_propagate_stage): Do not call
32989 ipa_update_after_lto_read.
32990 * ipa-inline.c (ipa_inline): Likewise.
32992 2017-05-03 Martin Jambor <mjambor@suse.cz>
32994 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32995 tag. Added a default constructor and a destructor.
32996 (ipa_edge_args_sum_t): New class;
32997 (ipa_edge_args_sum): Declare.
32998 (ipa_edge_args_vector): Remove declaration.
32999 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33000 (ipa_free_edge_args_substructures): Remove declaration.
33001 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33002 (ipa_edge_args_info_available_for_edge_p): Likewise.
33003 * ipa-prop.c (ipa_edge_args_vector): Removed.
33004 (edge_removal_hook_holder): Likewise.
33005 (edge_duplication_hook_holder): Likewise.
33006 (ipa_edge_args_sum): New variable.
33007 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33008 ipa_edge_args_vector.
33009 (ipa_free_edge_args_substructures): Likewise.
33010 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33011 ipa_edge_args_vector.
33012 (ipa_edge_removal_hook): Turned into method
33013 ipa_edge_args_sum_t::remove.
33014 (ipa_edge_duplication_hook): Turned into method
33015 ipa_edge_args_sum_t::duplicate.
33016 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33017 registering edge hooks.
33018 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33019 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33020 ipa_edge_args_sum instead of ipa_edge_args_vector.
33021 * ipa-profile.c (ipa_profile): Likewise.
33023 2017-05-03 Martin Jambor <mjambor@suse.cz>
33025 * symbol-summary.h (function_summary): New method exists.
33026 (function_summary::symtab_removal): Deallocate through release.
33027 (call_summary): New class.
33028 (gt_ggc_mx): New overload.
33029 (gt_pch_nx): Likewise.
33030 (gt_pch_nx): Likewise.
33032 2017-05-03 Jeff Law <law@redhat.com>
33034 PR tree-optimization/78496
33035 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33036 from simplify_cond_using_ranges. Split off code to walk
33037 backwards through casts into ...
33038 (simplify_cond_using_ranges_2): New function.
33039 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33040 (execute_vrp): After identifying jump threads, call
33041 simplify_cond_using_ranges_2.
33043 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33046 * ipa-inline.h (inline_summary): Add ctor.
33047 (create_ggc): Do not use ggc_cleared_alloc.
33049 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33050 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33052 * gcc.c (handle_braces): Support escaping in switch matching
33054 * doc/invoke.texi (Spec Files): Document it.
33055 Remove superfluous @code markup in items.
33057 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33059 * diagnostic-show-locus.c (struct column_range): New struct.
33060 (get_affected_columns): New function.
33061 (get_printed_columns): New function.
33062 (struct correction): New struct.
33063 (correction::ensure_capacity): New function.
33064 (correction::ensure_terminated): New function.
33065 (struct line_corrections): New struct.
33066 (line_corrections::~line_corrections): New dtor.
33067 (line_corrections::add_hint): New function.
33068 (layout::print_trailing_fixits): Reimplement in terms of the new
33070 (selftest::test_overlapped_fixit_printing): New function.
33071 (selftest::diagnostic_show_locus_c_tests): Call it.
33073 2017-05-03 Nathan Sidwell <nathan@acm.org>
33075 Canonicalize canonical type hashing
33076 * tree.h (type_hash_canon_hash): Declare.
33077 * tree.c (type_hash_list, attribute_hash_list): Move into
33078 type_hash_canon_hash.
33079 (build_type_attribute_qual_variant): Break out hash code calc into
33080 type_hash_canon_hash.
33081 (type_hash_canon_hash): New. Generic type hash computation.
33082 (build_range_type_1, build_array_type_1, build_function_type,
33083 build_method_type_directly, build_offset_type, build_complex_type,
33084 make_vector_type): Call it.
33086 2017-05-03 Richard Biener <rguenther@suse.de>
33088 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33089 When all DRs have unknown misaligned do not always peel
33090 when there is a store but apply the same costing model as if
33091 there were only loads.
33093 2017-05-03 Richard Biener <rguenther@suse.de>
33096 PR tree-optimization/80492
33097 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33098 compare_base_decls returning dont-know properly.
33100 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33102 * config/arm/iterators.md (CCSI): New mode iterator.
33103 (arch): New mode attribute.
33104 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33105 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33106 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33107 code iterator for success result mode.
33108 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33109 the corresponding new insn generators.
33111 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33114 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33115 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33117 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33119 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33120 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33121 (DDR_A): Wrap DDR argument in brackets.
33122 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33123 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33124 (DDR_REVERSED_P): Likewise.
33126 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33128 PR tree-optimization/79472
33129 * tree-switch-conversion.c (struct switch_conv_info): Add
33130 contiguous_range and default_case_nonstandard fields.
33131 (collect_switch_conv_info): Compute contiguous_range and
33132 default_case_nonstandard fields, don't clear final_bb if
33133 contiguous_range and only the default case doesn't have the required
33135 (check_all_empty_except_final): Set default_case_nonstandard instead
33136 of failing if contiguous_range and the default case doesn't have empty
33138 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33139 and only the default case doesn't have the required constants. Skip
33141 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33142 if default_case_nonstandard.
33143 (build_constructors): Build constant 1 just once. Assert that default
33144 values aren't inserted in between cases if contiguous_range. Skip
33146 (build_arrays): Skip virtual phis.
33147 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33148 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33149 Handle virtual phis.
33150 (gen_inbound_check): Handle default_case_nonstandard case.
33151 (process_switch): Adjust check_final_bb caller. Call
33152 gather_default_values with the first non-default case instead of
33153 default case if default_case_nonstandard.
33155 2017-05-02 Nathan Sidwell <nathan@acm.org>
33157 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33158 check. Fix formatting.
33160 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33162 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33163 errors when comparing specialized and unspecialized times.
33165 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33167 * diagnostic-show-locus.c
33168 (layout::should_print_annotation_line_p): Make private.
33169 (layout::print_annotation_line): Make private.
33170 (layout::annotation_line_showed_range_p): Make private.
33171 (layout::show_ruler): Make private.
33172 (layout::print_source_line): Make private. Pass in line and
33173 line_width, rather than calling location_get_source_line. Drop
33175 (layout::print_leading_fixits): New method.
33176 (layout::print_any_fixits): Rename to...
33177 (layout::print_trailing_fixits): ...this, and make private.
33178 Don't print newline fixits.
33179 (diagnostic_show_locus): Move logic for printing one row into...
33180 (layout::print_line): ...this new function. Move the
33181 location_get_source_line call and error-handling from
33182 print_source_line to here. Call print_leading_fixits, and rename
33183 print_any_fixits to print_trailing_fixits.
33184 (selftest::test_fixit_insert_containing_newline): Update now that
33185 newlines are partially supported.
33186 (selftest::test_fixit_insert_containing_newline_2): New test.
33187 (selftest::test_fixit_replace_containing_newline): Update comments.
33188 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33189 * edit-context.c (class added_line): New class.
33190 (class edited_line): Describe newline handling in comment.
33191 (edited_line::actually_edited_p): New method.
33192 (edited_line::print_content): Delete redundant decl.
33193 (edited_line::m_predecessors): New field.
33194 (edited_file::print_content): Call edited_line::print_content.
33195 (edited_file::print_diff): Update to support newlines.
33196 (edited_file::print_diff_hunk): Likewise.
33197 (edited_file::print_run_of_changed_lines): New function.
33198 (edited_file::print_diff_line): Convert to...
33199 (print_diff_line): ...this.
33200 (edited_file::get_effective_line_count): New function.
33201 (edited_line::edited_line): Initialize new field m_predecessors.
33202 (edited_line::~edited_line): Clean up m_predecessors.
33203 (edited_line::apply_fixit): Handle newlines.
33204 (edited_line::get_effective_line_count): New function.
33205 (edited_line::print_content): New function.
33206 (edited_line::print_diff_lines): New function.
33207 (selftest::test_applying_fixits_insert_containing_newline): New
33209 (selftest::test_applying_fixits_replace_containing_newline): New
33211 (selftest::insert_line): New function.
33212 (selftest::test_applying_fixits_multiple_lines): Add example of
33214 (selftest::edit_context_c_tests): Call the new tests.
33216 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33218 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33219 parameter cand. Update dump information.
33220 (get_computation_cost): Update uses.
33222 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33224 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33225 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33226 (get_computation_at, rewrite_use_address): Update use of
33227 get_computation_aff.
33229 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33231 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33232 (get_computation): Delete.
33233 (get_computation_cost): Implement like get_computation_cost_at.
33234 Use get_computation_at.
33235 (get_computation_cost_at): Delete.
33236 (rewrite_use_nonlinear_expr): Use get_computation_at.
33237 (rewrite_use_compare, remove_unused_ivs): Ditto.
33239 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33241 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33243 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33245 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33246 (ivopts_global_cost_for_size): Rename parameter and update uses.
33247 (iv_ca_recount_cost): Update uses.
33248 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33249 candidates seperately in n_invs and n_cands.
33250 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33252 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33254 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33255 (find_inv_vars_cb): New.
33256 (find_depends): Renamed to ...
33257 (find_inv_vars): ... this.
33258 (add_candidate_1, force_var_cost): Call find_inv_vars.
33259 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33263 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33264 inv_vars. Add inv_exprs.
33265 (struct iv_cand): Rename depends_on to inv_vars.
33266 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33267 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33268 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33269 (dump_cand): Dump inv_vars.
33270 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33271 (record_invariant, find_depends, add_candidate_1): Ditto.
33272 (set_group_iv_cost, force_var_cost): Ditto.
33273 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33274 (get_computation_cost_at, get_computation_cost): Ditto.
33275 (determine_group_iv_cost_generic): Ditto.
33276 (determine_group_iv_cost_address): Ditto.
33277 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33278 (determine_group_iv_costs): Ditto.
33279 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33280 (iv_ca_set_remove_invariants): Renamed to ...
33281 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33282 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33283 (iv_ca_set_add_invariants): Renamed to ...
33284 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33285 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33286 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33287 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33288 (create_new_ivs): Remove useless dump.
33290 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33292 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33294 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33295 (iv_ca_set_no_cp, create_new_iv): Ditto.
33297 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33299 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33301 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33303 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33304 function tree_check2.
33306 2017-05-02 Martin Liska <mliska@suse.cz>
33308 * doc/gcov.texi: Add missing preposition.
33309 * gcov.c (function_info::function_info): Properly fill up
33310 all member variables.
33312 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33314 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33316 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33318 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33320 2017-05-02 Martin Liska <mliska@suse.cz>
33323 * lto-streamer-in.c (lto_read_tree_1): Remove
33324 LTO_STREAMER_DEBUG.
33325 * lto-streamer.c (struct tree_hash_entry): Likewise.
33326 (struct tree_entry_hasher): Likewise.
33327 (tree_entry_hasher::hash): Likewise.
33328 (tree_entry_hasher::equal): Likewise.
33329 (lto_streamer_init): Likewise.
33330 (lto_orig_address_map): Likewise.
33331 (lto_orig_address_get): Likewise.
33332 (lto_orig_address_remove): Likewise.
33333 * lto-streamer.h: Likewise.
33334 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33335 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33337 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33339 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33340 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33341 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33342 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33343 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33344 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33345 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33346 (mm_maskz_sub_ss): New intrinsics.
33347 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33348 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33349 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33350 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33351 (__builtin_ia32_subss_mask_round): New builtins.
33352 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33353 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33354 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33356 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33357 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33359 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33362 2017-05-02 Martin Jambor <mjambor@suse.cz>
33364 PR tree-optimization/78687
33365 * tree-sra.c (access): New field parent.
33366 (process_subtree_disqualification): New function.
33367 (disqualify_candidate): Call it.
33368 (build_accesses_from_assign): Reset write flag if creating an
33370 (build_access_subtree): Fill in parent field and also prpagate
33371 down grp_write flag.
33372 (create_artificial_child_access): New parameter set_grp_write, set
33373 grp_write to its value.
33374 (propagate_subaccesses_across_link): Also propagate grp_write flag
33376 (propagate_all_subaccesses): Push the closest parent back to work
33377 queue if add_access_to_work_queue returned true.
33379 2017-05-02 Richard Biener <rguenther@suse.de>
33381 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33382 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33383 -fstrict-overflow documentation.
33384 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33385 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33386 flag_strict_overflow.
33387 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33388 * lto-opts.c (lto_write_options): Do not stream it.
33389 * lto-wrapper.c (merge_and_complain): Do not handle it.
33390 * opts.c (default_options_table): Do not set -fstrict-overflow.
33391 (finish_options): Likewise do not clear it when sanitizing.
33392 * simplify-rtx.c (simplify_const_relational_operation): Do not
33393 test flag_strict_overflow.
33395 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33397 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33398 using enabled attribute.
33399 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33400 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33401 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33402 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33403 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33404 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33405 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33406 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33407 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33408 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33410 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33412 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33414 2017-05-02 Richard Biener <rguenther@suse.de>
33416 PR tree-optimization/80591
33418 2017-04-10 Richard Biener <rguenther@suse.de>
33420 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33423 2017-05-02 Richard Biener <rguenther@suse.de>
33425 PR tree-optimization/80549
33426 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33427 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33428 headers if they do not have a preheader.
33430 2017-05-02 Martin Liska <mliska@suse.cz>
33433 * common.opt: Fix typo.
33434 * doc/invoke.texi: Likewise.
33436 2017-05-01 Jan Beulich <jbeulich@suse.com>
33438 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33439 swapping, add (x,x,m,x,n) alternative.
33441 2017-05-01 Nathan Sidwell <nathan@acm.org>
33443 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33444 unnecessary unadjusted_alignment check.
33446 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33449 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33451 * gimplify.c (gimplify_cilk_detach): New function.
33452 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33453 * tree-core.h: Document EXPR_CILK_SPAWN.
33454 * tree.h (EXPR_CILK_SPAWN): Define.
33456 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33458 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33459 to use new fixit_hint representation, using the "replace" logic.
33460 (get_line_span_for_fixit_hint): Likewise.
33461 (layout::print_any_fixits): Likewise.
33462 (selftest::test_one_liner_many_fixits): Rename to...
33463 (selftest::test_one_liner_many_fixits_1): ...this, and update
33464 comment and expected output to reflect that the multiple fix-it
33465 hints are now consolidated into one insertion.
33466 (selftest::test_one_liner_many_fixits_2): New test.
33467 (selftest::test_diagnostic_show_locus_one_liner): Update for
33469 (selftest::test_fixit_consolidation): Update for fix-it API
33471 * diagnostic.c (print_parseable_fixits): Likewise.
33472 * edit-context.c (edited_line::m_line_events): Convert from
33473 auto_vec <line_event *> to auto_vec <line_event>.
33474 (class line_event): Convert from abstract base class to a concrete
33475 class, taking over the role of replace_event.
33476 (class insert_event): Delete.
33477 (class replace_event): Rename to class line_event. Convert to
33479 (edit_context::add_fixits): Reimplement.
33480 (edit_context::apply_insert): Delete.
33481 (edit_context::apply_replace): Rename to...
33482 (edit_context::apply_fixit): ...this. Convert to half-open range.
33483 (edited_file::apply_insert): Delete.
33484 (edited_file::apply_replace): Rename to...
33485 (edited_file::apply_fixit): ...this.
33486 (edited_line::~edited_line): Drop deletion of events.
33487 (edited_line::apply_insert): Delete.
33488 (edited_line::apply_replace): Rename to...
33489 (edited_line::apply_fixit): ...this. Convert to half-open range.
33490 Update for change to type of m_line_events.
33491 * edit-context.h (edit_context::apply_insert): Delete.
33492 (edit_context::apply_replace): Rename to...
33493 (edit_context::apply_fixit): ...this.
33495 2017-05-01 Martin Sebor <msebor@redhat.com>
33497 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33500 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33503 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33504 __get_cpuid_max returns 0.
33505 (__get_cpuid_count): Ditto.
33507 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33509 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33510 replacement expression is another instance of one of its arguments.
33512 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33515 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33516 check for stack push/pop autoinc.
33517 * config/i386/i386.c (ix86_agi_dependent): Return false
33518 if the only reason why modified_in_p returned true is that
33519 addr is SP based and set_insn is a push or pop.
33521 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33523 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33526 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33529 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33530 (account_size_time): Use two predicates - exec_pred and
33532 (evaluate_conditions_for_known_args): Compute both clause and
33534 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33535 (inline_summary_t::duplicate): Update.
33536 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33538 (compute_inline_parameters): Likewise.
33539 (estimate_edge_size_and_time): Update caluclation of time.
33540 (estimate_node_size_and_time): Compute both time and nonspecialized
33542 (estimate_ipcp_clone_size_and_time): Update.
33543 (inline_merge_summary): Update.
33544 (do_estimate_edge_time): Update.
33545 (do_estimate_edge_size): Update.
33546 (do_estimate_edge_hints): Update.
33547 (inline_read_section, inline_write_summary): Stream both new predicates.
33548 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33550 (compute_inlined_call_time): Cleanup.
33551 (big_speedup_p): Update.
33552 (edge_badness): Update.
33553 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33554 (size_time_entry): Replace predicate by exec_predicate and
33555 nonconst_predicate.
33556 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33557 (estimate_edge_time): Return also nonspec_time.
33558 (reset_edge_growth_cache): Update.
33560 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33562 PR rtl-optimization/80491
33563 * ifcvt.c (noce_process_if_block): When looking for x setter
33564 with missing else_bb, don't check only the insn right before
33565 cond_earliest, but look for the last insn that x is modified in
33566 within the same bb.
33568 PR rtl-optimization/80491
33569 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33571 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33573 PR tree-optimization/80487
33574 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33576 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33578 PR tree-optimization/79697
33579 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33580 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33581 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33583 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33584 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33586 2017-04-28 Martin Sebor <msebor@redhat.com>
33588 PR tree-optimization/80523
33589 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33590 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33592 (maybe_warn, format_directive, parse_directive): Use new functions.
33593 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33595 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33597 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33599 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33601 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33602 target_header_dir): Set correctly.
33603 * configure: Regenerated.
33604 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33605 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33606 instead of SYSTEM_HEADER_DIR.
33608 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33610 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33611 (estimate_local_effects): Likewise.
33612 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33613 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33614 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33615 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33616 do_estimate_edge_time, estimate_edge_time): Likewise.
33617 * ipa-inline-analysis.c (estimate_node_size_and_time,
33618 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33619 (estimate_time_after_inlining): Remove.
33621 2017-04-28 Martin Liska <mliska@suse.cz>
33623 * doc/gcov.texi: Enhance documentation of gcov.
33625 2017-04-28 Martin Liska <mliska@suse.cz>
33627 * doc/gcov.texi: Sort options in alphabetic order.
33628 * doc/gcov-dump.texi: Likewise.
33629 * doc/gcov-tool.texi: Likewise.
33630 * gcov.c (print_usage): Likewise.
33631 * gcov-dump.c (print_usage): Likewise.
33632 * gcov-tool.c (print_merge_usage_message): Likewise.
33633 (print_rewrite_usage_message): Likewise.
33634 (print_overlap_usage_message): Likewise.
33636 2017-04-28 Martin Liska <mliska@suse.cz>
33638 PR gcov-profile/53915
33639 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33641 2017-04-28 Martin Liska <mliska@suse.cz>
33643 PR gcov-profile/79891
33644 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33645 is marked by compiler as living on a line.
33646 (get_cycles_count): Remove usage of the union.
33647 (output_intermediate_file): Likewise.
33648 (find_source): Fix GNU coding style.
33649 (accumulate_line_counts): Remove old non-all block mode.
33650 (output_lines): Remove usage of the union.
33651 * profile.c (output_location): Include all BBs, even if
33652 belonging to a same line (and file) as a previous BB.
33654 2017-04-28 Martin Liska <mliska@suse.cz>
33656 * gcov.c (process_args): Handle new argument 'w'.
33657 (read_graph_file): Assign ID to BBs.
33658 (output_branch_count): Display BB # if verbose flag is set.
33659 (output_lines): Likewise for arcs.
33660 (print_usage): Add '--verbose' option help.
33661 * doc/gcov.texi: Document --verbose (-w) option.
33663 2017-04-28 Martin Liska <mliska@suse.cz>
33665 * gcov.c (struct block_location_info): New struct.
33666 (process_file): Fill up the new structure.
33667 (read_graph_file): Replace usage of encoding by the newly added
33669 (add_line_counts): Likewise.
33670 (accumulate_line_counts): Remove usage of the union.
33671 (function_info::function_info): New function.
33672 (function_info::~function_info): Likewise.
33673 (process_file): Call delete instead of release_function.
33674 (release_function): Release the function.
33675 (release_structures): Call delete instead of release_function.
33676 (solve_flow_graph): Replace usage of num_blocks.
33677 (find_exception_blocks): Likewise.
33678 (output_lines): Fix GNU coding style.
33680 2017-04-28 Martin Liska <mliska@suse.cz>
33683 * coverage.c (coverage_remove_note_file): New function.
33684 * coverage.h: Declare the function.
33685 * toplev.c (finalize): Clean if an error has been seen.
33687 2017-04-28 Martin Liska <mliska@suse.cz>
33689 PR gcov-profile/80031
33690 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33691 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33692 * gcov.c (read_graph_file): Read just number of blocks.
33693 * profile.c (branch_prob): Do not stream 0 flags per a basic
33696 2017-04-28 Martin Liska <mliska@suse.cz>
33698 * gcov-dump.c (tag_*): Add new argument to declarations.
33699 (dump_gcov_file): Likewise.
33700 (tag_blocks): Add and use new argument depth.
33701 (tag_arcs): Likewise.
33702 (tag_lines): Likewise.
33703 (tag_counters): Likewise.
33704 (tag_summary): Likewise.
33705 (dump_working_sets): Use depth to do a proper indentation.
33707 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33710 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33711 * symtab.c (symtab_node::debug_symtab): Move definition here.
33713 2017-04-28 Richard Biener <rguenther@suse.de>
33715 * lto-streamer.h (LTO_major_version): Bump to 7.
33717 2017-04-28 Richard Biener <rguenther@suse.de>
33719 * tree-vrp.c (assert_info): New struct.
33720 (add_assert_info): New helper.
33721 (register_edge_assert_for_2): Refactor to add asserts to a vector
33723 (register_edge_assert_for_1): Likewise.
33724 (register_edge_assert_for): Likewise.
33725 (finish_register_edge_assert_for): New helper actually registering
33726 asserts where live on edge.
33727 (find_conditional_asserts): Adjust.
33728 (find_switch_asserts): Likewise.
33729 (evrp_dom_walker::try_find_new_range): Generalize.
33730 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33732 2017-04-27 Marek Polacek <polacek@redhat.com>
33735 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33736 arg10 and arg11 to itype.
33738 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33740 * doc/extend.texi (Object Size Checking): Improve grammar.
33742 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33745 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33746 that the logic for permitting reciprocal estimates matches that
33749 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33752 * tree.c (type_cache_hasher::equal): Only compare
33753 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33754 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33755 non-aggregate element types.
33756 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33757 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33759 2017-04-27 Richard Biener <rguenther@suse.de>
33761 PR middle-end/80533
33762 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33763 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33764 keeping a reference to a trailing array.
33766 2017-04-27 Richard Biener <rguenther@suse.de>
33768 PR middle-end/80539
33769 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33770 being in loop-closed SSA form conservatively.
33771 (chrec_fold_multiply_poly_poly): Likewise.
33773 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33775 PR middle-end/79665
33776 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33777 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33779 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33782 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33783 (aarch64_function_arg_alignment): Return unsigned int again, but still
33784 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33785 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33786 Don't emit -Wpsabi note.
33787 (aarch64_function_arg_boundary): Likewise.
33788 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33791 2017-04-26 Nathan Sidwell <nathan@acm.org>
33793 * tree.h (crc32_unsigned_n): Declare.
33794 (crc32_unsigned, crc32_unsigned): Make inline.
33795 * tree.c (crc32_unsigned_bits): Replace with ...
33796 (crc32_unsigned_n): ... this.
33797 (crc32_unsigned, crc32_byte): Remove.
33798 (crc32_string): Remove unnecessary braces.
33800 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33802 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33803 * ipa-inline-analysis.c (MAX_TIME): Remove.
33804 (account_size_time): Use sreal for time.
33805 (dump_inline_summary): Update.
33806 (estimate_function_body_sizes): Update.
33807 (estimate_edge_size_and_time): Update.
33808 (estimate_calls_size_and_time): Update.
33809 (estimate_node_size_and_time): Update.
33810 (inline_merge_summary): Update.
33811 (inline_update_overall_summary): Update.
33812 (estimate_time_after_inlining): Update.
33813 (inline_read_section): Update.
33814 (inline_write_summary): Update.
33815 * ipa-inline.c (compute_uninlined_call_time): Update.
33816 (compute_inlined_call_time): Update.
33817 (recursive_inlining): Update.
33818 (inline_small_functions): Update.
33819 (dump_overall_stats): Update.
33820 * ipa-inline.h: Include sreal.h.
33821 (size_time_entry): Turn time to sreal.
33822 (inline_summary): Turn self_time nad time to sreal.
33824 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33826 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33828 (sreal::stream_out, sreal::stream_in): New.
33829 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33831 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33833 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33836 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33839 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33841 (dimode_scalar_chain::compute_convert_gain): Ditto.
33842 (dimode_scalar_chain::make_vector_copies): Ditto.
33843 (dimode_scalar_chain::convert_reg): Ditto.
33844 (dimode_scalar_chain::convert_insn): Ditto.
33845 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33846 (VI248_AVX512BW_1): New mode iterator.
33847 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33848 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33851 2017-04-25 Martin Sebor <msebor@redhat.com>
33853 PR tree-optimization/80497
33854 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33855 constants are representable in HOST_WIDE_INT.
33856 (parse_directive): Ditto.
33858 2017-04-25 Martin Sebor <msebor@redhat.com>
33861 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33862 (new_zero_array): Adjust signature.
33863 (dom_info::dom_init): Used unsigned rather that size_t.
33864 (dom_info::dom_info): Same.
33866 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33867 Jakub Jelinek <jakub@redhat.com>
33870 * config/arm/arm.c: Include gimple.h.
33871 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33872 returns negative, increment ncrn only if it returned positive.
33873 (arm_needs_doubleword_align): Return int instead of bool,
33874 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33875 members, but if there is any such non-FIELD_DECL
33876 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33877 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33878 returns negative, increment nregs only if it returned positive.
33879 (arm_setup_incoming_varargs): Likewise.
33880 (arm_function_arg_boundary): Emit -Wpsabi note if
33881 arm_needs_doubleword_align returns negative, return
33882 DOUBLEWORD_ALIGNMENT only if it returned positive.
33884 2017-04-25 Marek Polacek <polacek@redhat.com>
33887 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33888 first argument to type.
33890 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33893 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33894 type checks to test for compatibility instead of equality.
33896 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33897 Jakub Jelinek <jakub@redhat.com>
33900 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33902 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33903 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33904 the alignment computation, but return their maximum in warn_alignment.
33905 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33906 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33908 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33909 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33912 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33914 * config/arc/simdext.md (dmpyh): Fix typo.
33916 2017-04-25 Richard Biener <rguenther@suse.de>
33918 PR tree-optimization/80492
33919 * alias.c (compare_base_decls): Handle registers with asm
33920 specification conservatively.
33921 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33922 compare_base_decls returning dont-know properly.
33924 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33926 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33927 (legitimate_offset_address_p): New function.
33928 (arc_legitimate_address_p): Use above function.
33930 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33932 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33934 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33936 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33937 ACCH registers whenever they are available.
33939 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33941 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33942 double regs fix when not used.
33944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33946 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33948 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33949 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33951 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33953 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33954 when emitting short ADD instructions.
33956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33958 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33960 (cmpsi_cc_c_insn): Likewise.
33961 (cbranchsi4_scratch): Compute proper instruction length using
33962 compact_hreg_operand.
33963 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33965 2017-04-25 Richard Biener <rguenther@suse.de>
33967 PR middle-end/80509
33968 * passes.c (pass_manager::pass_manager): Initialize
33969 m_name_to_pass_map.
33971 2017-04-25 Richard Biener <rguenther@suse.de>
33973 PR tree-optimization/79201
33974 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33976 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33979 * config/s390/vector.md: Split MEM->GPR vector moves for
33980 non-s_operand addresses.
33982 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33985 * config/s390/predicates.md (reload_const_wide_int_operand): New
33987 * config/s390/s390.md ("movti"): Remove d/P alternative.
33988 ("movti_bigconst"): New pattern definition.
33990 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33993 * s390-protos.h (s390_expand_cs_hqi): Removed.
33994 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33995 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33996 modes as well as CCZ1mode and CCZmode.
33997 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33998 signature of s390_emit_compare_and_swap.
33999 (s390_expand_cs_hqi): Likewise, make static.
34000 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34001 compare-and-swap, in some cases.
34002 (s390_expand_cs): Wrapper function.
34003 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34005 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34006 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34007 patterns for small and large integers. Forbid symref memory operands.
34008 Move expander to s390.c. Require cc register.
34009 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34010 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34011 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34012 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34013 symref memory operands. Remove CC mode and call s390_match_ccmode
34015 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34017 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34019 * config/s390/s390.md (define_peephole2): New peephole to help
34020 combining the load-and-test pattern with volatile memory.
34022 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34024 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34025 with CCZmode for TARGET_Z196.
34027 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34029 PR rtl-optimization/80501
34030 * combine.c (make_compound_operation_int): Set subreg_code to SET
34031 even for AND with mask of the sign bit of mode.
34033 PR rtl-optimization/80500
34034 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34035 sum's initial value.
34037 2017-04-25 Julian Brown <julian@codesourcery.com>
34038 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34040 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34042 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34044 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34046 2017-04-25 Julian Brown <julian@codesourcery.com>
34047 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34049 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34050 (thunderx2t99_sha): New Reservation.
34052 2017-04-25 Julian Brown <julian@codesourcery.com>
34053 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34055 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34056 type for 1-element load.
34058 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34060 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34062 2017-04-24 Martin Jambor <mjambor@suse.cz>
34064 PR tree-optimization/80293
34065 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34066 char arrays not totally scalarizable if it is false.
34067 (analyze_all_variable_accesses): Pass correct value in the new
34068 parameter. Add a statistics counter.
34070 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34072 PR middle-end/79931
34073 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34075 2017-04-24 Richard Biener <rguenther@suse.de>
34077 PR tree-optimization/80494
34078 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34079 out for complex types.
34081 2017-04-24 Richard Biener <rguenther@suse.de>
34083 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34084 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34085 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34086 (DFS): Adjust and never fail.
34087 (sccvn_dom_walker::fail): Remove.
34088 (sccvn_dom_walker::before_dom_children): Adjust.
34089 (run_scc_vn): Likewise and never fail.
34090 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34091 (pass_fre::execute): Likewise.
34093 2017-04-24 Richard Biener <rguenther@suse.de>
34095 PR tree-optimization/79725
34096 * tree-ssa-sink.c (statement_sink_location): Return whether
34097 failure reason was zero uses. Move that check later.
34098 (sink_code_in_bb): Deal with zero uses by removing the stmt
34101 2017-04-24 Richard Biener <rguenther@suse.de>
34104 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34105 pointer-based references.
34107 2017-04-24 Richard Biener <rguenther@suse.de>
34110 * pass_manager.h (pass_manager::operator new): Remove.
34111 (pass_manager::operator delete): Likewise.
34112 * passes.c (pass_manager::operator new): Remove.
34113 (pass_manager::operator delete): Likewise.
34114 (pass_manager::pass_manager): Zero individual pass members.
34116 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34119 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34120 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34121 Check "XEXP (src, 1)" operand here.
34122 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34123 Check "XEXP (src, 1)" operand here.
34124 (dimode_scalar_chain::make_vector_copies): Detect count register
34125 of a shift instruction. Zero extend count register from QImode
34126 to DImode to satisfy vector shift pattern count operand predicate.
34127 Substitute vector shift count operand with a DImode copy.
34128 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34131 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34133 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34134 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34135 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34136 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34137 (UNSPEC_NOREX_MEM): Remove definition.
34139 2017-04-21 Richard Biener <rguenther@suse.de>
34141 PR tree-optimization/79547
34142 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34143 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34144 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34145 without any constraints.
34147 2017-04-21 Richard Biener <rguenther@suse.de>
34149 PR tree-optimization/78847
34150 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34152 2017-04-21 Richard Biener <rguenther@suse.de>
34154 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34155 (build_distinct_type_copy): Likewise.
34156 (build_variant_type_copy): Likewise.
34157 * tree.c (build_qualified_type): Pass down mem-stat info.
34158 (build_distinct_type_copy): Likewise.
34159 (build_variant_type_copy): Likewise.
34161 2017-04-21 Richard Biener <rguenther@suse.de>
34163 PR tree-optimization/80237
34164 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34166 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34167 for a simplified result.
34169 2016-04-21 Richard Biener <rguenther@suse.de>
34171 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34172 sth as strict as a simple_iv but a chrec without symbols and an
34173 operand defined in the loop we are peeling (and not some subloop).
34174 (propagate_constants_for_unrolling): Propagate all constants.
34176 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34179 * config/i386/i386.c (print_reg): Remove assert for disalowed
34180 regno values, call output_operand_lossage instead.
34182 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34185 * config/i386/constraints.md (Yc): New register constraint.
34186 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34187 Use Yc constraint for alternative 2 of operand 0. Remove
34188 preferred_for_speed attribute.
34190 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34192 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34193 lastprivate clauses in SIMT case.
34195 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34197 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34199 2017-04-20 Richard Biener <rguenther@suse.de>
34201 PR tree-optimization/57796
34202 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34203 as N scalar stores.
34204 (vect_model_load_cost): Cost gathers as N scalar loads.
34206 2017-04-20 Richard Biener <rguenther@suse.de>
34208 * ggc-page.c (ggc_allocated_p): Rename to ...
34209 (safe_lookup_page_table_entry): ... this and return the lookup
34211 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34213 2017-04-20 Richard Biener <rguenther@suse.de>
34215 PR tree-optimization/80453
34216 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34217 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34218 from the conditions.
34219 (vn_phi_eq): Pass them down.
34220 (vn_phi_lookup): Record them.
34221 (vn_phi_insert): Likewise.
34223 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34225 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34226 uninitialized variable warning to avoid buffer overrun.
34228 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34231 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34232 is suppressed for '{ 0 }' in C.
34234 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34236 * BASE-VER: Set to 8.0.0.
34238 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34240 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34241 priority .init_array and .fini_array section with SECTION_NOTYPE
34244 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34246 PR middle-end/80423
34247 * tree.h (build_array_type): Add typeless_storage default argument.
34248 * tree.c (type_cache_hasher::equal): Also compare
34249 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34250 (build_array_type): Add typeless_storage argument, set
34251 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34253 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34254 (build_array_type): Likewise. Add typeless_storage argument.
34256 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34257 Jakub Jelinek <jakub@redhat.com>
34259 PR tree-optimization/80426
34260 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34261 operation on symbolic operands, also compute the overflow for the
34262 invariant part when the operation degenerates into a negation.
34264 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34267 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34268 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34271 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34273 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34276 * config/avr/avr.c (tree.h): Include it.
34277 (cgraph.h): Include it.
34278 (avr_encode_section_info): Don't warn for uninitialized progmem
34279 variable if it's just an alias.
34281 2017-04-19 Richard Biener <rguenther@suse.de>
34284 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34285 when needed by AutoPGO.
34287 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34290 * doc/lto.texi: Remove an extra 'that'.
34292 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34294 PR rtl-optimization/80429
34295 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34296 are only used in debug insns.
34298 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34299 Vladimir Makarov <vmakarov@redhat.com>
34301 * config/sparc/predicates.md (input_operand): Add comment. Return
34302 true for any memory operand when LRA is in progress.
34303 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34305 2017-04-18 Jeff Law <law@redhat.com>
34308 * mips.md ({return,simple_return}_internal): Do not overwrite
34311 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34313 PR tree-optimization/80443
34314 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34315 instead of adding 1, subtract -1 and similarly instead of subtracting
34318 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34320 PR rtl-optimization/80357
34321 * haifa-sched.c (tmp_bitmap): New variable.
34322 (model_recompute): Handle duplicate use records.
34323 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34324 (free_global_sched_pressure_data): Free it.
34326 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34329 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34330 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34331 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34332 instead of SYSTEM_HEADER_DIR.
34334 2017-04-18 Jeff Law <law@redhat.com>
34336 PR middle-end/80422
34337 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34338 predecessors after walking up the insn chain.
34340 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34343 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34344 sizetype type into debug info.
34346 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34349 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34350 unneeded test for TARGET_UPPER_REGS_SF.
34351 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34353 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34356 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34357 instead of gsi_after_labels.
34359 2017-04-18 Jeff Law <law@redhat.com>
34361 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34365 2017-04-13 Jeff Law <law@redhat.com>
34366 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34367 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34369 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34372 * config/avr/avr.c (intl.h): Include it.
34373 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34375 2017-04-18 Martin Liska <mliska@suse.cz>
34377 PR gcov-profile/78783
34378 * gcov-tool.c (gcov_output_files): Validate that destination
34379 file is either removed by the tool or by a user.
34381 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34382 Guy Benyei <guybe@mellanox.com>
34384 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34385 block, and do not negate it, the stored id is already negative.
34387 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34389 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34391 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34394 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34395 masks of options that should be turned off if the VSX vector
34396 options are turned off.
34397 (OTHER_P8_VECTOR_MASKS): Likewise.
34398 (OTHER_VSX_VECTOR_MASKS): Likewise.
34399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34400 rs6000_disable_incompatible_switches to validate no type switches
34402 (rs6000_incompatible_switch): New function to disallow turning on
34403 other vector options if -mno-vsx, -mno-power8-vector, or
34404 -mno-power9-vector are specified.
34406 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34408 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34410 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34412 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34413 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34414 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34415 (ARG_POINTER_CFA_OFFSET): Likewise.
34417 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34419 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34420 conditions to take advantage of various optimizations.
34422 2017-04-13 Jeff Law <law@redhat.com>
34424 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34425 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34426 (zero_extendsidi2_dext): Likewise.
34428 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34431 * fold-const.c (fold_ternary_loc): Revert
34432 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34435 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34437 PR rtl-optimization/80343
34438 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34439 new scratch pseudo.
34441 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34444 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34445 to ubsan_encode_value.
34447 2017-04-13 Jeff Law <law@redhat.com>
34449 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34450 appearing in DEBUG_INSNs.
34452 2017-04-13 Martin Liska <mliska@suse.cz>
34454 PR gcov-profile/80413
34455 * gcov-io.c (gcov_write_string): Copy to buffer just when
34456 allocated size is greater than zero.
34458 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34461 * dwarf2out.c (decls_for_scope): Ignore declarations of
34462 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34464 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34467 * ipa-visibility.c (non_local_p): Fix typos.
34468 (localize_node): When localizing symbol in same comdat group,
34469 dissolve the group only when we know external symbols are going
34471 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34473 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34475 PR tree-optimization/79390
34476 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34477 order does not result in usable sequence, retry with reversed operand
34483 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34484 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34485 op0 instead of fold_convert_loc (loc, type, arg0).
34487 2017-04-12 Jeff Law <law@redhat.com>
34489 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34490 has a delay slot in the generated code.
34492 * config/cris/cris.md (cris_preferred_reload_class): Return
34493 GENNONACR_REGS rather than GENERAL_REGS.
34495 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34498 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34499 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34500 signedness of the result type.
34502 2017-04-12 Richard Biener <rguenther@suse.de>
34503 Jeff Law <law@redhat.com>
34505 PR tree-optimization/80359
34506 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34507 trim stores to TARGET_MEM_REFs.
34509 2017-04-12 Richard Biener <rguenther@suse.de>
34511 PR tree-optimization/79390
34512 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34513 threading case even more.
34515 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34518 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34519 for quad_address_p for TImode, instead of just not indexed_address.
34521 2017-04-12 Richard Biener <rguenther@suse.de>
34522 Bernd Edlinger <bernd.edlinger@hotmail.de>
34524 PR middle-end/79671
34525 * alias.c (component_uses_parent_alias_set_from): Handle
34526 TYPE_TYPELESS_STORAGE.
34527 (get_alias_set): Likewise.
34528 * tree-core.h (tree_type_common): Add typeless_storage flag.
34529 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34530 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34531 for types containing members with TYPE_TYPELESS_STORAGE.
34532 (place_field): Likewise.
34533 (layout_type): Likewise for ARRAY_TYPE.
34534 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34535 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34536 TYPE_TYPELESS_STORAGE.
34537 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34539 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34542 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34543 first argument to type.
34545 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34549 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34550 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34551 (rs6000_expand_binop_builtin): Likewise.
34552 (rs6000_expand_ternop_builtin): Likewise; also add missing
34553 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34554 vshasigma built-ins.
34555 * doc/extend.texi: Document that vec_xxpermdi's third argument
34556 must be a constant.
34558 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34560 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34561 Use shift_const cost parameter when calculating gain of STV shifts.
34563 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34565 PR rtl-optimization/70478
34566 * lra-constraints.c (process_alt_operands): Check memory for
34567 disfavoring memory insn operand.
34569 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34571 PR middle-end/80100
34572 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34573 left shift in unsigned HOST_WIDE_INT type.
34575 PR rtl-optimization/80385
34576 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34577 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34580 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34581 if they have any depend clauses.
34583 2017-04-11 Martin Liska <mliska@suse.cz>
34586 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34587 * ipa-split.c (split_function): Create a local comdat symbol
34588 if caller is in a comdat group.
34590 2017-04-11 Martin Liska <mliska@suse.cz>
34593 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34596 2017-04-11 Martin Sebor <msebor@redhat.com>
34598 PR middle-end/80364
34599 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34600 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34602 (directive::set_width, directive::set_precision, format_character):
34604 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34607 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34610 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34611 conflict, set target->arch_name instead of target->cpu_name.
34613 2017-04-11 Richard Biener <rguenther@suse.de>
34615 PR tree-optimization/80374
34616 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34617 build_zero_cst, remove fold_convertible_p check again.
34619 2017-04-11 Martin Liska <mliska@suse.cz>
34622 * ubsan.c (instrument_object_size): Do not instrument register
34625 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34628 * config/i386/i386-builtin-types.def
34629 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34630 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34631 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34632 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34633 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34634 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34635 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34636 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34637 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34638 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34639 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34640 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34641 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34642 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34643 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34644 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34645 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34646 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34647 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34648 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34649 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34650 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34651 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34652 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34653 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34654 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34655 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34656 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34657 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34658 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34659 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34660 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34661 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34662 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34663 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34664 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34665 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34666 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34667 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34668 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34669 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34670 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34671 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34672 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34673 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34674 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34676 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34677 flag to second_arg_count, handle 4 argument function type _COUNT
34678 aliases, handle second_arg_count on second argument rather than last.
34680 2017-04-10 Jeff Law <law@redhat.com>
34682 PR tree-optimization/80374
34683 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34684 record anything if we can not convert integer_zero_node to the
34687 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34690 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34691 Enhance special handling given to the TARGET_P9_MINMAX option in
34692 relation to certain other options.
34694 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34696 PR tree-optimization/80153
34697 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34698 remove POINTER_PLUS_EXPR's base part directly, rather than through
34701 2017-04-10 Richard Biener <rguenther@suse.de>
34702 Bin Cheng <bin.cheng@arm.com>
34704 PR tree-optimization/80153
34705 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34706 the first element of pointer type aff_tree. Build result expr in
34708 (add_elt_to_tree): Convert to type unconditionally. Remove other
34709 fold_convert calls.
34710 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34711 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34713 2017-04-10 Richard Biener <rguenther@suse.de>
34715 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34718 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34720 PR rtl-optimization/70478
34721 * lra-constraints.c (curr_small_class_check): New.
34722 (update_and_check_small_class_inputs): New.
34723 (process_alt_operands): Update curr_small_class_check. Disfavor
34724 alternative insn memory operands. Check available regs for small
34727 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34730 * config/mips/mips.opt (-mvirt): Update description.
34731 * doc/invoke.texi (-mvirt): Likewise.
34733 2017-04-10 Richard Biener <rguenther@suse.de>
34735 PR middle-end/80362
34736 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34737 looking for NEGATE_EXPR in -A / -B to A / B folding.
34739 2017-04-10 Martin Liska <mliska@suse.cz>
34741 PR gcov-profile/80224
34742 * gcov.c (print_usage): Fix usage string.
34743 (get_gcov_intermediate_filename): Remove.
34744 (output_gcov_file): Use both for normal and intermediate format.
34745 (generate_results): Do not initialize special file for
34746 intermediate format.
34748 2017-04-10 Richard Biener <rguenther@suse.de>
34750 PR tree-optimization/80304
34751 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34754 2017-04-10 Nathan Sidwell <nathan@acm.org>
34757 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34758 (rs6000_init_builtins): Use it.
34760 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34762 * config/arm/arm.md (<mrc>): Add mode to SET source.
34763 (<mrrc>): Likewise.
34765 2017-04-10 Richard Biener <rguenther@suse.de>
34767 PR middle-end/80344
34768 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34770 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34773 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34774 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34775 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34776 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34777 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34778 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34779 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34780 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34781 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34782 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34783 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34784 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34785 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34786 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34787 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34788 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34789 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34790 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34791 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34792 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34793 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34794 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34795 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34797 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34799 PR rtl-optimization/70478
34800 * lra-constraints.c: Reverse the last patch.
34802 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34804 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34805 Add comment for WCHAR_T.
34807 2017-04-08 Martin Liska <mliska@suse.cz>
34810 2017-04-07 Martin Liska <mliska@suse.cz>
34813 * ipa-split.c (split_function): Add function part to a same comdat
34816 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34819 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34821 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34823 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34824 (vec_pairing): ...to this.
34825 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34826 (rs6000_sched_init): Adjust for name change.
34827 (struct rs6000_sched_context): Likewise.
34828 (rs6000_init_sched_context): Likewise.
34829 (rs6000_set_sched_context): Likewise.
34831 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34837 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34839 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34840 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34841 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34843 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34845 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34847 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34849 PR rtl-optimization/70703
34850 * ira-color.c (update_conflict_hard_regno_costs): Use
34851 int64_t instead of HOST_WIDE_INT.
34853 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34855 PR rtl-optimization/70478
34856 * lra-constraints.c (process_alt_operands): Disfavor alternative
34857 insn memory operands.
34859 2017-04-07 Jeff Law <law@redhat.com>
34861 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34862 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34864 2017-04-07 Martin Liska <mliska@suse.cz>
34867 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34868 Show error message instead of an ICE.
34870 2017-04-07 Martin Liska <mliska@suse.cz>
34873 * ipa-split.c (split_function): Add function part to a same comdat
34876 2017-04-07 Richard Biener <rguenther@suse.de>
34878 PR middle-end/80341
34879 * tree.c (get_unwidened): Also handle ! for_type case for
34881 * convert.c (do_narrow): Split out from ...
34882 (convert_to_integer_1): ... here. Do not pass final truncation
34883 type to get_unwidened for TRUNC_DIV_EXPR.
34885 2017-04-07 Richard Biener <rguenther@suse.de>
34887 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34889 (aff_combination_const): Adjust.
34890 (aff_combination_scale): Likewise.
34891 (aff_combination_add_elt): Likewise.
34892 (aff_combination_add_cst): Likewise.
34893 (aff_combination_convert): Likewise.
34894 (add_elt_to_tree): Likewise. Remove unused argument.
34895 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34897 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34899 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34901 * config/arm/arm.c (arm_default_short_enums): Use
34902 ARM_DEFAULT_SHORT_ENUMS.
34903 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34905 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34908 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34909 members with redundant out-of-class redeclaration.
34911 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34914 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34915 * config/i386/i386.md (*zero_extendsidi2):
34916 Add (?*x,*x) and (?*v,*v) alternatives.
34918 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34921 * config/i386/i386.c (ix86_expand_builtin)
34922 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34923 mode from insn data. Convert operands to insn operand mode.
34924 Copy operands that don't satisfy insn predicate to a register.
34926 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34928 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34931 2017-04-06 Richard Biener <rguenther@suse.de>
34933 PR tree-optimization/80334
34934 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34935 preserve alignment of accesses.
34937 2017-04-06 Richard Biener <rguenther@suse.de>
34939 PR tree-optimization/80262
34940 * tree-sra.c (build_ref_for_offset): Preserve address-space
34942 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34943 Drop useless address-space information on MEM_REF offsets.
34945 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34947 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34949 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34951 PR rtl-optimization/70703
34952 * ira-color.c (update_conflict_hard_regno_costs): Use
34953 HOST_WIDE_INT instead of long.
34955 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34958 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34959 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34961 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34962 for x86_64 target. Handle -m3dnowa option.
34964 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34966 PR rtl-optimization/70703
34967 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34968 (update_conflict_hard_regno_costs): Use long instead of unsigned
34969 arithmetic for cost calculation.
34971 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34972 Bernd Edlinger <bernd.edlinger@hotmail.de>
34975 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34978 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34981 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34982 ptr_mode with Pmode throughout.
34983 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34984 into probe_stack_range and use DImode.
34986 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34989 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34990 call_eh_return is true.
34992 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34994 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34995 Initialize last_match_fntype_index.
34997 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35000 * tree-nvr.c: Include internal-fn.h.
35001 (pass_return_slot::execute): Ignore internal calls without
35004 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35005 Richard Biener <rguenther@suse.de>
35008 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35009 captures used multiple times, except for the last use.
35010 * generic-match-head.c: Include gimplify.h.
35012 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35014 PR tree-optimization/79390
35015 * target.h (struct noce_if_info): Declare.
35016 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35017 * target.def (noce_conversion_profitable_p): New target hook.
35018 * ifcvt.h (struct noce_if_info): New type, moved from ...
35019 * ifcvt.c (struct noce_if_info): ... here.
35020 (noce_conversion_profitable_p): Renamed to ...
35021 (default_noce_conversion_profitable_p): ... this. No longer
35023 (noce_try_store_flag_constants, noce_try_addcc,
35024 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35025 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35026 instead of noce_conversion_profitable_p.
35027 * config/i386/i386.c: Include ifcvt.h.
35028 (ix86_option_override_internal): Don't override
35029 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35030 (ix86_noce_conversion_profitable_p): New function.
35031 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35032 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35033 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35034 * doc/tm.texi: Regenerated.
35036 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35038 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35041 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35044 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35045 instructions for small multiply cores.
35047 2017-04-04 Jeff Law <law@redhat.com>
35049 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35051 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35052 that are not set by the loop over the elements.
35054 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35057 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35058 int mode, convert_modes it to mode as unsigned, otherwise use
35059 lowpart_subreg to mode rather than SImode.
35060 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35061 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35062 Use DImode instead of SImode for the shift count operand.
35063 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35066 2017-04-04 Richard Biener <rguenther@suse.de>
35068 PR middle-end/80281
35069 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35070 arithmetic done for the negate or the plus. Simplify.
35071 (A - (-B) -> A + B): Likewise.
35072 * fold-const.c (split_tree): Make sure to not negate pointers.
35074 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35076 PR rtl-optimization/60818
35077 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35078 a compare of comparisons with the thing compared if this results
35079 in a different machine mode.
35081 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35083 * alias.c (base_alias_check): Fix typo in comment.
35084 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35085 * cgraphunit.c (symbol_table::compile): Likewise.
35086 * collect2.c (maybe_run_lto_and_relink): Likewise.
35087 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35088 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35089 * config/avr/avr.c (avr_map_op_t): Likewise.
35090 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35091 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35092 * config/epiphany/epiphany.md (movcc): Likewise.
35093 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35094 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35096 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35097 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35098 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35099 * config/sh/sh.c (sh_rtx_costs): Likewise.
35100 * fold-const.c (fold_truth_andor): Likewise.
35101 * genautomata.c (collapse_flag): Likewise.
35102 * gengtype.h (struct type::u::s): Likewise.
35103 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35104 * input.c (FORMAT_AMOUNT): Likewise.
35105 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35106 (known_aggs_to_agg_replacement_list): Likewise.
35107 * ipa-inline-analysis.c: Likewise.
35108 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35109 * ipa-polymorphic-call.c
35110 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35111 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35112 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35114 * modulo-sched.c (apply_reg_moves): Likewise.
35115 * omp-expand.c (build_omp_regions_1): Likewise.
35116 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35117 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35118 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35119 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35120 * value-prof.c: Likewise.
35121 * var-tracking.c (val_reset): Likewise.
35123 2017-04-03 Richard Biener <rguenther@suse.de>
35125 PR tree-optimization/80275
35126 * fold-const.c (split_address_to_core_and_offset): Handle
35129 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35131 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35132 descriptors is at least equal to that of functions.
35134 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35136 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35138 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35141 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35142 (mov<IMOD4:mode>): New expander.
35143 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35145 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35147 PR rtl-optimization/79405
35148 * fwprop.c (propagations_left): New variable.
35149 (forward_propagate_into): Decrement it.
35150 (fwprop_init): Initialize it.
35151 (fw_prop): If the variable has reached zero, stop propagating.
35152 (fwprop_addr): Ditto.
35154 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35157 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35158 a FUNCTION_DECL, pass it as decl instead of origin to
35161 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35163 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35166 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35169 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35170 TARGET_VSX_SMALL_INTEGER.
35172 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35174 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35175 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35177 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35179 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35180 extraction from odd-numbered MSA register.
35182 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35184 PR middle-end/80173
35185 * expmed.c (store_bit_field_1): Don't attempt to create
35186 a word subreg out of hard registers wider than word if they
35187 have HARD_REGNO_NREGS of 1 for their mode.
35189 PR middle-end/80163
35190 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35191 conversions to integer types wider than word and pointer.
35194 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35195 (rtx_equal_for_cselib_p): Pass 0 to it.
35196 * cselib.c (cselib_hasher::equal): Likewise.
35197 (rtx_equal_for_cselib_1): Add depth argument. If depth
35198 is 128, don't look up VALUE locs and punt. Increment
35199 depth in recursive calls when walking VALUE locs.
35201 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35203 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35204 (make_gcov_file_name): Use the canonical path name for generating
35206 (read_line): Fix handling of files with ascii null bytes.
35208 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35210 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35211 to initialise a vector register instead
35212 of using a const_int.
35214 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35216 PR translation/80189
35217 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35218 diagnostic messages.
35220 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35223 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35224 (dfp_diex_<mode>): Update mode of operand 1.
35225 * doc/extend.texi (dxex, dxexq): Document change to return type.
35226 (diex, diexq): Document change to argument type.
35228 2017-03-30 Martin Jambor <mjambor@suse.cz>
35231 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35232 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35233 it reflects the signature changes performed at the callee side.
35234 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35235 to cgraph_build_function_type_skip_args.
35236 (build_function_decl_skip_args): Adjust call to the above function.
35238 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35241 * config/i386/sse.md
35242 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35243 register as dest whenever it is a MEM not rtx_equal_p to the
35244 corresponding dup operand, and when forcing into reg move the
35245 reg into the memory afterwards.
35246 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35247 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35248 for the force_reg mode.
35249 (avx512vl_vextractf128<mode>): Use register as dest either
35250 always when a MEM, or when it is a MEM not rtx_equal_p to the
35251 corresponding dup operand, or even not when it is a CONST_VECTOR
35252 depending on the mode and lo vs. hi.
35253 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35255 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35256 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35257 Likewise. Require that operands[2] is even.
35258 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35259 Remove extraneous parens. Require that operands[2] is a multiple
35261 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35262 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35263 disallow memory then.
35265 2017-03-30 Richard Biener <rguenther@suse.de>
35267 PR tree-optimization/77498
35268 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35269 to non-constants over backedges.
35271 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35273 PR rtl-optimization/80233
35274 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35275 as last_combined_insn. Do not test for BARRIER_P separately.
35277 2017-03-29 Andreas Schwab <schwab@suse.de>
35280 * calls.c (prepare_call_address): Convert funexp to Pmode before
35281 copying to temp reg.
35283 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35285 PR tree-optimization/80158
35286 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35287 Handle possible future case of more than one alternate
35289 (replace_rhs_if_not_dup): Likewise.
35290 (replace_one_candidate): Likewise.
35292 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35294 PR rtl-optimization/80193
35295 * ira.c (ira): Do not check allocation for LRA.
35297 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35299 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35300 (nvptx_output_simt_exit): Declare.
35301 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35302 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35303 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35304 (nvptx_declare_function_name): ...here. Emit declaration of local
35305 memory space buffer for omp_simt_enter insn.
35306 (nvptx_output_unisimt_switch): New.
35307 (nvptx_output_softstack_switch): New.
35308 (nvptx_output_simt_enter): New.
35309 (nvptx_output_simt_exit): New.
35310 * config/nvptx/nvptx.h (struct machine_function): New fields
35311 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35312 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35313 (UNSPECV_SIMT_EXIT): Ditto.
35314 (omp_simt_enter_insn): New insn.
35315 (omp_simt_enter): New expansion.
35316 (omp_simt_exit): New insn.
35317 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35319 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35320 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35321 (expand_GOMP_SIMT_EXIT): New.
35322 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35323 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35324 (GOMP_SIMT_EXIT): Ditto.
35325 * target-insns.def (omp_simt_enter): New insn.
35326 (omp_simt_exit): Ditto.
35327 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35329 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35330 (lower_rec_input_clauses): Likewise.
35331 (lower_lastprivate_clauses): Handle SIMT privatization.
35333 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35334 (ompdevlow_adjust_simt_enter): New.
35335 (find_simtpriv_var_op): New.
35336 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35337 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35339 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35340 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35341 (copy_decl_for_dup_finish): Ditto.
35343 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35345 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35348 * config/i386/i386.c (ix86_option_override_internal): Always
35349 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35351 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35353 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35355 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35357 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35358 mark new edge's irreducible flag accordign to it.
35359 (vect_do_peeling): Check loop preheader edge's irreducible flag
35360 and pass it to function slpeel_add_loop_guard.
35362 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35364 PR tree-optimization/80218
35365 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35366 Update block frequencies and counts.
35368 2017-03-28 Richard Biener <rguenther@suse.de>
35370 PR tree-optimization/78644
35371 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35372 of a simplification result we may not use it at all.
35374 2017-03-28 Richard Biener <rguenther@suse.de>
35377 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35378 without arguments, generate default definition of a SSA name.
35380 2017-03-28 Richard Biener <rguenther@suse.de>
35382 PR middle-end/80222
35383 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35384 TYPE_REF_CAN_ALIAS_ALL references.
35385 * fold-const.c (fold_indirect_ref_1): Likewise.
35387 2017-03-28 Martin Liska <mliska@suse.cz>
35390 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35391 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35393 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35394 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35396 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35397 (EXTRA_SPECS): Define.
35398 (SUBTARGET_EXTRA_SPECS): Likewise.
35399 (SUBTARGET_CPP_SPEC): Likewise.
35400 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35401 SUBTARGET_EXTRA_SPECS.
35402 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35404 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35406 * config/arc/simdext.md (vst64_insn): Update pattern.
35407 (vld32wh_insn): Likewise.
35408 (vld32wl_insn): Likewise.
35409 (vld64_insn): Likewise.
35410 (vld32_insn): Likewise.
35412 2017-03-28 Marek Polacek <polacek@redhat.com>
35415 * fold-const.c (fold_comparison): Use protected_set_expr_location
35416 instead of SET_EXPR_LOCATION.
35418 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35420 * tree.c (add_expr): Avoid name lookup warning.
35422 2017-03-27 Jeff Law <law@redhat.com>
35424 PR tree-optimization/80216
35425 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35426 function name. Limit recursion depth.
35427 (record_temporary_equivalences): Corresponding changes.
35429 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35431 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35434 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35437 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35439 * cfgcleanup.c (reg_note_cfa_p): New array.
35440 (insns_have_identical_cfa_notes): New function.
35441 (old_insns_match_p): Don't cross-jump in between /f
35442 and non-/f instructions. If both i1 and i2 are frame related,
35443 verify all CFA notes, their order and content.
35445 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35448 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35449 HImode and SImode with zero extend to DImode to one insn.
35450 (bswap<mode>2_extenddi): Likewise.
35451 (bswapsi2_extenddi): Likewise.
35452 (bswaphi2_extendsi): Likewise.
35453 (bswaphi2): Combine bswap HImode and SImode into one insn.
35454 Separate memory insns from swapping register.
35455 (bswapsi2): Likewise.
35456 (bswap<mode>2): Likewise.
35457 (bswaphi2_internal): Delete, no longer used.
35458 (bswapsi2_internal): Likewise.
35459 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35460 store, and gpr<-gpr swap insns.
35461 (bswap<mode>2_store): Likewise.
35462 (bswaphi2_reg): Register only splitter, combine with the splitter.
35463 (bswaphi2 splitter): Likewise.
35464 (bswapsi2_reg): Likewise.
35465 (bswapsi2 splitter): Likewise.
35466 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35467 the insns into load, store, and register/register insns.
35468 (bswapdi2_ldbrx): Likewise.
35469 (bswapdi2_load): Likewise.
35470 (bswapdi2_store): Likewise.
35471 (bswapdi2_reg): Likewise.
35473 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35475 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35476 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35478 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35481 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35484 special handling for target option conflicts between dform
35485 options (-mpower9-dform, -mpower9-dform-vector,
35486 -mpower9-dform-scalar) and -mno-direct-move.
35488 2017-03-27 Richard Biener <rguenther@suse.de>
35490 PR tree-optimization/80181
35491 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35493 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35495 * config/arc/predicates.md (move_double_src_operand): Replace the
35496 call to move_double_src_operand with a call to address_operand.
35498 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35500 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35501 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35502 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35504 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35506 * config/arc/predicates.md (long_immediate_loadstore_operand):
35507 Consider scaled addresses cases.
35509 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35511 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35512 restored when in interrupt.
35513 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35514 doesn't have delay slot.
35516 2017-03-27 Richard Biener <rguenther@suse.de>
35519 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35520 inlined thunk clones.
35522 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35525 * asan.c (instrument_derefs): Copy over last operand from
35526 original COMPONENT_REF to the new COMPONENT_REF with
35527 DECL_BIT_FIELD_REPRESENTATIVE.
35528 * ubsan.c (instrument_object_size): Likewise.
35530 2017-03-27 Richard Biener <rguenther@suse.de>
35532 PR tree-optimization/80170
35533 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35534 sure DR/SCEV didnt fold in constants we do not see when looking
35535 at the reference base alignment.
35537 2017-03-27 Richard Biener <rguenther@suse.de>
35539 PR middle-end/80171
35540 * gimple-fold.c (fold_ctor_reference): Properly guard against
35541 NULL return value from canonicalize_constructor_val.
35543 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35546 * config/i386/i386.c (ix86_expand_builtin)
35547 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35548 flags reg setting and flags reg using instructions.
35549 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35550 clobbering instructions to zero extend op2.
35552 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35554 * doc/install.texi (Configuration) <--with-aix-soname>:
35555 Update link to AIX ld.
35557 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35559 PR rtl-optimization/80160
35560 PR rtl-optimization/80159
35561 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35562 reg_alternate_class into account.
35564 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35567 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35568 to consider in curr_insn_transform.
35570 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35572 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35573 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35574 and emit_mode_inner.
35576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35578 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35579 argument to the overloaded builtin variants. Use the new flag to
35580 deprecate certain builtin variants.
35581 * config/s390/s390-builtin-types.def: Add new builtin types.
35582 * config/s390/s390-builtins.h: Support new flags field for
35583 overloaded builtins.
35584 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35585 (s390_macro_to_expand): Enable vector float data type.
35586 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35587 builtins by incrementing the __VEC__ version number.
35588 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35590 (s390_resolve_overloaded_builtin): Emit error messages depending
35591 on the builtin flags.
35592 * config/s390/s390.c (s390_expand_builtin): Support additional
35593 flags argument. Change error message to match the messages
35594 emitted in s390-c.c.
35595 * config/s390/s390.md: New UNSPEC_* constants.
35596 (op_type): Add new instruction types.
35597 * config/s390/vecintrin.h: Add new builtins and test data class
35599 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35600 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35601 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35602 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35603 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35604 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35606 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35607 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35608 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35609 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35611 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35612 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35613 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35614 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35616 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35617 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35618 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35619 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35620 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35621 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35622 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35624 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35625 ("vec_scatter_element<V_HW_4:mode>_DI")
35626 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35627 ("vec_fpint<mode>", "vflls")
35628 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35629 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35630 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35631 ("*vec_cmphe<mode>_cc"): ... these.
35633 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35634 mode constant instead of magic value.
35636 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35638 * config/s390/s390.c (s390_expand_vec_compare): Support other
35639 vector floating point modes than just V2DF.
35640 (s390_expand_vcond): Likewise.
35641 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35642 (s390_cannot_change_mode_class): Prevent mode changes between TF
35643 and V1TF in vector registers.
35644 * config/s390/s390.md (DF, SF): New mode attributes.
35645 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35646 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35647 SFmode support for VRs.
35648 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35650 (VFT, VF_HW): New mode iterators.
35651 (vw, sdx): New mode attributes.
35652 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35653 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35654 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35655 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35656 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35657 also the new vector floating point modes. Renaming to ...
35659 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35660 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35661 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35662 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35663 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35664 ("vec_unordered<mode>"): ... these.
35666 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35667 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35668 ("*vec_extendv2df"): New insn definitions.
35670 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35672 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35673 ("mulditi3_2", "*muldi3_sign"): New patterns.
35674 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35675 rename the pattern definition.
35677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35679 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35681 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35683 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35685 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35686 instruction if possible.
35687 * config/s390/vector.md (vec_halfnumelts): New mode
35689 ("*vec_vllezlf<mode>"): New pattern.
35691 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35693 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35694 ("popcountv4si2", "popcountv2di2"): Rename to ...
35695 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35696 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35698 ("popcount<mode>2_vxe"): New pattern.
35700 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35702 * common/config/s390/s390-common.c (processor_flags_table): Add
35704 * config.gcc: Add arch12.
35705 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35706 Default to arch12 for unknown CPU model numbers.
35707 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35708 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35709 PROCESSOR_max sanity check.
35710 * config/s390/s390-opts.h (enum processor_type): Add
35712 * config/s390/s390.c (processor_table): Add arch12.
35713 (s390_expand_builtin): Add check for B_VXE flag.
35714 (s390_issue_rate): Add PROCESSOR_ARCH12.
35715 (s390_get_sched_attrmask): Likewise.
35716 (s390_get_unit_mask): Likewise.
35717 (s390_sched_score): Enable z13 scheduling for arch12.
35718 (s390_sched_reorder): Likewise.
35719 (s390_sched_variable_issue): Likewise.
35720 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35722 (s390_tune_attr): Use z13 scheduling also for arch12.
35723 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35724 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35725 (TARGET_VXE_P): New macros.
35726 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35727 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35728 * config/s390/s390.opt: Add arch12 as processor_type.
35730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35732 * config/s390/s390.md
35733 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35734 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35735 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35737 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35738 Rename expanders to ...
35740 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35741 ("fixuns_truncdddi2_emu"): ... these.
35743 ("fixuns_trunc<mode>si2_emu"): New expander.
35745 ("*fixuns_truncdfdi2_z13"): Rename to ...
35746 ("*fixuns_truncdfdi2_vx"): ... this.
35748 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35750 * config/s390/2964.md: Remove the single element vector compare
35751 instructions which are no longer used.
35752 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35754 (s390_canonicalize_comparison): Remove handling of DFmode
35756 (s390_expand_vec_compare_scalar): Remove function.
35757 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35758 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35760 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35764 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35766 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35767 will anyway by matched by mov<mode>_64dfp.
35769 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35771 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35772 vlef/vstef. Add missing operand to vleif.
35774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35776 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35777 pair for all vector types with 64 bit elements.
35778 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35779 * config/s390/vector.md (V_HW_64): ... here.
35780 (V_128_NOSINGLE): New mode iterator.
35781 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35782 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35783 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35784 ("*vec_load_pairv2di"): Change to ...
35785 ("*vec_load_pair<mode>"): ... this one.
35787 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35789 * config/s390/constraints.md: Add comments.
35790 (jKK): Reject element sizes > 8 bytes.
35791 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35793 * config/s390/s390.md: Add the s_operand checks formerly in
35794 s390_split_ok_p to various splitters where they are still
35796 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35797 for 128 bit vectors. Plus two splitters.
35799 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35801 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35804 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35807 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35808 error if the boundary argument is not constant.
35810 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35812 PR rtl-optimization/80112
35813 * loop-doloop.c (doloop_condition_get): Don't check condition
35814 if cmp isn't SET with IF_THEN_ELSE src.
35816 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35818 PR tree-optimization/80158
35819 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35820 replacing a candidate statement, also replace it for the
35821 candidate's alternate interpretation.
35822 (replace_rhs_if_not_dup): Likewise.
35823 (replace_one_candidate): Likewise.
35825 2017-03-24 Richard Biener <rguenther@suse.de>
35827 PR tree-optimization/80167
35828 * graphite-isl-ast-to-gimple.c
35829 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35831 (translate_isl_ast_to_gimple::get_rename): Likewise.
35833 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35836 handling of certain combinations of target options, including the
35837 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35838 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35840 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35843 * config/arm/arm.md (*load_multiple): Add reload_completed to
35844 matching condition.
35846 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35847 Richard Biener <rguenth@suse.de>
35849 PR tree-optimization/79908
35850 PR tree-optimization/80136
35851 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35852 been cast away, gimplify_and_add suffices.
35854 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35856 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35858 2017-03-23 Richard Biener <rguenther@suse.de>
35860 PR tree-optimization/80032
35861 * gimplify.c (gimple_push_cleanup): Forced unconditional
35862 cleanups still have to go to the conditional_cleanups
35865 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35867 PR tree-optimization/80072
35868 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35870 (next_operand_entry_id): Change type to unsigned int.
35871 (sort_by_operand_rank): Make sure to return the right return value
35872 even if unsigned fields are bigger than INT_MAX.
35873 (struct oecount): Change cnt and id type to unsigned int.
35874 (oecount_hasher::equal): Formatting fix.
35875 (oecount_cmp): Make sure to return the right return value
35876 even if unsigned fields are bigger than INT_MAX.
35877 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35880 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35881 TREE_READONLY on result if writing it more than once.
35884 * doc/invoke.texi (-fsanitize=thread): Document that with
35885 -fnon-call-exceptions atomics are not able to throw
35889 * tsan.c: Include tree-eh.h.
35890 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35891 maybe_clean_or_replace_eh_stmt where needed.
35892 (instrument_memory_accesses): Add cfg_changed argument.
35893 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35894 if it returned true.
35895 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35897 PR rtl-optimization/63191
35898 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35899 wrapper function, moved the whole old content into ...
35900 (ix86_delegitimize_address_1): ... this. New inline function.
35901 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35902 true as last argument instead of ix86_delegitimize_address.
35904 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35906 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35907 cortexa57_branch_cost.
35909 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35911 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35913 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35916 * doc/md.texi (Constraints): Document wA constraint.
35917 * config/rs6000/constraints.md (wA): New.
35918 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35919 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35920 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35921 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35923 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35926 * gimplify.c (is_oacc_declared): New function.
35927 (oacc_default_clause): Use it to set default flags for acc declared
35928 variables inside parallel regions.
35929 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35930 declared variables.
35931 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35932 declare attribute to any decl as necessary.
35934 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35937 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35938 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35939 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35940 (arm_arch_lpae): This.
35941 * config/arm/arm.c (arm_arch7ve): Rename into ...
35942 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35943 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35946 2017-03-22 Martin Liska <mliska@suse.cz>
35949 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35950 error message instead of an ICE.
35952 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35954 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35956 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35958 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35960 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35963 2017-03-21 Martin Sebor <msebor@redhat.com>
35965 * doc/extend.texi: Use "cannot" instead of "can't."
35966 * doc/hostconfig.texi: Same.
35967 * doc/install.texi: Same.
35968 * doc/invoke.texi: Same.
35969 * doc/loop.texi: Same.
35970 * doc/md.texi: Same.
35971 * doc/objc.texi: Same.
35972 * doc/rtl.texi: Same.
35973 * doc/tm.texi: Same.
35974 * doc/tm.texi.in: Same.
35975 * doc/trouble.texi: Same.
35977 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35980 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35981 (collect_checksum_attributes): Set it.
35982 (die_checksum_ordered): Use it.
35984 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35986 PR tree-optimization/79908
35987 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35988 change: For a VA_ARG whose LHS has been cast away, use
35989 force_gimple_operand to construct the side effects.
35991 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35993 PR translation/80001
35994 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35995 more amenable to translation.
35996 (oacc_loop_auto_partitions): Likewise.
35998 2017-03-21 Marek Polacek <polacek@redhat.com>
35999 Martin Sebor <msebor@redhat.com>
36001 PR tree-optimization/80109
36002 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36003 on INTEGRAL_TYPE_P.
36005 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36006 Segher Boessenkool <segher@kernel.crashing.org>
36009 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36010 check reg_used_between_p between insn and one of succ or succ2
36011 depending on if succ is artificial insn not inserted into insn
36014 2017-03-21 Martin Liska <mliska@suse.cz>
36016 PR gcov-profile/80081
36017 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36018 * doc/gcc.texi: Include gcov-dump stuff.
36019 * doc/gcov-dump.texi: New file.
36021 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36023 PR rtl-optimization/79150
36024 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36025 conditional jump, if the jump is the last insn of the loop.
36027 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36028 Richard Biener <rguenth@suse.de>
36030 PR tree-optimization/79908
36031 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36032 been cast away, use force_gimple_operand to construct the side
36035 2017-03-21 Martin Liska <mliska@suse.cz>
36037 PR libfortran/79956
36038 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36041 2017-03-21 Brad Spengler <spender@grsecurity.net>
36044 * plugin.c (htab_hash_plugin): New function.
36045 (add_new_plugin): Use it and adjust.
36046 (parse_plugin_arg_opt): Adjust.
36047 (init_one_plugin): Likewise.
36049 2017-03-21 Richard Biener <rguenther@suse.de>
36051 PR tree-optimization/80032
36052 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36053 if set force the cleanup to happen unconditionally.
36054 (gimplify_target_expr): Push inserted clobbers with force_uncond
36055 to avoid them being removed by control-dependent DCE.
36057 2017-03-21 Richard Biener <rguenther@suse.de>
36059 PR tree-optimization/80122
36060 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36061 va_arg_pack_len when the inlined call stmt requires pack
36063 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36065 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36068 * tsan.c (instrument_builtin_call): If the memory model argument
36069 is not a constant, assume it is valid.
36072 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36075 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36077 PR rtl-optimization/79910
36078 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36079 if its dest is used by an insn before I2 (other than the combined
36080 insns themselves, which are properly handled already).
36082 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36085 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36087 * combine.c (record_used_regs): New static function.
36088 (try_combine): Handle situations where there is an additional
36089 instruction between I2 and I3 which needs to have a LOG_LINK
36093 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36095 * combine.c (try_combine): Delete redundant i1 test. Call
36096 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36098 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36101 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36102 alternatives 13/14.
36104 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36106 PR tree-optimization/80054
36107 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36108 the optimization if a PHI or any of its arguments is not dominated
36109 by the candidate's basis. Use gphi* rather than gimple* as
36111 (replace_profitable_candidates): Clean up a gimple* variable that
36112 should be a gphi* variable.
36114 2017-03-20 Martin Sebor <msebor@redhat.com>
36117 * doc/extend.texi (attribute constructor): Document present limitation.
36119 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36122 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36123 __POWER9_VECTOR__ #ifdef control, change template definition to
36124 use Power9-specific built-in function.
36125 (vec_any_eq): Likewise.
36126 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36127 to control outcomes from this test.
36128 (vector_ae_<mode>p): For VEC_F modes, likewise.
36130 2017-03-20 Ian Lance Taylor <iant@google.com>
36132 * config/i386/i386.c (ix86_function_regparm): Save an extra
36133 register for -fsplit-stack with DECL_STATIC_CHAIN.
36135 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36138 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36139 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36141 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36143 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36145 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36148 2017-03-20 Marek Polacek <polacek@redhat.com>
36151 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36153 2017-03-20 Richard Biener <rguenther@suse.de>
36155 PR tree-optimization/80113
36156 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36157 allocate extra SSA name for PHI def.
36158 (add_close_phis_to_outer_loops): Likewise.
36159 (add_close_phis_to_merge_points): Likewise.
36160 (copy_loop_close_phi_args): Likewise.
36161 (copy_cond_phi_nodes): Likewise.
36163 2017-03-20 Martin Liska <mliska@suse.cz>
36165 PR middle-end/79753
36166 * tree-chkp.c (chkp_build_returned_bound): Do not build
36167 returned bounds for a LHS that's not a BOUNDED_P type.
36169 2017-03-20 Martin Liska <mliska@suse.cz>
36173 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36174 COMPLEX_CST and VECTOR_CST.
36176 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36179 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36180 target operand. A new splitter adds the clobber statement in case
36181 the target operand is dead anyway.
36183 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36185 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36186 to age-old versions of binutils and glibc.
36188 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36190 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36192 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36194 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36196 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36198 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36199 requirement for binutils 2.13.
36201 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36203 * combine.c (try_combine): Delete redundant i1 test. Call
36204 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36206 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36208 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36209 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36211 <riscv64-*-elf>: Re-arrange section
36212 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36213 <riscv32-*-linux>: Likewise.
36214 <riscv64-*-elf>: Likewise
36215 <riscv64-*-linux>: Likewise.
36217 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36220 * aarch64.opt(verbose-cost-dump): Fix typo.
36222 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36225 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36226 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36228 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36230 * reload.c (find_reloads): When reloading a nonoffsettable address,
36231 use RELOAD_OTHER for it and its address reloads.
36233 PR rtl-optimization/79910
36234 * combine.c (record_used_regs): New static function.
36235 (try_combine): Handle situations where there is an additional
36236 instruction between I2 and I3 which needs to have a LOG_LINK
36239 2017-03-17 Jeff Law <law@redhat.com>
36241 PR tree-optimization/71437
36242 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36243 conditional in the hash table first.
36244 (vrp_dom_walker::before_dom_children): Extract condition from
36245 ASSERT_EXPR. Record condition, its inverion and any implied
36246 conditions as well.
36248 2017-03-17 Marek Polacek <polacek@redhat.com>
36249 Markus Trippelsdorf <markus@trippelsdorf.de>
36251 PR tree-optimization/80079
36252 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36255 2017-03-17 Richard Biener <rguenther@suse.de>
36257 PR middle-end/80075
36258 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36259 Properly verify the LHS before the RHS possibly claims to be
36261 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36264 2017-03-17 Martin Jambor <mjambor@suse.cz>
36266 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36267 (List of -O2 options): Likewise.
36268 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36271 2017-03-17 Tom de Vries <tom@codesourcery.com>
36273 * gcov-dump.c (print_usage): Print bug_report_url.
36275 2017-03-17 Richard Biener <rguenther@suse.de>
36277 PR middle-end/80050
36278 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36279 (parser::peek): Likewise.
36281 2017-03-17 Richard Biener <rguenther@suse.de>
36283 PR tree-optimization/80048
36284 * sese.c (free_sese_info): Properly release rename_map and
36285 copied_bb_map elements.
36287 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36289 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36290 Add linked-list forward and backlinks. Insert on
36291 construction, remove on destruction.
36292 (class pass_store_merging): Add m_stores_head field.
36293 (pass_store_merging::terminate_and_process_all_chains):
36294 Iterate over m_stores_head list.
36295 (pass_store_merging::terminate_all_aliasing_chains):
36297 (pass_store_merging::execute): Check for debug stmts first.
36298 Push new chains onto the m_stores_head stack.
36300 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36303 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36304 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36305 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36307 2017-03-16 Jeff Law <law@redhat.com>
36309 PR tree-optimization/71437
36310 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36311 member function. Implementation moved into after_dom_children
36312 member function and into the threader's thread_outgoing_edges
36314 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36315 some code into new thread_outgoing_edges.
36316 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36317 definition. Simplify marker handling (do it here). Assume we always
36318 have the available expression and the const/copies tables.
36319 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36321 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36322 * tree-vrp.c (equiv_stack): No longer file scoped.
36323 (vrp_dom_walker): New class.
36324 (vrp_dom_walker::before_dom_children): New member function.
36325 (vrp_dom_walker::after_dom_children): Likewise.
36326 (identify_jump_threads): Setup domwalker. Use it rather than
36327 walking edges in a random order by hand. Simplify setup/finalization.
36328 (finalize_jump_threads): Remove.
36329 (vrp_finalize): Do not call identify_jump_threads here.
36330 (execute_vrp): Do it here instead and call thread_through_all_blocks
36333 PR tree-optimization/71437
36334 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36336 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36338 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36339 (dom_opt_dom_walker::thread_across_edge): Remove
36340 handle_dominating_asserts argument. All callers changed.
36341 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36342 changes. Remove calls to lhs_of_dominating_assert. Other
36343 uses of handle_dominating_asserts turn into unconditional code
36344 (simplify_control_stmt_condition_1): Likewise.
36345 (simplify_control_stmt_condition): Likewise.
36346 (thread_through_normal_block, thread_across_edge): Likewise.
36347 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36348 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36349 object if it is not an SSA_NAME.
36350 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36351 before calling into the VRP specific simplifiers.
36352 (identify_jump_threads): Remove handle_dominating_asserts
36355 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36358 * tree-diagnostic.c (default_tree_printer): No longer static.
36359 * tree-diagnostic.h (default_tree_printer): New prototype.
36361 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36363 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36364 Change ins into fmov.
36366 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36368 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36369 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36370 Use h_con constraint for operand 1.
36371 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36372 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36374 2017-03-15 Jeff Law <law@redhat.com>
36376 PR tree-optimization/71437
36377 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36378 (record_temporary_equivalences): Use it.
36380 PR tree-optimization/71437
36381 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36382 tree-ssa-scopedtables.
36383 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36384 (record_conditions, record_cond, vuse_eq): Likewise.
36385 (record_edge_info): Adjust to API tweak of record_conditions.
36386 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36387 (record_temporary_equivalences, optimize_stmt): Likewise.
36388 (eliminate_redundant_computations): Likewise.
36389 (record_equivalences_from_stmt): Likewise.
36390 * tree-ssa-scopedtables.c: Include options.h and params.h.
36391 (vuse_eq): New function, moved from tree-ssa-dom.c
36392 (build_and_record_new_cond): Likewise.
36393 (record_conditions): Likewise. Accept vector of conditions rather
36394 than edge_equivalence structure for first argument.
36395 for the first argument.
36396 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36397 from tree-ssa-dom.c.
36398 (avail_exprs_stack::record_cond): Likewise.
36399 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36400 from tree-ssa-dom.c.
36401 (avail_exprs_stack): Add new member functions lookup_avail_expr
36403 (record_conditions): Declare.
36405 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36408 * lra-constraints.c (process_alt_operands): Increase reject for
36409 reloading an input/output operand.
36411 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36414 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36415 insns to convert from signed/unsigned char/short to IEEE 128-bit
36417 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36419 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36422 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36423 subreg of inner mode for values already in registers.
36425 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36427 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36428 iteration reg is used after the loop.
36430 2017-03-14 Martin Sebor <msebor@redhat.com>
36432 PR tree-optimization/79800
36433 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36434 precision in negative-positive range.
36435 (format_floating): Call non-const overload with adjusted precision.
36437 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36440 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36443 2017-03-14 Martin Sebor <msebor@redhat.com>
36445 PR middle-end/80020
36446 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36447 * builtins.def (aligned_alloc): Use it.
36450 * Makefile.in (GTFILES): Add calls.c.
36451 * calls.c: Include "gt-calls.h".
36453 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36455 PR rtl-optimization/79728
36456 * regs.h (struct target_regs): New field
36457 x_contains_allocatable_regs_of_mode.
36458 (contains_allocatable_regs_of_mode): New macro.
36459 * reginfo.c (init_reg_sets_1): Initialize it, and change
36460 contains_reg_of_mode so it includes global regs as well.
36461 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36462 rather than contains_regs_of_mode.
36464 2017-03-14 Martin Liska <mliska@suse.cz>
36466 * doc/invoke.texi: Document options that can't be combined with
36467 -fcheck-pointer-bounds.
36469 2017-03-14 Martin Liska <mliska@suse.cz>
36471 PR middle-end/79831
36472 * doc/invoke.texi (-Wchkp): Document the option.
36474 2017-03-14 Martin Liska <mliska@suse.cz>
36476 * Makefile.in: Install gcov-dump.
36478 2017-03-14 Martin Liska <mliska@suse.cz>
36480 * multiple_target.c (expand_target_clones): Bail out for
36481 an invalid attribute.
36483 2017-03-14 Richard Biener <rguenther@suse.de>
36485 * alias.c (struct alias_set_entry): Pack properly.
36486 * cfgloop.h (struct loop): Likewise.
36487 * cse.c (struct set): Likewise.
36488 * ipa-utils.c (struct searchc_env): Likewise.
36489 * loop-invariant.c (struct invariant): Likewise.
36490 * lra-remat.c (struct cand): Likewise.
36491 * recog.c (struct change_t): Likewise.
36492 * rtl.h (struct address_info): Likewise.
36493 * symbol-summary.h (function_summary): Likewise.
36494 * tree-loop-distribution.c (struct partition): Likewise.
36495 * tree-object-size.c (struct object_size_info): Likewise.
36496 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36497 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36498 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36499 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36500 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36501 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36502 (struct _stmt_vec_info): Likewise.
36504 2017-03-14 Martin Liska <mliska@suse.cz>
36507 * multiple_target.c (create_dispatcher_calls): Check that
36508 a target can create a function dispatcher.
36510 2017-03-14 Martin Liska <mliska@suse.cz>
36513 * multiple_target.c (expand_target_clones): Drop local.local
36514 flag for default implementation.
36516 2017-03-14 Richard Biener <rguenther@suse.de>
36518 PR tree-optimization/80030
36519 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36521 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36523 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36524 gcc_fallthrough() instead of __attribute__((fallthrough));
36526 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36528 * doc/gcc.texi: Remove "up" link to (DIR).
36529 * doc/gccint.texi: Ditto.
36531 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36533 * doc/install.texi (Specific) <avr>: Remove reference to
36536 2017-03-13 Jeff Law <law@redhat.com>
36538 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36539 attribute rather than comments.
36541 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36542 match_scratch operand is highest.
36544 2017-03-13 Martin Liska <mliska@suse.cz>
36546 PR middle-end/78339
36547 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36548 is a CHKP clone, use original declaration.
36550 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36552 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36553 (arc_conditional_register_usage): Use a different allocation order
36554 when optimizing for size.
36555 * common/config/arc/arc-common.c (arc_option_optimization_table):
36556 Section anchors default on when optimizing for size.
36558 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36560 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36562 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36564 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36565 * config/arc/arc.md (cpu_facility): Add cd variant.
36566 (*movqi_insn): Add code density variant.
36567 (*movhi_insn): Likewise.
36568 (*movqi_insn): Likewise.
36569 (*addsi3_mixed): Likewise.
36570 (subsi3_insn): Likewise.
36572 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36574 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36576 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36578 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36579 expressions with MINUS and UNARY ops.
36581 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36584 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36586 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36587 between vec_select and vector argument.
36588 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36589 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36590 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36591 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36592 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36593 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36595 2017-03-13 Richard Biener <rguenther@suse.de>
36598 * params.def (vect-max-peeling-for-alignment): Fix typo.
36600 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36602 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36603 issue that only occurred with binutils below 2.18.
36605 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36607 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36608 refer to binutils 2.11/2.12 minimum.
36610 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36612 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36613 ftp.kernel.org and simplify binutils requirement.
36615 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36617 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36619 (Optimize Options): Ditto. Also remove redundancy.
36621 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36623 PR translation/79848
36624 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36626 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36627 to G_ to avoid double translation.
36629 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36631 PR translation/79923
36632 * auto-profile.c (get_combined_location): Convert leading
36633 character of diagnostics to lower case and remove trailing period.
36634 (read_profile): Likewise for various diagnostics.
36635 * config/arm/arm.c (arm_option_override): Remove trailing period
36636 from various diagnostics.
36637 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36638 (msp430_expand_delay_cycles): Likewise.
36640 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36643 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36644 full command-line argument, rather than just "str".
36645 (aarch64_validate_march): Likewise.
36646 (aarch64_validate_mtune): Likewise.
36648 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36650 PR rtl-optimization/78911
36651 * lra-assigns.c (must_not_spill_p): New function.
36652 (spill_for): Use it.
36654 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36656 PR tree-optimization/79981
36657 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36658 ATOMIC_COMPARE_EXCHANGE ifn result.
36659 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36660 IFN_ATOMIC_COMPARE_EXCHANGE.
36662 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36665 * opts.c (parse_sanitizer_options): Add missing question mark to
36666 "did you mean" message.
36668 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36670 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36672 (VMULEUH_UNS): Likewise.
36673 (VMULOUB_UNS): Likewise.
36674 (VMULOUH_UNS): Likewise.
36675 * config/rs6000/rs6000.c (builtin_function_type): Remove
36676 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36678 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36681 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36682 x with result of extra_parsing_for_operand_code_0.
36683 (function_reader::extra_parsing_for_operand_code_0): Convert
36684 return type from void to rtx, returning x. When reading
36685 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36686 larger size containing struct block_symbol.
36688 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36691 -mfloat128-hardware without -m64.
36693 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36696 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36697 entries to the case statement that marks unsigned arguments to
36698 overloaded functions.
36700 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36703 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36705 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36708 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36709 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36711 2017-03-10 Martin Liska <mliska@suse.cz>
36715 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36716 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36719 2017-03-10 Olivier Hainque <hainque@adacore.com>
36721 * tree-switch-conversion (array_value_type): Start by resetting
36722 candidate type to it's main variant.
36724 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36726 PR rtl-optimization/79909
36727 * combine.c (try_combine): Use simplify_replace_rtx on individual
36728 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36729 of the whole CALL_INSN_FUNCTION_USAGE.
36731 PR tree-optimization/79972
36732 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36733 get_range_info on SSA_NAMEs. Formatting fixes.
36735 2017-03-10 Richard Biener <rguenther@suse.de>
36736 Jakub Jelinek <jakub@redhat.com>
36738 PR tree-optimization/77975
36739 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36740 edge to be constant.
36741 (get_val_for): For constant x return it. Formatting fix.
36742 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36743 would use the same bases as the current one.
36745 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36747 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36748 instead of vec_select for V1TImode.
36749 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36751 (VSX_LE_128): Add V1TI to this mode iterator.
36752 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36753 (*vsx_le_perm_store_<mode>): Likewise.
36754 (pre-reload splitter for VSX stores): Likewise.
36755 (post-reload splitter for VSX stores): Likewise.
36756 (*vsx_xxpermdi2_le_<mode>): Likewise.
36757 (*vsx_lxvd2x2_le_<mode>): Likewise.
36758 (*vsx_stxvd2x2_le_<mode>): Likewise.
36760 2017-03-09 Michael Eager <eager@eagercon.com>
36762 Correct failures with --enable-checking=yes,rtl.
36764 * config/microblaze/microblaze.c (microblaze_expand_shift):
36765 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36766 test for const0_rtx.
36767 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36768 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36770 2017-03-09 Richard Biener <rguenther@suse.de>
36772 PR tree-optimization/79977
36773 * graphite-scop-detection.c (scop_detection::merge_sese):
36774 Handle the case of extra exits to blocks dominating the entry.
36776 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36778 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36781 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36783 PR rtl-optimization/79949
36784 * lra-constraints.c (process_alt_operands): Check memory when
36785 trying to predict a cycle. Print about the overall increase.
36787 2017-03-09 Richard Biener <rguenther@suse.de>
36789 PR middle-end/79971
36790 * gimple-expr.c (useless_type_conversion_p): Preserve
36791 TYPE_SATURATING for fixed-point types.
36793 2017-03-09 Richard Biener <rguenther@suse.de>
36796 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36797 alignment of BLKmode params.
36799 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36802 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36803 (VALL_NO_V2Q): Likewise.
36805 * config/aarch64/aarch64-simd.md
36806 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36808 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36809 VALL_NO_V2Q mode iterator.
36810 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36812 2017-03-09 Martin Liska <mliska@suse.cz>
36814 PR tree-optimization/79631
36815 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36816 tree_int_cst_sign_bit just for INTEGER constants.
36818 2017-03-09 Martin Liska <mliska@suse.cz>
36822 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36825 2017-03-09 Marek Polacek <polacek@redhat.com>
36828 * tree.c (inchash::add_expr): Handle TREE_VEC.
36830 2017-03-09 Martin Liska <mliska@suse.cz>
36833 (chkp_narrow_size_and_offset): New function.
36834 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36835 (void chkp_parse_bit_field_ref): New function.
36836 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36837 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36839 2017-03-09 Martin Liska <mliska@suse.cz>
36842 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36843 (chkp_find_bounds_1): Remove gcc_unreachable.
36845 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36848 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36849 BUILT_IN_SYNC*, determine the access type from the size suffix and
36850 always build a MEM_REF with that type. Handle forgotten
36851 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36854 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36855 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36856 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36857 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36858 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36859 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36860 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36861 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36862 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36863 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36864 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36865 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36866 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36867 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36868 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36869 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36870 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36871 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36872 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36873 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36874 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36875 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36876 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36877 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36878 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36879 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36880 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36881 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36882 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36883 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36884 definitions outside of __OPTIMIZE__ guarded section.
36887 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36888 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36889 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36890 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36893 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36895 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36896 ("vfenez<mode>"): Add missing constraints.
36898 2017-03-08 Martin Sebor <msebor@redhat.com>
36901 * config/nds32/nds32.c (nds32_option_override):
36902 Fix misspelled diagnostic.
36904 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36907 * gimplify.c (gimplify_omp_for): Replace index var in outer
36908 taskloop statement with an artificial variable and add
36909 OMP_CLAUSE_PRIVATE clause for it.
36911 2017-03-08 Richard Biener <rguenther@suse.de>
36913 PR tree-optimization/79955
36914 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36915 for accesses that are completely outside of the variable.
36917 2017-03-08 Andrew Haley <aph@redhat.com>
36919 PR tree-optimization/79943
36920 * tree-ssa-loop-split.c (compute_new_first_bound): When
36921 calculating the new upper bound, (END-BEG) should be added, not
36924 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36926 * config/avr/avr.md (setmemhi): Make sure match_dup
36927 operand number comes before match_scratch.
36929 2017-03-08 Richard Biener <rguenther@suse.de>
36931 PR tree-optimization/79920
36932 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36933 with ncopies == 1 to ...
36934 (vect_transform_slp_perm_load): ... here. Properly compute
36935 all element loads by iterating VF times over the group. Do
36936 not handle ncopies (computed in a broken way) in
36937 vect_create_mask_and_perm.
36939 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36942 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36943 is a uniform vector, use uniform_vector_p return value instead of
36944 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36946 2017-03-07 Marek Polacek <polacek@redhat.com>
36948 PR middle-end/79809
36949 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36950 (alloca_call_type): Likewise.
36952 2017-03-07 Martin Liska <mliska@suse.cz>
36954 * gcov.c (process_args): Put comment to correct location.
36956 2017-03-07 Martin Liska <mliska@suse.cz>
36958 PR middle-end/68270
36959 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36960 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36961 (chkp_narrow_bounds_for_field): Likewise.
36962 (chkp_parse_array_and_component_ref): Pass one more argument to
36965 2017-03-07 Richard Biener <rguenther@suse.de>
36967 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36970 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36972 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36973 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36975 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36978 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36979 to end of description.
36980 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36982 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36984 PR rtl-optimization/79901
36985 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36987 (*avx512f_<code><mode>3<mask_name>): ... this.
36988 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36989 iterator instead of VI8_AVX2_AVX512BW.
36991 PR rtl-optimization/79901
36992 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36993 min/max expander, expand it using expand_vec_cond_expr.
36996 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36999 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37002 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37003 to void * for PCH reasons.
37004 * dwarf2out.c (output_loc_operands, output_die): Cast
37005 v.val_vec.array to unsigned char *.
37007 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37010 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37013 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37015 PR rtl-optimization/79571
37016 * lra-constraints.c (process_alt_operands): Calculate static
37017 reject and subtract it from overall when only addresses will be
37020 2017-03-06 Julia Koval <julia.koval@intel.com>
37023 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37024 incoming stack boundary to 128 for 64-bit targets.
37026 2017-03-06 Richard Biener <rguenther@suse.de>
37028 PR tree-optimization/79894
37029 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37030 to NULL after folding it.
37032 2017-03-06 Richard Biener <rguenther@suse.de>
37034 PR tree-optimization/79824
37035 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37036 check disabling peeling for gaps.
37038 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37040 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37041 attributes): Document gettimeofday.
37043 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37045 * config/s390/s390.c (s390_option_override_internal): Set
37046 PARAM_MIN_VECT_LOOP_BOUND
37048 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37050 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37051 * config/s390/s390.md: Likewise.
37053 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37056 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37057 (<avx2_avx512>_perm<mode>): Rename to ...
37058 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37060 (<avx512>_perm<mode>_mask): Rename to ...
37061 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37063 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37064 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37065 instead of VI8F_256_512.
37066 (avx512f_perm<mode>): New define_expand.
37067 (avx512f_perm<mode>_mask): Likewise.
37068 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37069 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37071 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37073 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37074 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37076 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37078 2017-03-06 Martin Liska <mliska@suse.cz>
37081 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37082 when having a SSA NAME w/o VAR_DECL assigned to it.
37084 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37086 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37087 msa_dpsub_<su>_d): Fix MODE for vec_select.
37089 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37091 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37093 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37095 2017-03-06 Richard Biener <rguenther@suse.de>
37097 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37098 * plugin.c (register_plugin_info): Likewise.
37099 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37101 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37103 * config/i386/sse.md (sse_storehps, sse_storelps,
37104 avx_<castmode><avxsizesuffix>_<castmode>,
37105 avx512f_<castmode><avxsizesuffix>_<castmode>,
37106 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37107 in condition that at least one operand is not a MEM.
37109 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37111 PR middle-end/79805
37112 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37113 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37115 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37116 gimple_call_nothrow_p flag based on whether original builtin can throw.
37117 If it can, emit following stmts on the fallthrough edge.
37118 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37119 don't create new bb if inserting just debug stmts on the edge, try to
37120 insert them on the fallthru bb or just reset debug stmts.
37122 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37125 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37126 restore recog_data (including the operand rtxes inside it) around
37127 the call to get_insn_template.
37129 2017-03-03 Martin Sebor <msebor@redhat.com>
37131 PR tree-optimization/79699
37132 * context.c (context::~context): Free MPFR caches to avoid
37133 a memory leak on program exit.
37135 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37137 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37138 Use wide_int::ulow () instead of .elt (0).
37140 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37142 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37143 (*pushxf): Limit oF constraint to 32bit targets and add oC
37144 constraint for 64bit targets.
37145 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37146 (*pushdf): Change rmF constraint to rmC.
37148 2017-03-03 Martin Liska <mliska@suse.cz>
37150 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37151 Remove unused variable.
37153 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37156 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37157 is a memory operand, increase num_memory.
37158 (ix86_expand_args_builtin): Likewise.
37160 2017-03-03 Jan Hubicka <jh@suse.cz>
37163 * ipa-devirt.c (maybe_record_node): Properly handle
37164 __cxa_pure_virtual visibility.
37166 2017-03-03 Martin Liska <mliska@suse.cz>
37168 PR tree-optimization/79803
37169 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37171 (pass_loop_prefetch::execute): Disabled optimization if an
37172 assumption about L1 cache size is not met.
37174 2017-03-03 Martin Liska <mliska@suse.cz>
37176 PR rtl-optimization/79574
37177 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37178 (hash_scan_set): Likewise.
37179 (dump_hash_table): Likewise.
37180 (hoist_code): Likewise.
37182 2017-03-03 Richard Biener <rguenther@suse.de>
37184 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37185 in place of uhigh ().
37186 (fixed_convert_from_real): Likewise.
37188 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37191 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37193 2017-03-03 Richard Biener <rguenther@suse.de>
37195 PR middle-end/79818
37196 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37197 TYPE_OVERFLOW_UNDEFINED check.
37199 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37201 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37203 (vector_ae_<mode>_p): Likewise.
37204 (vector_nez_<mode>_p): Likewise.
37205 (vector_ne_v2di_p): Likewise.
37206 (vector_ae_v2di_p): Likewise.
37207 (vector_ne_<mode>_p): Likewise.
37208 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37210 (vsx_tsqrt<mode>2_fe): Likewise.
37212 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37215 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37217 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37219 PR rtl-optimization/79780
37220 * cprop.c (one_cprop_pass): When second and further conditional trap
37221 in a single basic block is turned into an unconditional trap, turn it
37222 into a deleted note to avoid RTL verification failures.
37224 2017-03-02 Richard Biener <rguenther@suse.de>
37226 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37228 2017-03-02 Richard Biener <rguenther@suse.de>
37230 PR tree-optimization/79345
37232 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37233 param and abort the walk, returning -1 if it is hit.
37234 (walk_aliased_vdefs): Take a limit param and pass it on.
37235 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37236 defaulting to 0 and return a signed int.
37237 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37238 (check_defs): New helper.
37239 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37240 about uninitialized memory.
37241 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37242 bogus uninitialized warning.
37243 (fixed_convert_from_real): Likewise.
37245 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37247 PR tree-optimization/66768
37248 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37249 iv_use if base object can't be determined.
37251 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37253 PR tree-optimization/79345
37254 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37255 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37256 (get_pattern_stats): Initialize it.
37257 * genemit.c (gen_expand): Verify match_scratch numbers come after
37258 match_operand/match_dup numbers.
37259 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37260 match_scratch numbers.
37261 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37263 * config/s390/s390.md (trunctdsd2): Likewise.
37265 2017-03-02 Richard Biener <rguenther@suse.de>
37267 * wide-int.h (wide_int_storage::operator=): Implement in terms
37270 2017-03-02 Richard Biener <rguenther@suse.de>
37272 PR tree-optimization/79777
37273 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37274 the to insert expression to sth existing.
37276 2017-03-01 Martin Sebor <msebor@redhat.com>
37278 PR middle-end/79692
37279 * gimple-ssa-sprintf.c
37280 (directive::known_width_and_precision): New function.
37281 (format_integer): Use it.
37282 (get_mpfr_format_length): Consider the full range of precision
37283 when computing %g output with the # flag. Set the likely byte
37284 count to 3 rather than 1 when precision is indeterminate.
37285 (format_floating): Correct the lower bound of precision.
37287 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37289 * doc/invoke.texi: Document default code model for 64-bit Linux.
37291 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37294 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37295 udiv rather than div since input pattern is unsigned.
37297 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37299 * config/i386/i386.c (print_reg): Warn for values of
37300 unsupported size in integer register.
37302 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37305 * config/rs6000/predicates.md (current_file_function_operand): Do
37306 not allow self calls to be local if the function is replaceable.
37308 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37311 * config/rs6000/altivec.h (vec_ctz and others): Change the
37312 preprocessor macro that controls conditional compilation from
37313 _ARCH_PWR9 to __POWER9_VECTOR__.
37314 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37315 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37316 control (instead of _ARCH_PWR9 control) so that template
37317 definition uses power9-specific function.
37318 (vec_any_eq): Likewise.
37319 (vec_all_ne): Change macro definition to use a power9-specific
37320 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37321 _ARCH_PWR9 control).
37322 (vec_any_eq) Likewise.
37323 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37324 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37325 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37326 support for xvcmpnedp instruction.
37327 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37328 macro expansion so that Power9 implementation of vec_all_ne does
37329 not use the AltiVec predicate framework.
37330 (VCMPNEH_P): Likewise.
37331 (VCMPNEW_P): Likewise.
37332 (VCMPNED_P): Likewise.
37333 (VCMPNEFP_P): Likewise.
37334 (VCMPNEDP_P): Likewise.
37335 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37336 implementation of vec_any_eq to not use AltiVec predicate
37338 (VCMPAEH_P): Likewise.
37339 (VCMPAEW_P): Likewise.
37340 (VCMPAED_P): Likewise.
37341 (VCMPAEFP_P): Likewise.
37342 (VCMPAEDP_P): Likewise.
37343 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37344 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37345 not use the AltiVec predicate framework.
37346 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37347 of vec_any_eq to not use AltiVec predicate framework.
37348 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37349 support for predefined __POWER9_VECTOR__ macro to indicate that
37350 Power9 instruction selection is enabled.
37351 (altivec_overloaded_builtins): Remove extraneous
37352 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37353 function argument types RS6000_BTI_bool_V16QI and
37354 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37355 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37356 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37357 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37358 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37359 Power9 for implementations of vec_cmpne. Change the signature for
37360 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37361 (representing vec_all_ne) to remove the previously described first
37362 argument of type RS6000_BTI_INTSI, as this was an artifact of
37363 reliance on the AltiVec predicate framework, which is no longer
37364 used in the implementation of these functions. Add
37365 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37366 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37367 since, unlike the AltiVec predicate framework implementation, we
37368 do not share function descriptors between vec_alle and vec_anyeq.
37369 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37370 set of modes that receive special treatment even when
37371 TARGET_P9_VECTOR is true. The special treatment emits code that
37372 does not depend on Power9 instructions.
37373 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37374 define_expand to not rely on AltiVec predicate framework.
37375 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37377 (vector_ne_v2di_p): Change this define_expand to not rely on
37378 AltiVec predicate framework.
37379 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37381 (vector_ne_<mode>_p): Change this define_expand to not rely on
37382 AltiVec predicate framework.
37383 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37385 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37386 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37387 define_insn pattern.
37388 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37389 define_insn pattern because the xvcmpne<VSs>. instruction is not
37391 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37392 instruction is not supported.
37394 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37396 * config/nvptx/nvptx.c: Include intl.h.
37398 2017-03-01 Martin Jambor <mjambor@suse.cz>
37401 * ipa-prop.h (ipa_bits): Removed field known.
37402 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37403 to pointers. Adjusted their comments to warn about their sharing.
37404 (ipcp_transformation_summary): Change bits to a vector of pointers.
37405 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37406 (ipa_get_ipa_bits_for_value): Declare.
37407 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37408 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37409 (ipa_bits_hash_table): Likewise.
37410 (ipa_vr_ggc_hash_traits): Likewise.
37411 (ipa_vr_hash_table): Likewise.
37412 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37413 being pointers and vr_known being removed.
37414 (ipa_set_jf_unknown): Likewise.
37415 (ipa_get_ipa_bits_for_value): New function.
37416 (ipa_set_jfunc_bits): Likewise.
37417 (ipa_get_value_range): New overloaded functions.
37418 (ipa_set_jfunc_vr): Likewise.
37419 (ipa_compute_jump_functions_for_edge): Use the above functions to
37420 construct bits and vr parts of jump functions.
37421 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37422 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37424 (ipcp_grow_transformations_if_necessary): Also allocate
37425 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37427 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37428 them. Fix too long lines.
37429 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37430 vr_known being removed.
37431 (ipa_read_jump_function): Use new setter functions to construct bits
37432 and vr parts of jump functions or set them to NULL.
37433 (write_ipcp_transformation_info): Adjust for bits being pointers.
37434 (read_ipcp_transformation_info): Likewise.
37435 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37437 Include gt-ipa-prop.h.
37438 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37440 (ipcp_store_bits_results): Likewise.
37441 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37442 Do not write to existing jump functions but use a temporary instead.
37444 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37447 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37448 attempt to use its first operand as BIT_FIELD_REF base.
37450 2017-03-01 Richard Biener <rguenther@suse.de>
37452 PR middle-end/79721
37453 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37454 interpolating formula in wrapping arithmetic.
37455 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37457 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37459 PR tree-optimization/79734
37460 * tree-vect-generic.c (expand_vector_condition): Optimize
37461 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37462 Handle VEC_COND_EXPR where comparison has different inner width from
37463 type's inner width.
37465 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37467 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37468 markup, and similar issues. Remove @opindex entries for things
37469 that aren't options. Add missing -mmpy-option entries.
37471 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37473 PR tree-optimization/79737
37474 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37475 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37476 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37477 instead of byte_size. Formatting fix.
37478 (shift_bytes_in_array_right): Formatting fix.
37480 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37483 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37484 condition on optimize for the leaf function test.
37486 2017-02-28 Martin Liska <mliska@suse.cz>
37489 * read-rtl-function.c (function_reader::handle_unknown_directive):
37490 Bail out when one uses -flto.
37492 2017-02-28 Martin Liska <mliska@suse.cz>
37494 * common.opt: Replace space with tabular for options of <number>
37496 * config/i386/i386.opt: Show <number> value for
37497 -mlarge-data-threshold.
37498 * opts.c (print_filtered_help): Do not display number in hexadecimal
37501 2017-02-28 Martin Liska <mliska@suse.cz>
37503 * common.opt: Fix --help=option -Q for options which are of
37506 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37508 * config/i386/i386.c (print_reg): Error out for values
37509 of 8-bit size in invalid integer register.
37511 2017-02-28 Martin Sebor <msebor@redhat.com>
37513 PR tree-optimization/79691
37514 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37516 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37519 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37520 gcc_unreachable with output_operand_lossage.
37522 2017-02-28 Richard Biener <rguenther@suse.de>
37524 PR tree-optimization/79740
37525 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37527 (visit_nary_op): Insert the nary into the hashtable if we
37528 pattern-matched sth.
37529 * tree-ssa-pre.c (eliminate_insert): Robustify.
37531 2017-02-28 Richard Biener <rguenther@suse.de>
37533 PR middle-end/79731
37534 * fold-const.c (decode_field_reference): Reject out-of-bound
37537 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37539 * config/i386/i386.c: Include intl.h.
37540 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37541 instead of just cond ? "..." : "...".
37542 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37543 * coverage.c (read_counts_file): Likewise.
37544 * omp-offload.c: Include intl.h.
37545 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37546 of just cond ? "..." : "...".
37547 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37548 of just cond ? "..." : "...".
37550 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37553 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37555 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37556 'tune for' CPU name.
37557 * config/arm/arm-cpu-data.h: Regenerated.
37559 2017-02-28 Richard Biener <rguenther@suse.de>
37561 PR tree-optimization/79732
37562 * tree-inline.c (expand_call_inline): Do not shadow var.
37564 2017-02-28 Richard Biener <rguenther@suse.de>
37566 PR tree-optimization/79723
37567 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37568 address-space properly.
37570 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37572 * doc/optinfo.texi (Optimization groups): Fix option used for
37574 * doc/invoke.texi (-fopt-info): Document "omp".
37575 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37576 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37577 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37578 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37579 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37581 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37583 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37584 OPTGROUP_OMP with "omp".
37586 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37589 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37590 for arithmetic shift of unsigned V2DI.
37592 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37594 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37595 arc/linux.h headers.
37596 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37597 (LINK_SPEC): Likewise.
37598 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37599 (EXTRA_SPECS): Likewise.
37600 (STARTFILE_SPEC): Likewise.
37601 (ENDFILE_SPEC): Likewise.
37602 (LIB_SPEC): Likewise.
37603 (TARGET_SDATA_DEFAULT): Likewise.
37604 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37605 (MULTILIB_DEFAULTS): Likewise.
37606 (DWARF2_UNWIND_INFO): Likewise.
37607 * config/arc/big.h: New file.
37608 * config/arc/elf.h: Likewise.
37609 * config/arc/linux.h: Likewise.
37610 * config/arc/t-uClibc: Remove.
37612 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37614 PR tree-optimization/77536
37615 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37616 (tree_transform_and_unroll_loop): Use above function to compute the
37617 estimated niter of unrolled loop and use it when scaling profile.
37618 Also use count info rather than frequency if it's non-zero.
37619 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37620 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37621 (vect_transform_loop): Call above function.
37623 2017-02-27 Richard Biener <rguenther@suse.de>
37625 PR tree-optimization/45397
37626 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37627 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37628 (visit_nary_op): Add pattern matching for CSEing sign-changed
37629 or truncated operations with wider ones.
37631 2017-02-27 Richard Biener <rguenther@suse.de>
37633 PR tree-optimization/79690
37634 * tree-vect-stmts.c (vectorizable_store): Use vector type
37635 built from the DR with address-space.
37637 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37639 * doc/invoke.texi (Optimize Options): Refine the description
37640 of asan-use-after-return.
37642 2017-02-25 Alan Modra <amodra@gmail.com>
37644 PR rtl-optimization/79584
37645 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37646 base, not ad->base_term, the reg within base. Remove assertion
37647 that ad->base == ad->base_term. Replace gen_int_mode using
37648 bogus mode with const0_rtx.
37650 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37652 PR middle-end/79396
37653 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37654 FMA_EXPR like tcc_binary or tcc_unary.
37656 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37659 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37661 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37662 (output_loc_operands): Handle DW_OP_call_ref and
37663 DW_OP_GNU_variable_value.
37664 (struct variable_value_struct): New type.
37665 (struct variable_value_hasher): Likewise.
37666 (variable_value_hash): New variable.
37667 (string_types): Remove.
37668 (copy_loc_descr): New function.
37669 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37670 (prepend_loc_descr_to_each): New function.
37671 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37672 instead of add_loc_descr_to_each if the first argument is single
37673 location list and the second has multiple.
37674 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37675 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37676 when looking for variable value which doesn't have other location info.
37677 (loc_list_from_tree): Formatting fix.
37678 (gen_array_type_die): Simplify DW_AT_string_length handling.
37679 (adjust_string_types): Remove.
37680 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37681 string_types. Call resolve_variable_values.
37682 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37683 (resolve_addr_in_expr): Likewise. Add A argument.
37684 (copy_deref_exprloc): Remove deref argument. Adjust for the
37685 original expression being DW_OP_GNU_variable_value with optionally
37686 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37687 optionally after it.
37688 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37689 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37690 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37691 (variable_value_hasher::hash, variable_value_hasher::equal): New
37693 (resolve_variable_value_in_expr, resolve_variable_value,
37694 resolve_variable_values, note_variable_value_in_expr,
37695 note_variable_value): New functions.
37696 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37699 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37702 * opts.h (handle_generated_option): Add GENERATED_P argument.
37703 * opts-common.c (handle_option): Adjust function comment.
37704 (handle_generated_option): Add GENERATED_P argument, pass it to
37706 (control_warning_option): Pass false to handle_generated_option
37708 * opts.c (maybe_default_option): Pass true to handle_generated_option
37710 * optc-gen.awk: Likewise.
37712 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37714 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37715 a REG, look at the REG it is a SUBREG of.
37716 (splitter for cmpeqsi_t): Ditto.
37718 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37720 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37721 the special USEs with the pattern of the insn, not the insn itself.
37723 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37726 * doc/invoke.texi: Document -mload-store-pairs.
37728 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37729 Sandra Loosemore <sandra@codesourcery.com>
37731 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37732 argument isn't a CONST_INT.
37733 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37734 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37735 (nios2_validate_compare): Bypass alternate compare logic if *op2
37736 is not a CONST_INT.
37737 (ldstwm_operation_p): Return false if first_base is not a REG or
37738 if first_offset is not a CONST_INT.
37740 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37742 * config/cris/cris.md: Use correct operand in a define_peephole2.
37744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37746 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37748 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37750 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37751 this_insn if it is an INSN or JUMP_INSN.
37752 (force_offsettable): Look at base, not at addr.
37753 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37754 on things that aren't necessarily CONST_INTs.
37756 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37758 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37759 -mfpmath=sse is the default also for x86-32 targets with SSE2
37760 instruction set when @option{-ffast-math} is enabled
37762 2017-02-24 Jeff Law <law@redhat.com>
37764 PR rtl-optimizatoin/79286
37765 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37768 2017-02-24 Richard Biener <rguenther@suse.de>
37770 PR tree-optimization/79389
37771 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37774 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37776 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37777 function comment to reflect reality.
37778 (loop_exits_before_overflow): Fix typo in function description.
37780 2017-02-24 Richard Biener <rguenther@suse.de>
37782 PR tree-optimization/79389
37783 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37784 properly that a threading opportunity exists. Detect conditional
37785 copy/constant propagation opportunities.
37787 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37789 * config/visium/visium.md (type): Add trap.
37790 (b): New mode attribute.
37791 (*btst): Rename into...
37792 (*btst<mode>): ...this and adjust.
37793 (*cbranchsi4_btst_insn): Rename into...
37794 (*cbranch<mode>4_btst_insn): ...this and adjust.
37795 (trap): New define_insn.
37797 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37799 PR tree-optimization/79389
37800 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37801 (noce_reversed_cond_code): New function.
37802 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37803 reversed_comparison_code. Formatting fix.
37804 (noce_try_store_flag): Test rev_cond != NULL in addition to
37805 reversed_comparison_code.
37806 (noce_try_store_flag_constants): Likewise.
37807 (noce_try_store_flag_mask): Likewise.
37808 (noce_try_addcc): Use rev_cond if non-NULL instead of
37809 reversed_comparison_code.
37810 (noce_try_cmove_arith): Likewise. Formatting fixes.
37811 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37812 (noce_find_if_block): Initialize rev_cond.
37813 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37814 instead of false as last argument never attempt to reverse it
37817 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37819 PR tree-optimization/79663
37820 * tree-predcom.c (combine_chains): Process refs in reverse order
37821 only for ZERO length chains, and add explaining comment.
37823 2017-02-23 Jeff Law <law@redhat.com>
37825 PR tree-optimization/79578
37826 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37827 in call to operand_equal_p.
37829 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37832 * config/i386/cpuid.h: Fix another undefined behavior.
37834 2017-02-23 Richard Biener <rguenther@suse.de>
37836 PR tree-optimization/79683
37837 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37838 vector types for data-refs.
37840 2017-02-23 Martin Liska <mliska@suse.cz>
37842 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37844 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37846 PR middle-end/79665
37847 * internal-fn.c (get_range_pos_neg): Moved to ...
37848 * tree.c (get_range_pos_neg): ... here. No longer static.
37849 * tree.h (get_range_pos_neg): New prototype.
37850 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37851 are known to be in between 0 and signed maximum inclusive, try to
37852 expand both unsigned and signed divmod and use the cheaper one from
37855 2017-02-22 Jeff Law <law@redhat.com>
37857 PR tree-optimization/79578
37858 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37859 to compare base operands.
37861 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37864 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37865 gpc_reg_operand instead of fpr_reg_operand.
37867 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37869 * config/mips/mips.c (mips_return_in_memory): Force FP
37870 vector types to be returned in memory for o32 ABI.
37872 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37874 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37875 instead of DW_TAG_member for static data member declarations and don't
37876 set no_linkage_name for static inline data members.
37877 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37880 2017-02-22 Martin Liska <mliska@suse.cz>
37882 * doc/invoke.texi: Replace inequality signs with square brackets
37885 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37888 * lra-constraints.c (simplify_operand_subreg): Handle
37889 WORD_REGISTER_OPERATIONS targets.
37891 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37894 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37895 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37896 elimination by swapping fld*.
37898 2017-02-22 Richard Biener <rguenther@suse.de>
37900 PR tree-optimization/79673
37901 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37902 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37903 irrelevant address-space qualifiers and avoiding a
37904 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37906 2017-02-22 Richard Biener <rguenther@suse.de>
37908 PR tree-optimization/79666
37909 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37910 to not symbolically negate if that may introduce undefined
37913 2017-02-22 Martin Liska <mliska@suse.cz>
37916 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37917 * data-streamer-out.c (streamer_write_gcov_count_stream):
37919 * value-prof.c (stream_out_histogram_value): Make assert more
37920 precise based on type of counter.
37922 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37925 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37926 Use nonimmediate_operand instead of memory_operand for operand 1.
37927 (float-extend standard_x87sse_constant_load splitter): Ditto.
37929 2017-02-21 Jeff Law <law@redhat.com>
37931 PR tree-optimization/79621
37932 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37933 blocks with edges to themselves.
37935 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37938 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37939 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37940 Use gimple_call_builtin_p.
37943 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37944 on temporarily removed DEBUG_INSNs.
37946 PR tree-optimization/79649
37947 * tree-loop-distribution.c (classify_partition): Give up on
37948 non-generic address space loads/stores.
37950 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37952 * doc/loop.texi (Loop manipulation): Remove nonexistent
37953 tree_ssa_loop_version from the documentation.
37954 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37956 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37959 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37960 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37961 * config/rs6000/rs6000.c: Include except.h.
37962 (rs6000_expand_split_stack_prologue): Call
37963 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37965 2017-02-21 Martin Jambor <mjambor@suse.cz>
37968 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37969 have been analyzed.
37971 2017-02-21 Martin Jambor <mjambor@suse.cz>
37973 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37974 for backward compatibility only.
37975 * doc/invoke.texi (Option Summary): Remove all references to
37976 -fipa-cp-alignment.
37978 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37982 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37984 * lra-constraints.c (curr_insn_transform): Handle
37985 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37987 2017-02-21 Martin Liska <mliska@suse.cz>
37989 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37991 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37993 PR translation/79638
37994 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37996 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37999 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38000 (arm_function_ok_for_sibcall): Return false for an indirect call by
38001 descriptor if all the argument registers are used.
38002 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38003 alignment of the function.
38005 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38007 PR tree-optimization/61441
38008 * simplify-rtx.c (simplify_const_unary_operation): For
38009 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38010 the sNaN unmodified.
38012 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38014 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38015 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38016 instead of SYSTEM_HEADER_DIR.
38018 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38019 Martin Liška <mliska@suse.cz>
38021 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38022 Fix typos and grammar, use active voice, and clarify.
38024 2017-02-20 Marek Polacek <polacek@redhat.com>
38026 PR middle-end/79537
38027 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38030 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38032 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38035 * config/i386/i386.c (ix86_expand_builtin): Handle
38036 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38037 ix86_builtins_isa[fcode].isa as a requirement of those
38038 flags and any other flag in the bitmask.
38039 (ix86_init_mmx_sse_builtins): Use 0 instead of
38040 ~OPTION_MASK_ISA_64BIT as mask.
38041 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38042 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38043 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38044 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38046 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38049 * lra-constraints.c (split_reg): Check requested split mode
38050 is supported by the register.
38052 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38054 * lra-constraints.c (simplify_operand_subreg): Remove early
38057 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38060 * lra-constraints.c (curr_insn_transform): Tighten condition
38061 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38063 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38066 * lra-constraints.c (curr_insn_transform): Handle
38067 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38069 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38072 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38074 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38076 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38079 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38082 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38084 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38087 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38089 * final.c (last_columnnum, override_columnnum): New variables.
38090 (final_start_function): Set last_columnnum, pass it to begin_prologue
38091 hook and pass 0 to dwarf2out_begin_prologue.
38092 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38093 to source_line debug hook.
38094 (notice_source_line): Compute last_columnnum and for debug_column_info
38095 return true on column changes.
38096 * debug.h (struct gcc_debug_hooks): Add column argument to
38097 source_line and begin_prologue hooks.
38098 (debug_nothing_int_charstar_int_bool): Remove prototype.
38099 (debug_nothing_int_int_charstar,
38100 debug_nothing_int_int_charstar_int_bool): New prototypes.
38101 (dwarf2out_begin_prologue): Add column argument.
38102 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38103 begin_prologue hooks.
38104 (debug_nothing_int_charstar_int_bool): Remove.
38105 (debug_nothing_int_int_charstar,
38106 debug_nothing_int_int_charstar_int_bool): New functions.
38107 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38108 through to dwarf2out_source_line.
38109 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38110 (dwarf2out_source_line): Add column argument, emit it if requested.
38111 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38113 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38114 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38115 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38116 through to dwarf2out_begin_prologue.
38117 (vmsdbgout_source_line): Add column argument, pass it through to
38118 dwarf2out_source_line.
38119 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38120 dbxout_source_line caller.
38121 (dbxout_source_line): Add column argument.
38123 * common.opt (gno-column-info, gcolumn-info): New options.
38124 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38125 (check_die): Also test for multiple DW_AT_decl_column attributes.
38126 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38127 DW_AT_decl_column if requested.
38128 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38130 (gen_variable_die): Likewise.
38131 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38132 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38135 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38136 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38137 (ix86_handle_option): Handle OPT_m3dnowa.
38138 * doc/invoke.texi (-m3dnowa): Document.
38139 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38140 -m3dnowa instead of -m3dnow -march=athlon.
38143 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38144 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38146 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38149 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38150 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38151 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38152 generator for vsx_xxpermdi_<mode>_be.
38153 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38154 force big-endian semantics.
38155 (vsx_xxpermdi_<mode>_be): New define_expand with same
38156 implementation as previous version of vsx_xxpermdi_<mode>.
38158 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38160 PR tree-optimization/79327
38161 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38162 variable, its initialization and use.
38164 2017-02-17 Julia Koval <julia.koval@intel.com>
38166 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38167 (OPTION_MASK_ISA_PKU_UNSET): New.
38168 (ix86_handle_option): Handle -mrdpid.
38169 * config/i386/cpuid.h (bit_RDPID): New.
38170 * config/i386/driver-i386.c (host_detect_local_cpu):
38171 Detect RDPID feature.
38172 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38173 * config/i386/i386-c.c (ix86_target_macros_internal):
38175 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38176 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38177 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38178 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38179 * config/i386/i386.md (define_insn "rdpid"): New.
38180 * config/i386/i386.opt Add -mrdpid.
38181 * config/i386/immintrin.h (_rdpid_u32): New.
38183 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38185 PR rtl-optimization/79541
38186 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38187 instead of transforming it into USE.
38189 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38191 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38192 If HONOR_SNANS (SFmode) force the input to a register.
38193 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38194 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38195 an frsp or similar insn.
38197 2017-02-17 Martin Liska <mliska@suse.cz>
38199 PR rtl-optimization/79577
38200 * params.def (selsched-max-sched-times): Increase minimum to 1.
38202 2017-02-17 Martin Liska <mliska@suse.cz>
38204 PR rtl-optimization/79574
38205 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38207 2017-02-17 Martin Liska <mliska@suse.cz>
38209 PR tree-optimization/79529
38210 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38211 ssa_defined_default_def_p to handle cases which are implicitly
38213 * tree-ssa.c (ssa_defined_default_def_p): New function.
38214 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38215 which are implicitly defined.
38216 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38218 2017-02-17 Richard Biener <rguenther@suse.de>
38220 PR middle-end/79576
38221 * params.def (max-ssa-name-query-depth): Limit to 10.
38223 2017-02-17 Richard Biener <rguenther@suse.de>
38225 PR tree-optimization/79552
38226 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38229 2017-02-17 Richard Biener <rguenther@suse.de>
38232 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38234 2017-02-17 Marek Polacek <polacek@redhat.com>
38236 PR middle-end/79536
38237 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38238 (fold_negate_expr): New wrapper.
38240 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38242 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38243 Correct terminology and de-emphasize pre-standard behavior.
38245 2017-02-16 Alan Modra <amodra@gmail.com>
38247 PR rtl-optimization/79286
38248 * ira.c (def_dominates_uses): New function.
38249 (update_equiv_regs): Don't create an equivalence for insns that
38250 may trap where the register def does not dominate the use.
38252 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38254 PR rtl-optimization/78127
38255 * lra.c (lra): Call lra_eliminate before finish the loop after
38258 2017-02-16 Richard Biener <rguenther@suse.de>
38260 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38262 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38263 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38264 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38265 (isl_val_int_from_wi): New function.
38266 (extract_affine_gmp): Rename to ...
38267 (extract_affine_wi): ... this, take a widest_int.
38268 (extract_affine_int): Just wrap extract_affine_wi.
38269 (add_param_constraints): Use isl_val_int_from_wi.
38270 (add_loop_constraints): Likewise, and extract_affine_wi.
38272 2017-02-15 Jeff Law <law@redhat.com>
38274 PR middle-end/79521
38275 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38276 ira_init_register_move_cost_if_necessary.
38278 2017-02-15 Martin Sebor <msebor@redhat.com>
38280 PR middle-end/32003
38281 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38282 removed in a prior commit.
38284 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38286 PR tree-optimization/79347
38287 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38288 counters during peeling.
38290 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38292 * Makefile.in (site.exp): Remove "set ISLVER".
38294 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38297 * real.c (real_from_integer): Call real_convert even for decimal.
38299 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38302 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38304 2017-02-14 Andrew Pinski <apinski@cavium.com>
38306 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38307 cores and change the partno/implementer to be correct.
38308 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38309 the 'B" as the implementer.
38310 * config/aarch64/aarch64-tune.md: Regenerate.
38312 2017-02-14 Carl Love <cel@us.ibm.com>
38314 * config/rs6000/rs6000.c: Add case statement entry to make the
38315 xvcvuxdsp built-in argument unsigned.
38316 * config/rs6000/vsx.md: Fix the source and return operand types so they
38317 match the instruction definitions from the ISA document. Fix typo
38318 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38321 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38324 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38325 member early_clobber_alts.
38326 * lra-lives.c (reg_early_clobber_p): New.
38327 (process_bb_lives): Use it.
38328 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38329 (debug_operand_data): Initialize early_clobber_alts.
38330 (setup_operand_alternative): Set up early_clobber_alts.
38331 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38332 alternatives to new_insn_reg.
38333 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38335 (lra_update_insn_regno_info): Pass the new arg.
38337 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38339 PR middle-end/79505
38340 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38341 (new_oacc_loop_raw): Don't clear already cleared fields.
38344 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38345 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38346 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38348 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38351 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38353 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38356 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38357 the extra instruction to the right place to store 128-bit constant
38360 2017-02-14 Martin Sebor <msebor@redhat.com>
38362 PR middle-end/79448
38363 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38364 warning for strings of unknown length.
38366 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38368 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38370 2017-02-14 Jeff Law <law@redhat.com>
38373 * ira-costs.c (scan_one_insn): Initialize register move costs
38374 for pseudos seen in USE/CLOBBER insns.
38376 PR tree-optimization/79095
38377 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38378 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38379 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38380 if the operands are known to be not equal, then the resulting range
38382 (intersect_ranges): If the new range is ~[0,0] and the old range is
38383 wide, then prefer ~[0,0].
38384 * tree-vrp.c (overflow_comparison_p_1): New function.
38385 (overflow_comparison_p): New function.
38386 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38387 if NAME is used in an overflow test.
38388 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38389 overflow check that can be expressed as an equality test, then adjust
38390 ops to be that equality test.
38392 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38394 * config/s390/s390-builtin-types.def: Remove flags argument.
38395 * config/s390/s390.c (s390_init_builtins): Likewise.
38397 2017-02-14 Martin Liska <mliska@suse.cz>
38399 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38400 vector. Fix trailing white spaces.
38402 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38404 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38407 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38409 PR rtl-optimization/68664
38410 * config/arm/arm.c (arm_sched_can_speculate_insn):
38411 New function. Declare prototype.
38412 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38414 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38416 PR rtl-optimization/68664
38417 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38419 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38421 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38423 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38424 max skip bytes for function, loop and jump.
38426 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38428 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38429 ABS_EXPR for gimple dump.
38431 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38434 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38436 PR tree-optimization/79408
38437 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38438 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38439 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38440 also if rhs1 is INTEGER_CST.
38442 2017-02-14 Richard Biener <rguenther@suse.de>
38444 PR middle-end/79432
38445 * tree-into-ssa.c (insert_phi_nodes): When the function can
38446 have abnormal edges rewrite SSA names with broken use-def
38447 dominance out of SSA and register them for PHI insertion.
38449 2017-02-13 Martin Sebor <msebor@redhat.com>
38451 PR middle-end/79496
38452 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38453 clearing info.nowrite flag when snprintf size argument is a range.
38455 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38457 * cprop.c (cprop_jump): Add missing space in string literal.
38458 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38459 (get_constraint_for_component_ref): Likewise.
38460 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38461 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38462 * lra-constraints.c (process_alt_operands): Likewise.
38463 * ipa-inline.c (inline_small_functions): Likewise.
38464 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38465 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38466 * trans-mem.c (diagnose_tm_1_op): Likewise.
38467 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38468 (grid_parallel_clauses_gridifiable): Likewise.
38470 * config/nvptx/mkoffload.c (process): Add space in between
38473 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38474 "MOD4_SSE_REGS" and "ALL_REGS".
38476 * spellcheck.c (test_data): Add , in between "foo" and "food".
38478 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38481 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38482 boundary crossing check and subsequent code generation agree.
38484 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38486 * config/aarch64/aarch64.c (has_memory_op): Delete.
38487 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38490 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38492 PR rtl-optimization/79388
38493 PR rtl-optimization/79450
38494 * combine.c (distribute_notes): When removing TEM_INSN for which
38495 corresponding dest has last value recorded, invalidate that last
38498 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38500 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38501 of explicit '@'. Add missing assembly comment marker on branch costs
38504 2017-02-13 Nathan Sidwell <nathan@acm.org>
38506 * gengtype-lex.l (<in_struct>): Add '/'.
38508 2017-02-13 Martin Liska <mliska@suse.cz>
38511 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38513 2017-02-13 Richard Biener <rguenther@suse.de>
38515 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38517 * configure: Re-generate.
38518 * config.in: Likewise.
38519 * graphite-dependences.c: Simplify as if
38520 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38521 * graphite-isl-ast-to-gimple.c: Likewise.
38522 * graphite-optimize-isl.c: Likewise.
38523 * graphite-poly.c: Likewise.
38524 * graphite-sese-to-poly.c: Likewise.
38525 * graphite.h: Likewise.
38526 * toplev.c: Include isl/version.h and use isl_version () for
38527 printing the ISL version.
38528 * doc/install.texi: Update ISL requirement.
38530 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38532 * doc/standards.texi (Standards): Update reference to
38535 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38537 * doc/extend.texi (Named Address Spaces): sourceware.org now
38539 * doc/install.texi (Binaries): Ditto.
38542 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38544 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38545 standard terminology "stringize"/"stringizing" throughout.
38546 * doc/cppinternals.texi: Likewise.
38548 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38550 * doc/extend.texi: Fix some spelling mistakes and typos.
38551 * doc/invoke.texi: Likewise.
38553 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38556 * params.def (inline-min-speedup) Change from 10 to 8.
38558 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38560 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38563 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38566 * ipa-inline-analysis.c (get_minimal_bb): New function.
38567 (record_modified): Use it.
38568 (remap_edge_change_prob): Handle also ancestor functions.
38570 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38572 * doc/contrib.texi (Contributors): Remove broken link into
38573 the Mauve CVS repository.
38575 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38577 PR middle-end/79454
38578 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38579 result computation whenever lhs doesn't have vector mode, not
38580 just when it has BLKmode.
38582 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38584 * doc/makefile.texi (profiledbootstrap): Refer to the
38585 installation instructions only in textual form.
38587 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38590 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38592 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38594 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38595 (Specific): Update mingw-w64 reference.
38597 (Specific): Remove broken link to Renesas RX processor.
38599 2017-02-10 Richard Biener <rguenther@suse.de>
38601 * toplev.c (process_options): Do not mention obsolete graphite
38602 options when printing sorry message about missing graphite support.
38603 Mention -floop-nest-optimize.
38605 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38607 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38608 (vtst_p16): Likewise.
38609 (vtstq_p8): Likewise.
38610 (vtstq_p16): Likewise.
38612 (vtstq_p64): Likewise.
38613 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38614 (vset_lane_p64): New.
38615 (vsetq_lane_p64): New.
38617 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38619 PR tree-optimization/79411
38620 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38621 stmt operands are SSA_NAMEs used in abnormal phis.
38622 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38625 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38628 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38631 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38633 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38635 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38637 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38641 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38642 not arbitrary TREE_CONSTANT.
38645 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38646 "omp declare target link" attribute unless is_global_var.
38647 * omp-offload.c (find_link_var_op): Likewise.
38649 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38650 Chung-Lin Tang <cltang@codesourcery.com>
38652 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38654 (gimplify_adjust_omp_clauses): Don't delete TILE.
38655 (gimplify_omp_for): Deal with TILE.
38656 * internal-fn.c (expand_GOACC_TILE): New function.
38657 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38659 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38660 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38662 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38663 avoid DIV for outermost collapse var.
38664 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38665 Remove out of date comments, fix whitespace.
38666 * omp-general.c (omp_extract_for_data): Deal with tiling.
38667 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38668 adjust OLF_DIM_BASE value.
38669 (struct omp_for_data): Add tiling field.
38670 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38671 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38672 for auto loops. Remove default auto determining, moved to
38673 oacc_loop_fixed_partitions.
38674 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38675 stmts, add e_mask field.
38676 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38677 (oacc_thread_numbers): Use oacc_dim_call.
38678 (oacc_xform_tile): New.
38679 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38680 (finish_oacc_loop): Adjust for ifns vector.
38681 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38682 add case for GOACC_TILE fns.
38683 (oacc_loop_xform_loop): Delete.
38684 (oacc_loop_process): Iterate over call list directly, and add
38685 handling for GOACC_TILE fns.
38686 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38688 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38689 vector partitioning to outer loops. Assign 2 partitions to loops
38690 when available. Add TILE handling.
38691 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38692 (execite_oacc_device_lower): Process GOACC_TILE fns,
38693 ignore unknown specs.
38694 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38695 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38696 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38698 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38700 * configure.ac (ACX_BUGURL): Update.
38701 * configure: Regenerate.
38703 2017-02-09 Richard Biener <rguenther@suse.de>
38705 PR tree-optimization/69823
38706 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38707 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38709 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38711 * config/arc/arc-c.def: Add __NPS400__ definition.
38712 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38713 (TARGET_NPS400): Define.
38715 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38717 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38719 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38720 pointer, arch_info.
38721 (arc_cpu_types): Fill the arch_info field with a pointer into the
38722 arc_arch_types table.
38723 (arc_selected_cpu): Declare.
38724 * config/arc/arc.c (arc_selected_cpu): Make global.
38725 (arc_selected_arch): Delete.
38726 (arc_base_cpu): Delete.
38727 (arc_override_options): Remove references to deleted variables,
38728 update access to arch information.
38729 (ARC_OPT): Update access to arch information.
38730 (ARC_OPTX): Likewise.
38731 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38732 (TARGET_ARC600): Update access to arch information.
38733 (TARGET_ARC601): Likewise.
38734 (TARGET_ARC700): Likewise.
38735 (TARGET_EM): Likewise.
38736 (TARGET_HS): Likewise.
38737 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38740 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38743 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38744 condition/operands for integer GE/LE/GEU/LEU operations.
38746 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38748 PR translation/79397
38749 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38752 2017-02-08 Martin Jambor <mjambor@suse.cz>
38755 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38756 whether allocation happened.
38757 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38758 nothing was allocated.
38760 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38762 PR tree-optimization/79408
38763 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38764 constant, but SSA_NAME with a known integer range, use the minimum
38765 of that range instead of op1 to determine if modulo can be replaced
38766 with its first operand.
38768 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38770 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38772 2017-02-08 Richard Biener <rguenther@suse.de>
38774 PR tree-optimization/71824
38775 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38776 Check all loops contained in the merged region.
38778 2017-02-07 Andrew Pinski <apinski@cavium.com>
38780 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38782 2017-02-07 Andrew Pinski <apinski@cavium.com>
38784 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38785 (thunderxt88): Likewise.
38786 (thunderxt81): Disable LSE and change v8.1 to v8.
38787 (thunderxt83): Likewise.
38789 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38790 Richard Biener <rguenther@suse.de>
38792 PR middle-end/79399
38793 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38794 type from int to size_t.
38795 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38797 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38799 PR rtl-optimization/79386
38800 * cprop.c (bypass_conditional_jumps): Initialize
38801 bypass_last_basic_block already before splitting bbs after
38802 unconditional traps...
38803 (bypass_conditional_jumps): ... rather than here.
38806 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38807 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38808 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38809 fix -masm=intel patterns.
38811 2017-02-07 Richard Biener <rguenther@suse.de>
38813 PR tree-optimization/79256
38814 PR middle-end/79278
38815 * builtins.c (get_object_alignment_2): Use min_align_of_type
38816 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38817 and ADJUST_FIELD_ALIGN.
38819 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38821 * doc/tm.texi: Regenerate.
38822 * stor-layout.c (layout_decl): Adjust.
38823 (update_alignment_for_field): Likewise.
38824 (place_field): Likewise.
38825 (min_align_of_type): Likewise.
38826 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38827 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38828 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38829 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38830 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38831 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38832 * config/i386/i386.c (x86_field_alignment): Likewise.
38833 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38834 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38835 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38836 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38837 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38838 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38842 2017-01-30 Richard Biener <rguenther@suse.de>
38844 PR tree-optimization/79256
38845 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38846 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38849 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38851 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38852 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38853 builtins to SImode and emit a zero-extend, if necessary.
38855 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38857 * docs/invoke.texi (RISC-V Options): Alphabetize.
38859 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38861 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38864 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38866 * config/riscv/riscv.c: New file.
38867 * common/config/riscv/riscv-common.c: Likewise.
38868 * config.gcc: Likewise.
38869 * config/riscv/constraints.md: Likewise.
38870 * config/riscv/elf.h: Likewise.
38871 * config/riscv/generic.md: Likewise.
38872 * config/riscv/linux.h: Likewise.
38873 * config/riscv/multilib-generator: Likewise.
38874 * config/riscv/peephole.md: Likewise.
38875 * config/riscv/pic.md: Likewise.
38876 * config/riscv/predicates.md: Likewise.
38877 * config/riscv/riscv-builtins.c: Likewise.
38878 * config/riscv/riscv-c.c: Likewise.
38879 * config/riscv/riscv-ftypes.def: Likewise.
38880 * config/riscv/riscv-modes.def: Likewise.
38881 * config/riscv/riscv-opts.h: Likewise.
38882 * config/riscv/riscv-protos.h: Likewise.
38883 * config/riscv/riscv.h: Likewise.
38884 * config/riscv/riscv.md: Likewise.
38885 * config/riscv/riscv.opt: Likewise.
38886 * config/riscv/sync.md: Likewise.
38887 * config/riscv/t-elf-multilib: Likewise.
38888 * config/riscv/t-linux: Likewise.
38889 * config/riscv/t-linux-multilib: Likewise.
38890 * config/riscv/t-riscv: Likewise.
38891 * configure.ac: Likewise.
38892 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38893 Waterman as RISC-V maintainers.
38894 * doc/install.texi: Add RISC-V entries.
38895 * doc/invoke.texi: Add RISC-V options section.
38896 * doc/md.texi: Add RISC-V constraints section.
38897 * configure: Regenerated.
38899 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38902 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38903 false values to be constant vectors with all 0 or all 1 bits set.
38904 (vcondu<mode><mode>): Likewise.
38905 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38907 (fpmask_comparison_operator): Update comment.
38908 (vecint_comparison_operator): New predicate.
38909 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38910 vector conditionals when the true and false values are constant
38911 vectors with all 0 bits or all 1 bits set.
38913 2017-02-06 Martin Sebor <msebor@redhat.com>
38915 PR tree-optimization/79376
38916 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38918 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38920 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38921 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38922 to simplify split condition.
38924 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38926 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38927 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38930 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38932 PR rtl-optimization/68664
38933 * target.def (can_speculate_insn): New hook.
38934 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38935 * doc/tm.texi: Regenerate.
38936 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38937 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38938 (rs6000_sched_can_speculate_insn): New function.
38940 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38942 PR tree-optimization/79284
38943 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38944 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38945 vectorizable_mask_load_store, vectorizable_operation,
38946 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38947 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38948 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38949 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38950 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38951 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38952 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38953 is_gimple_assign (stmt). Replace another such test with
38954 is_gimple_assign (stmt).
38956 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38959 * config/avr/avr.c (rtl-iter.h): Include it.
38960 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38961 (avr_legitimate_combined_insn): ...and implementation.
38963 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38965 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38966 * config/s390/s390.c (s390_const_operand_ok)
38967 (s390_canonicalize_comparison, s390_extract_part)
38968 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38969 (s390_contiguous_bitmask_p, s390_rtx_costs)
38970 (legitimize_pic_address): Likewise.
38971 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38972 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38973 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38974 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38975 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38977 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38979 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38980 REGNO($0) == REGNO($1).
38982 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38984 * config/s390/linux.h(SIZE_TYPE): Add comment.
38986 2017-02-06 Julian Brown <julian@codesourcery.com>
38987 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38988 Virendra Pathak <virendra.pathak@broadcom.com>
38990 * config/aarch64/aarch64-cores.def: Change the scheduler
38992 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38993 * config/aarch64/thunderx2t99.md: New file.
38995 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38997 * doc/standards.texi (Go Language): Update link to language
39000 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39002 * tree-eh.c (lower_resx): Sanitize profile.
39003 (cleanup_empty_eh_move_lp): Likewise.
39005 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39008 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39010 * cfgloopmanip.h (loop_version): Update prototype.
39011 * modulo-sched.c (sms_schedule): Update call of loop_version.
39012 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39013 * tree-parloops.c (gen_parallel_loop): Likewise.
39014 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39015 * tree-ssa-loop-split.c (split_loop): Likewise.
39016 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39017 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39019 2017-02-05 Martin Liska <mliska@suse.cz>
39022 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39024 (print_operand_address): Initialize a struct to zero.
39026 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39028 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39029 garbage collector only in textual form.
39031 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39033 * doc/extend.texi (x86 specific memory model extensions for
39034 transactional memory): Simplify a phrase.
39036 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39039 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39040 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39041 (atomic_storedi_1): Likewise.
39043 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39045 PR tree-optimization/79338
39046 * tree-parloops.c (gather_scalar_reductions): Don't call
39047 vect_analyze_loop_form for loop->inner before destroying loop's
39050 2017-02-03 Martin Sebor <msebor@redhat.com>
39052 PR tree-optimization/79327
39053 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39054 when precision has resulted in leading zeros.
39055 (format_integer): Adjust the likely counter to assume an unknown
39056 argument that may be zero is non-zero.
39058 2017-02-03 Jason Merrill <jason@redhat.com>
39061 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39062 avoid copying non-taken branch.
39064 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39066 PR tree-optimization/79340
39067 * tree-vect-loop.c (vectorizable_reduction): Release
39068 vec_defs elements after safe_splicing them into other vectors.
39071 PR tree-optimization/79327
39072 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39073 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39075 (format_integer): Use wide_int_to_tree instead of build_int_cst
39076 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39077 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39078 of shortest and longest sequence.
39080 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39082 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39083 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39085 2017-02-03 Walter Lee <walt@tilera.com>
39088 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39089 after initial stackframe link reg save.
39090 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39092 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39095 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39096 wu for stxssp alternative.
39098 2017-02-03 Martin Sebor <msebor@redhat.com>
39100 PR tree-optimization/79352
39101 * gimple-fold.c (get_range_strlen): Add argument.
39102 (get_range_strlen): Change return type to bool.
39103 (get_maxval_strlen): Pass in a dummy argument.
39104 * gimple-fold.h (get_range_strlen): Change return type to bool.
39105 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39106 * tree.h (array_at_struct_end_p): Add argument.
39107 * tree.c (array_at_struct_end_p): Handle it.
39109 2017-02-03 Martin Liska <mliska@suse.cz>
39112 * multiple_target.c (create_dispatcher_calls): Redirect edge
39113 from a caller of a dispatcher.
39114 (expand_target_clones): Make the clones local.
39115 (ipa_target_clone): Do both target clones and resolvers.
39116 (ipa_dispatcher_calls): Remove the pass.
39117 (pass_dispatcher_calls::gate): Likewise.
39118 (make_pass_dispatcher_calls): Likewise.
39119 * passes.def (pass_target_clone): Put as very first IPA early
39122 2017-02-03 Martin Liska <mliska@suse.cz>
39124 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39125 in case of a function with ifunc attribute.
39127 2017-02-03 Martin Liska <mliska@suse.cz>
39129 * cgraph.c (cgraph_node::dump): Dump function version info.
39130 * symtab.c (symtab_node::dump_base): Add missing new line.
39132 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39134 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39135 (ifcombine_ifandif): Use it.
39137 2017-02-03 Martin Liska <mliska@suse.cz>
39139 * doc/invoke.texi: Document default value for
39140 use-after-scope-direct-emission-threshold.
39142 2017-02-03 Martin Liska <mliska@suse.cz>
39144 PR tree-optimization/79339
39145 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39146 (format_floating): Likewise.
39148 2017-02-03 Martin Liska <mliska@suse.cz>
39151 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39153 (ipa_node_params_t::remove): Likewise.
39154 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39155 initialization from removed ipa_node_params_t::insert.
39156 (ipa_node_params::~ipa_node_params): Move from removed
39157 ipa_node_params_t::release.
39158 * symbol-summary.h (symbol_summary::m_released): New member.
39159 Do not release a summary twice. Do not allow to call finalizer
39160 for types of a summary that live in GGC memory.
39162 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39164 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39167 2017-02-02 Martin Sebor <msebor@redhat.com>
39169 PR middle-end/79275
39170 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39171 (format_string): Tighten up the range of output for non-constant
39172 strings and correct the expected range for wide non-constant strings.
39174 2017-02-02 Martin Sebor <msebor@redhat.com>
39176 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39178 PR middle-end/32003
39179 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39181 (-fdump-tree-@var): Add to index and document how to come up
39182 with pass-specific option and dump file names.
39183 (-fdump-passes): Clarify where to look for output.
39185 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39187 PR middle-end/77445
39188 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39189 statistics of the analyzed path; allow threading for speed when
39190 any of BBs along the path are optimized for speed.
39192 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39194 PR middle-end/78468
39195 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39196 settings of the virtual registers.
39199 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39201 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39202 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39205 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39207 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39208 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39210 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39212 * config/s390/s390.md: Add missing comments with the expanded
39214 * config/s390/vector.md: Likewise.
39215 * config/s390/vx-builtins.md: Likewise.
39217 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39220 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39221 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39222 conditions on a single line.
39224 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39226 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39227 __S390_VX__ to __VX__.
39229 2017-02-01 Andrew Pinski <apinski@cavium.com>
39231 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39232 stmt_info to record_stmt_cost.
39233 (vect_get_known_peeling_cost): Pass stmt_info if known to
39235 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39236 cpu_vector_cost field into
39237 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39238 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39239 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39240 splitting of scalar_stmt_cost and vec_stmt_cost.
39241 (thunderx_vector_cost): Likewise.
39242 (cortexa57_vector_cost): LIkewise.
39243 (exynosm1_vector_cost): Likewise.
39244 (xgene1_vector_cost): Likewise.
39245 (thunderx2t99_vector_cost): Improve after the splitting of the two
39247 (aarch64_builtin_vectorization_cost): Update for the splitting of
39248 scalar_stmt_cost and vec_stmt_cost.
39250 2017-02-01 Torvald Riegel <triegel@redhat.com>
39251 Richard Henderson <rth@redhat.com>
39253 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39254 conditional on existance of a fast atomic load.
39255 * optabs-query.c (can_atomic_load_p): New function.
39256 * optabs-query.h (can_atomic_load_p): Declare it.
39257 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39258 no fast atomic load is available for the particular size of access.
39259 (expand_atomic_compare_and_swap): Likewise.
39260 (expand_atomic_load): Likewise.
39261 (expand_atomic_store): Likewise.
39262 (expand_atomic_fetch_op): Likewise.
39263 * testsuite/lib/target-supports.exp
39264 (check_effective_target_sync_int_128): Remove x86 because it provides
39265 no fast atomic load.
39266 (check_effective_target_sync_int_128_runtime): Likewise.
39268 2017-02-01 Richard Biener <rguenther@suse.de>
39270 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39271 (graphite_transform_loops): Provide opt-info for optimized nests.
39272 * tree-parloop.c (parallelize_loops): Provide opt-info for
39273 parallelized loops.
39275 2017-02-01 Richard Biener <rguenther@suse.de>
39277 PR middle-end/79315
39278 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39279 was not set before.
39281 2017-02-01 Richard Biener <rguenther@suse.de>
39283 PR tree-optimization/71824
39284 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39285 Verify the loops are valid in the merged SESE region.
39286 (scop_detection::can_represent_loop_1): Check analyzing the
39287 evolution of the number of iterations in the region succeeds.
39289 2017-01-31 Ian Lance Taylor <iant@golang.org>
39291 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39292 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39294 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39296 PR preprocessor/79210
39297 * input.c (get_substring_ranges_for_loc): Replace line_width
39298 assertion with error-handling.
39300 2017-01-31 Richard Biener <rguenther@suse.de>
39302 PR tree-optimization/77318
39303 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39304 (create_pw_aff_from_tree): Take loop parameter.
39305 (add_condition_to_pbb): Pass loop of the condition to
39306 create_pw_aff_from_tree.
39308 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39310 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39311 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39313 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39317 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39319 (convert_int_to_float128): Likewise.
39320 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39321 (convert_int_to_float128): Likewise.
39322 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39323 (UNSPEC_IEEE128_CONVERT): Likewise.
39324 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39325 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39326 Use local variables for IBM extended format.
39327 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39328 (fix_trunc<mode>si2_fprs): Likewise.
39329 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39330 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39331 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39332 to know that we can now have integers of all sizes in vector
39334 (fix<uns>_<mode>di2_hw): Likewise.
39335 (float<uns>_<mode>si2_hw): Likewise.
39336 (fix_<mode>si2_hw): Likewise.
39337 (fixuns_<mode>si2_hw): Likewise.
39338 (float<uns>_<mode>di2_hw): Likewise.
39339 (float_<mode>di2_hw): Likewise.
39340 (float_<mode>si2_hw): Likewise.
39341 (floatuns_<mode>di2_hw): Likewise.
39342 (floatuns_<mode>si2_hw): Likewise.
39343 (xscvqp<su>wz_<mode>): Delete, no longer used.
39344 (xscvqp<su>dz_<mode>): Likewise.
39345 (xscv<su>dqp_<mode>): Likewise.
39346 (ieee128_mfvsrd_64bit): Likewise.
39347 (ieee128_mfvsrd_32bit): Likewise.
39348 (ieee128_mfvsrwz): Likewise.
39349 (ieee128_mtvsrw): Likewise.
39350 (ieee128_mtvsrd_64bit): Likewise.
39351 (ieee128_mtvsrd_32bit): Likewise.
39353 2017-01-31 Martin Liska <mliska@suse.cz>
39356 * ipa-prop.c (ipa_free_all_node_params): Call release method
39357 instead of ~sumbol_summary to not to trigger double times
39360 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39362 PR tree-optimization/71691
39363 * bitmap.h (class auto_bitmap): New.
39364 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39365 is_maybe_undefined instead of ssa_undefined_value_p.
39367 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39369 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39370 __S390_ARCH_LEVEL__ to __ARCH__.
39372 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39374 PR tree-optimization/79267
39375 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39376 if should_remove_lhs_p is true.
39378 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39381 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39382 (add_alignment_attribute): New.
39383 (base_type_die): Add alignment attribute.
39384 (subrange_type_die): Likewise.
39385 (modified_type_die): Likewise.
39386 (gen_array_type_die): Likewise.
39387 (gen_descr_array_type_die: Likewise.
39388 (gen_enumeration_type_die): Likewise.
39389 (gen_subprogram_die): Likewise.
39390 (gen_variable_die): Likewise.
39391 (gen_field_die): Likewise.
39392 (gen_ptr_to_mbr_type_die): Likewise.
39393 (gen_struct_or_union_type_die): Likewise.
39394 (gen_subroutine_type_die): Likewise.
39395 (gen_typedef_die): Likewise.
39396 (base_type_cmp): Compare alignment attribute.
39398 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39401 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39402 (setb_unsigned) New pattern for setb with CCUNS.
39403 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39404 subfc./subfe sequence to avoid overflow problems. Generate a
39405 shorter sequence with cmpld/setb for power9.
39406 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39407 for generating subfc. instruction.
39408 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39409 now uses this instruction.
39411 2017-01-30 Ian Lance Taylor <iant@google.com>
39414 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39415 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39417 2017-01-30 Martin Sebor <msebor@redhat.com>
39419 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39420 Move constant to the right of a relational operator.
39421 (get_mpfr_format_length, format_character, format_string): Ditto.
39422 (should_warn_p, maybe_warn): Same.
39424 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39426 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39429 * asan.c (get_translation_unit_decl): Remove function.
39430 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39432 2017-01-30 Martin Liska <mliska@suse.cz>
39434 PR gcov-profile/79259
39435 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39436 -fprofile-generate.
39438 2017-01-30 Martin Liska <mliska@suse.cz>
39441 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39442 Initialize variables with NULL value.
39444 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39447 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39449 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39451 2017-01-30 Richard Biener <rguenther@suse.de>
39453 PR tree-optimization/79276
39454 * tree-vrp.c (process_assert_insertions): Properly adjust common
39455 when removing a duplicate.
39457 2017-01-30 Richard Biener <rguenther@suse.de>
39459 PR tree-optimization/79256
39460 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39461 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39463 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39465 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39468 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39469 ("*r<noxa>sbg_<mode>_sll_bitmask")
39470 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39471 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39472 Use contiguous_bitmask_nowrap_operand.
39474 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39477 * config/rs6000/altivec.h (vec_xl): Revise #define.
39478 (vec_xst): Likewise.
39480 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39482 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39484 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39486 PR rtl-optimization/79194
39487 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39488 traps before call to bypass_conditional_jumps.
39490 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39492 PR tree-optimization/71374
39493 * lra-constraints.c (check_conflict_input_operands): New.
39494 (match_reload): Use it.
39496 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39499 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39500 account to calculate conflict_set.
39502 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39504 PR rtl-optimization/78559
39505 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39506 other_insn in combine.
39508 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39510 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39511 uint16_type_node for BT_UINT16.
39513 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39515 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39516 "RTL Tests" to menu.
39517 (GIMPLE Tests): New node.
39518 (RTL Tests): New node.
39520 2017-01-27 Richard Biener <rguenther@suse.de>
39522 PR tree-optimization/79245
39523 * tree-loop-distribution.c (distribute_loop): Apply cost
39524 modeling also to detected patterns.
39526 2017-01-27 Richard Biener <rguenther@suse.de>
39528 PR tree-optimization/71433
39529 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39530 (compare_assert_loc): New function.
39531 (process_assert_insertions): Sort and optimize assert locations
39532 to remove duplicates and push down identical assertions on
39533 edges to their destination block.
39535 2017-01-27 Richard Biener <rguenther@suse.de>
39537 PR tree-optimization/79244
39538 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39539 out SSA names even if abnormal.
39541 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39543 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39544 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39545 instead of MPFR_RNDN.
39547 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39550 * arm.c (arm_option_override): Don't call build_target_option_node
39551 until after doing all option overrides.
39552 (arm_valid_target_attribute_tree): Likewise.
39554 2017-01-27 Martin Liska <mliska@suse.cz>
39556 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39557 for {cd}tors and C++ {cd}tors.
39559 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39561 * config/s390/s390.md ("*setmem_long_and")
39562 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39564 2017-01-26 Martin Sebor <msebor@redhat.com>
39566 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39569 2017-01-26 Martin Sebor <msebor@redhat.com>
39571 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39572 HAVE_DFmode before using XFmode or DFmode.
39573 (parse_directive): Avoid using the z length modifier to avoid
39574 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39576 PR middle-end/78703
39577 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39578 to accept adjustment as an array.
39579 (get_int_range): New function.
39580 (struct directive): Make width and prec arrays.
39581 (directive::set_width, directive::set_precision): Call get_int_range.
39582 (format_integer, format_floating): Handle width and precision ranges.
39583 (format_string, parse_directive): Same.
39585 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39588 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39589 comdat_type_p set, just clone them, but keep the children in the
39593 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39594 which have direct callers with -fvar-tracking-assignments enabled
39596 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39597 inside of type units.
39599 2017-01-26 Martin Sebor <msebor@redhat.com>
39601 PR middle-end/78703
39602 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39604 (struct format_result): Replace number_chars, number_chars_min,
39605 and number_chars_max with a single member of struct result_range.
39607 (format_result::operator+=): Adjust.
39608 (struct fmtresult): Remove bounded. Handle likely and unlikely
39610 (fmtresult::adjust_for_width_or_precision): New function.
39611 (fmtresult:type_max_digits): New function.
39612 (bytes_remaining): Handle likely and unlikely counters.
39613 (min_bytes_remaining): Remove.
39614 (format_percent): Simplify.
39615 (format_integer, format_floating): Set likely and unlikely counters.
39616 (get_string_length, format_character, format_string): Same.
39617 (format_plain, should_warn_p): New function.
39618 (maybe_warn): Call should_warn_p. Update diagnostic messages
39619 and handle those for all directives, including plain strings.
39620 (format_directive): Handle likely and unlikely counters.
39621 Remove unnecessary quoting from diagnostics. Add an informational
39623 (add_bytes): Remove.
39624 (pass_sprintf_length::compute_format_length): Simplify.
39625 (try_substitute_return_value): Handle likely and unlikely counters.
39627 2017-01-26 Carl Love <cel@us.ibm.com>
39629 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39630 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39632 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39635 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39636 endianess for subregs into account.
39637 * lra-constraints.c (lra_constraints): Do risky transformations
39638 always on the first iteration.
39639 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39640 last_call_used_reg_set.
39641 (process_bb_lives): Define and use last_call_used_reg_set.
39642 * lra.c (lra): Always continue after lra_constraints on the first
39645 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39647 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39649 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39651 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39653 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39654 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39655 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39656 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39657 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39658 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39659 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39660 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39661 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39663 2017-01-26 Marek Polacek <polacek@redhat.com>
39666 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39667 for the third operand.
39669 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39671 PR middle-end/79236
39672 * omp-low.c (struct omp_context): Add simt_stmt field.
39673 (scan_omp_for): Return omp_context *.
39674 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39675 context to the _simt_ SIMD stmt.
39676 (lower_omp_for): For combined SIMD with sibling _simt_
39677 SIMD, make sure to use the same decls in _looptemp_
39678 clauses as in the sibling.
39680 2017-01-26 David Sherwood <david.sherwood@arm.com>
39682 PR middle-end/79212
39683 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39686 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39689 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39690 emit fld b; fld a; if possible.
39692 * brig-builtins.def: Update copyright years.
39693 * config/arm/arm_acle_builtins.def: Update copyright years.
39695 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39698 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39699 constraint instead of o for the stxsd instruction.
39701 2017-01-25 Carl Love <cel@us.ibm.com>
39703 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39704 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39706 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39708 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39710 2017-01-25 Richard Biener <rguenther@suse.de>
39712 PR tree-optimization/69264
39713 * target.def (vector_alignment_reachable): Improve documentation.
39714 * doc/tm.texi: Regenerate.
39715 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39717 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39718 earlier changes with respect to TYPE_USER_ALIGN.
39719 (vector_alignment_reachable_p): Likewise. Improve dumping.
39721 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39724 * config/arm/arm.md (xordi3): Force constant operand into a register
39727 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39729 * doc/invoke.texi (-fstore-merging): Correct default optimization
39730 levels at which it is enabled.
39731 (-O): Move -fstore-merging from list to...
39734 2017-01-25 Richard Biener <rguenther@suse.de>
39737 * omp-expand.c: Include debug.h.
39738 (expand_omp_taskreg): Make sure to generate early debug before
39739 outlining anything from a function.
39740 (expand_omp_target): Likewise.
39741 (grid_expand_target_grid_body): Likewise.
39743 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39746 * asan.c (get_translation_unit_decl): New function.
39747 (asan_add_global): Extract modules file name from globals
39748 TRANSLATION_UNIT_DECL name.
39750 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39753 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39754 for long calls with APCS frame and VFP.
39756 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39758 * cfg.c (original_copy_tables_initialized_p): New function.
39759 * cfg.h (original_copy_tables_initialized_p): New decl.
39760 * cfgrtl.c (relink_block_chain): Guard the call to
39761 free_original_copy_tables with a call to
39762 original_copy_tables_initialized_p.
39763 * cgraph.h (symtab_node::native_rtl_p): New decl.
39764 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39765 (symtab_node::needed_p): Don't assert for early assembly output
39766 for __RTL functions.
39767 (cgraph_node::finalize_function): Set "force_output" for __RTL
39769 (cgraph_node::analyze): Bail out early for __RTL functions.
39770 (analyze_functions): Update assertion to support __RTL functions.
39771 (cgraph_node::expand): Bail out early for __RTL functions.
39772 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39774 * function.h (struct function): Update comment for field
39776 * gimple-expr.c: Include "tree-pass.h".
39777 (gimple_has_body_p): Return false for __RTL functions.
39778 * Makefile.in (OBJS): Add run-rtl-passes.o.
39779 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39781 (gcc::pass_manager::get_clean_slate): New accessor.
39782 * passes.c: Include "insn-addr.h".
39783 (should_skip_pass_p): Add logging. Update logic for running
39784 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39785 property-provider override so it is only done for gimple passes.
39787 (skip_pass): New function.
39788 (execute_one_pass): Call skip_pass when skipping passes.
39789 * read-md.c (md_reader::read_char): Support filtering
39790 the input to a subset of line numbers.
39791 (md_reader::md_reader): Initialize fields
39792 m_first_line and m_last_line.
39793 (md_reader::read_file_fragment): New function.
39794 * read-md.h (md_reader::read_file_fragment): New decl.
39795 (md_reader::m_first_line): New field.
39796 (md_reader::m_last_line): New field.
39797 * read-rtl-function.c (function_reader::create_function): Only
39798 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39799 curr_properties. Set DECL_INITIAL to a dummy block.
39800 (read_rtl_function_body_from_file_range): New function.
39801 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39803 * run-rtl-passes.c: New file.
39804 * run-rtl-passes.h: New file.
39806 2017-01-24 Jeff Law <law@redhat.com>
39808 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39811 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39813 PR tree-optimization/79159
39814 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39815 (record_nonwrapping_iv): Improve boundary using above function if no
39816 value range information.
39818 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39819 Martin Jambor <mjambor@suse.cz>
39821 * brig-builtins.def: New file.
39822 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39823 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39824 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39825 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39826 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39827 * builtin-types.def (BT_INT8): New.
39828 (BT_INT16): Likewise.
39829 (BT_UINT8): Likewise.
39830 (BT_UINT16): Likewise.
39831 (BT_FN_ULONG): Likewise.
39832 (BT_FN_UINT_INT): Likewise.
39833 (BT_FN_UINT_ULONG): Likewise.
39834 (BT_FN_UINT_LONG): Likewise.
39835 (BT_FN_UINT_PTR): Likewise.
39836 (BT_FN_ULONG_PTR): Likewise.
39837 (BT_FN_INT8_FLOAT): Likewise.
39838 (BT_FN_INT16_FLOAT): Likewise.
39839 (BT_FN_UINT32_FLOAT): Likewise.
39840 (BT_FN_UINT16_FLOAT): Likewise.
39841 (BT_FN_UINT8_FLOAT): Likewise.
39842 (BT_FN_UINT64_FLOAT): Likewise.
39843 (BT_FN_UINT16_UINT32): Likewise.
39844 (BT_FN_UINT32_UINT16): Likewise.
39845 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39846 (BT_FN_INT_PTR_INT): Likewise.
39847 (BT_FN_UINT_PTR_UINT): Likewise.
39848 (BT_FN_LONG_PTR_LONG): Likewise.
39849 (BT_FN_ULONG_PTR_ULONG): Likewise.
39850 (BT_FN_VOID_UINT64_UINT64): Likewise.
39851 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39852 (BT_FN_INT8_INT8_INT8): Likewise.
39853 (BT_FN_INT16_INT16_INT16): Likewise.
39854 (BT_FN_INT_INT_INT): Likewise.
39855 (BT_FN_UINT_FLOAT_UINT): Likewise.
39856 (BT_FN_FLOAT_UINT_UINT): Likewise.
39857 (BT_FN_ULONG_UINT_UINT): Likewise.
39858 (BT_FN_ULONG_UINT_PTR): Likewise.
39859 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39860 (BT_FN_UINT_UINT_UINT): Likewise.
39861 (BT_FN_VOID_UINT_PTR): Likewise.
39862 (BT_FN_UINT_UINT_PTR: Likewise.
39863 (BT_FN_UINT32_UINT64_PTR): Likewise.
39864 (BT_FN_INT_INT_UINT_UINT): Likewise.
39865 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39866 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39867 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39868 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39869 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39870 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39871 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39872 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39873 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39874 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39875 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39876 * doc/frontends.texi: List BRIG FE.
39877 * doc/install.texi (Testing): Add BRIG tesring requirements.
39878 * doc/invoke.texi (Overall Options): Mention BRIG.
39879 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39881 2017-01-24 Richard Biener <rguenther@suse.de>
39883 PR translation/79208
39884 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39886 2017-01-24 Martin Jambor <mjambor@suse.cz>
39889 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39890 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39891 and known_contexts.
39893 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39895 PR middle-end/79123
39896 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39897 casts from signed to unsigned really don't have a range.
39899 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39901 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39902 GMP_RNDx for compatiblity.
39904 2017-01-24 Martin Liska <mliska@suse.cz>
39907 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39908 that would prevent us to call alloca with -1 as argument.
39910 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39912 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39913 Avoid -Wformat-security warning.
39915 2017-01-23 Andrew Pinski <apinski@cavium.com>
39917 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39920 2017-01-23 Martin Sebor <msebor@redhat.com>
39922 PR middle-end/78703
39923 * gimple-ssa-sprintf.c (warn_level): New global.
39924 (format_integer): Use it here and throughout the rest of the file.
39925 Use the same switch to compute sign as base.
39926 (maybe_warn): New function.
39927 (format_directive): Factor out warnings into maybe_warn.
39928 Add debugging output. Use warn_level.
39929 (add_bytes): Use warn_level.
39930 (pass_sprintf_length::compute_format_length): Add debugging output.
39931 (try_substitute_return_value): Same.
39932 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39934 PR middle-end/78703
39935 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39936 (struct fmtresult, format_integer, format_floating): Adjust.
39937 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39938 (get_string_length, format_string,format_directive): Same.
39939 (pass_sprintf_length::compute_format_length): Same.
39940 (try_substitute_return_value): Simplify slightly.
39942 PR middle-end/78703
39943 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39944 (fmtresult::operator+=): Outlined.
39945 (struct fmtresult): Add ctors.
39946 (struct conversion_spec): Rename...
39947 (struct directive): ...to this. Add and remove data members.
39948 (directive::set_width, directive::set_precision): New functions.
39949 (format_percent): Use fmtresult ctor.
39950 (get_width_and_precision): Remove.
39951 (format_integer): Make naming changes. Avoid computing width and
39953 (format_floating): Same. Adjust indentation.
39954 (format_character, format_none): New functions.
39955 (format_string): Moved character handling to format_character.
39956 (format_directive): Remove arguments, change return type.
39957 (parse_directive): New function.
39958 (pass_sprintf_length::compute_format_length): Move directive
39959 parsing to parse_directive.
39961 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39963 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39964 (assign_assembler_name_if_needed): ... this.
39965 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39966 (assign_assembler_name_if_needed): ... this.
39967 (free_lang_data_in_cgraph): Adjust callers.
39968 * cgraphunit.c (cgraph_node::analyze): Likewise.
39969 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39972 2017-01-23 Richard Biener <rguenther@suse.de>
39974 PR tree-optimization/79088
39975 PR tree-optimization/79188
39976 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39977 resetting loop bounds after last path deletion. Reset loop
39978 bounds of the target loop, make code match the comments.
39979 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39980 Make sure loops need no fixups.
39982 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39984 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39985 exponent support with double type for first argument.
39986 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39987 type returned by __builtin_vec_extract_sig,
39988 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39989 functions from "vector int" to "vector unsigned int" or from
39990 "vector long long int" to "vector unsigned long long int".
39991 Changed type returned by __builtin_vec_extract_exp,
39992 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39993 functions from "vector int" to "vector unsigned int" or from
39994 "vector long long int" to "vector unsigned long long int".
39995 Changed return type of __builtin_vec_test_data_class,
39996 __builtin_vec_test_data_class_sp, and
39997 __builtin_vec_test_data_class_dp from "vector int" to
39998 "vector bool int" or from "vector long long int" to "vector bool
39999 long long int" and changed second argument type from "unsigned
40000 int" to "int". Added new overloaded function forms "vector float
40001 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40002 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40003 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40004 double, vector unsigned long long int)" and "vector double
40005 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40006 long int)". Changed return type of
40007 __builtin_scalar_test_data_class and
40008 __builtin_scalar_test_data_class_sp and
40009 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40010 int" and changed second argument from "unsigned int" to "int".
40011 Changed type returned by __builtin_scalar_test_neg,
40012 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40013 from "int" to "bool int". Added new overloaded function form
40014 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40015 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40016 exponent double-precision with floating point first argument.
40017 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40018 documentation of scalar_test_data_class, scalar_test_neg,
40019 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40020 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40021 vec_test_data_class built-in functions to reflect refinements in
40022 their type signatures.
40024 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40026 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40028 (aarch64_elf_asm_destructor): Likewise.
40030 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40032 PR rtl-optimization/78634
40033 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40034 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40035 * ifcvt.c (noce_try_cmove): Add missing cost check.
40037 PR rtl-optimization/71724
40038 * combine.c (if_then_else_cond): Look for situations where it is
40039 beneficial to undo the work of one of the recursive calls.
40041 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40043 PR tree-optimization/70754
40044 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40045 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40046 combined stmt before it if not NULL.
40047 (combine_chains): Process refs reversely and compute dominance point
40050 2017-01-23 Martin Liska <mliska@suse.cz>
40052 PR tree-optimization/79196
40053 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40054 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40056 (strlen_optimize_stmt): Call the renamed function.
40058 2017-01-23 Michael Matz <matz@suse.de>
40060 PR tree-optimization/78384
40061 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40063 2017-01-23 Richard Biener <rguenther@suse.de>
40065 PR tree-optimization/79186
40066 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40067 both incoming edges before moving an assert.
40069 2017-01-23 Martin Jambor <mjambor@suse.cz>
40071 * ipa-prop.c (load_from_param_1): Removed.
40072 (load_from_unmodified_param): Bits from load_from_param_1 put back
40074 (load_from_param): Removed.
40075 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40076 with stmt. Reverted back to use of load_from_unmodified_param.
40078 2017-01-23 Martin Jambor <mjambor@suse.cz>
40081 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40082 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40083 field a pointer to garbage collected vector, mark lattices and
40084 ipcp_orig_node with GTY((skip)).
40085 (ipa_get_param_count): Adjust to descriptors being a pointer.
40086 (ipa_get_param): Likewise.
40087 (ipa_get_type): Likewise.
40088 (ipa_get_param_move_cost): Likewise.
40089 (ipa_set_param_used): Likewise.
40090 (ipa_get_controlled_uses): Likewise.
40091 (ipa_set_controlled_uses): Likewise.
40092 (ipa_is_param_used): Likewise.
40093 (ipa_node_params_t): Move into garbage collector. New methods insert
40095 (ipa_node_params_sum): Annotate wth GTY(()).
40096 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40098 (ipa_load_from_parm_agg): Adjust declaration.
40099 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40100 * ipa-profile.c (ipa_profile): Likewise.
40101 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40102 (ipa_populate_param_decls): Make descriptors parameter garbage
40104 (ipa_dump_param): Adjust to descriptors being a pointer.
40105 (ipa_alloc_node_params): Likewise.
40106 (ipa_initialize_node_params): Likewise.
40107 (load_from_param_1): Make descriptors parameter garbage collected.
40108 (load_from_unmodified_param): Likewise.
40109 (load_from_param): Likewise.
40110 (ipa_load_from_parm_agg): Likewise.
40111 (ipa_node_params::~ipa_node_params): Removed.
40112 (ipa_free_all_node_params): Remove call to delete operator.
40113 (ipa_node_params_t::insert): New.
40114 (ipa_node_params_t::remove): Likewise.
40115 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40116 copy known_csts and known_contexts vectors.
40117 (ipa_read_node_info): Adjust to descriptors being a pointer.
40118 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40120 (ipcp_transform_function): Make descriptors variable garbage
40123 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40125 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40126 * config/i386/avx512dqintrin.h: Ditto.
40127 * config/i386/avx512fintrin.h: Ditto.
40128 * config/i386/i386.c: Handle new builtins.
40129 * config/i386/i386-builtin.def: Add new builtins.
40130 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40131 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40133 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40134 Martin Liska <mliska@suse.cz>
40136 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40137 * asan.c (asan_expand_poison_ifn): Support stores and use
40138 appropriate ASAN report function.
40139 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40140 * internal-fn.def (ASAN_POISON_USE): Declare.
40141 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40142 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40143 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40144 ASAN_POISON calls w/o LHS.
40145 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40146 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40147 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40148 * gimplify.c (asan_poison_variables): Add attribute
40149 use_after_scope_memory to variables that really needs to live
40151 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40152 having the attribute.
40154 2017-01-23 Martin Liska <mliska@suse.cz>
40156 * asan.c (create_asan_shadow_var): New function.
40157 (asan_expand_poison_ifn): Likewise.
40158 * asan.h (asan_expand_poison_ifn): New declaration.
40159 * internal-fn.c (expand_ASAN_POISON): Likewise.
40160 * internal-fn.def (ASAN_POISON): New builtin.
40161 * sanopt.c (pass_sanopt::execute): Expand
40162 asan_expand_poison_ifn.
40163 * tree-inline.c (copy_decl_for_dup_finish): Make function
40165 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40166 * tree-ssa.c (is_asan_mark_p): New function.
40167 (execute_update_addresses_taken): Rewrite local variables
40168 (identified just by use-after-scope as addressable) into SSA.
40170 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40172 * doc/install.texi (Specific): opensource.apple.com uses https
40173 now. Remove trailing slash.
40175 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40177 * README.Portability: Remove note on an Irix compatibility issue.
40179 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40181 * gcov.c (INCLUDE_ALGORITHM): Define.
40182 (INCLUDE_VECTOR): Define.
40183 No longer include <vector> and <algorithm> directly.
40185 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40187 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40189 * doc/invoke.texi (Code Gen Options): Ditto.
40191 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40194 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40196 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40198 rtl-optimization/79125
40199 * cprop.c (local_cprop_pass): Handle cases where we make an
40200 unconditional trap.
40202 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40206 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40207 read from, for big endian.
40209 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40211 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40212 register pauth builtins for LP64 only.
40214 2017-01-20 Marek Polacek <polacek@redhat.com>
40217 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40220 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40222 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40224 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40225 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40226 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40228 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40231 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40232 in big-endian mode when they are not a single duplicated value.
40234 2017-01-20 Richard Biener <rguenther@suse.de>
40236 * BASE-VER: Bump to 7.0.1.
40238 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40240 * omp-low.c (omplow_simd_context): New struct. Use it...
40241 (lower_rec_simd_input_clauses): ...here and...
40242 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40243 references to idx, lane, max_vf, is_simt.
40245 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40247 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40250 2017-01-20 Martin Jambor <mjambor@suse.cz>
40252 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40253 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40255 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40256 (GTFILES): Rename hsa.c to hsa-common.c.
40257 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40258 * hsa-dump.c: Likewise.
40259 * hsa-gen.c: Likewise.
40260 * hsa-regalloc.c: Likewise.
40261 * ipa-hsa.c: Likewise.
40262 * omp-expand.c: Likewise.
40263 * omp-low.c: Likewise.
40264 * toplev.c: Likewise.
40266 2017-01-20 Marek Polacek <polacek@redhat.com>
40269 * doc/invoke.texi: Document -Wduplicated-branches.
40270 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40271 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40272 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40273 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40274 return 0 only when not OEP_LEXICOGRAPHIC.
40275 (fold_build_cleanup_point_expr): Use the expression
40276 location when building CLEANUP_POINT_EXPR.
40277 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40278 * tree.c (add_expr): Handle error_mark_node.
40280 2017-01-20 Martin Liska <mliska@suse.cz>
40283 * tree-profile.c (init_ic_make_global_vars): Do not call
40285 (gimple_init_gcov_profiler): Likewise.
40287 2017-01-20 Martin Liska <mliska@suse.cz>
40290 * cgraph.h (maybe_create_reference): Remove argument and
40292 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40294 * ipa-cp.c (create_specialized_node): Likewise.
40295 * symtab.c (symtab_node::maybe_create_reference): Handle
40296 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40298 2017-01-20 Martin Liska <mliska@suse.cz>
40300 * read-rtl-function.c (function_reader::create_function): Use
40301 build_decl instread of build_decl_stat.
40303 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40305 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40306 * config/i386/avx512dqintrin.h: Ditto.
40307 * config/i386/avx512fintrin.h: Ditto.
40308 * config/i386/i386-builtin-types.def: Add new types.
40309 * config/i386/i386.c: Handle new types.
40310 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40311 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40312 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40313 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40314 (__builtin_ia32_kshiftridi): New.
40315 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40317 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40321 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40322 define to rs6000_init_stack_protect_guard.
40323 (rs6000_init_stack_protect_guard): New function.
40325 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40326 Yunqiang Su <yunqiang.su@imgtec.com>
40328 * config.gcc (supported_defaults): Add madd4.
40329 (with_madd4): Add validation.
40330 (all_defaults): Add madd4.
40331 * config/mips/mips.opt (mmadd4): New option.
40332 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40334 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40336 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40337 (ISA_HAS_FUSED_MADD4): Likewise.
40338 * doc/invoke.texi (-mmadd4): Document the new option.
40339 * doc/install.texi (--with-madd4): Document the new option.
40341 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40343 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40344 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40345 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40346 (aarch64_init_pauth_hint_builtins): New.
40347 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40348 (aarch64_expand_builtin): Expand new builtins.
40350 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40352 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40353 * combine-stack-adj.c (no_unhandled_cfa): Handle
40354 REG_CFA_TOGGLE_RA_MANGLE.
40355 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40356 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40357 info for return address signing.
40358 (aarch64_expand_epilogue): Likewise.
40360 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40362 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40363 * config/aarch64/aarch64-protos.h
40364 (aarch64_return_address_signing_enabled): New declaration.
40365 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40367 (aarch64_expand_prologue): Sign return address before it's pushed onto
40369 (aarch64_expand_epilogue): Authenticate return address fetched from
40371 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40372 (aarch64_attributes): New function attributes for "sign-return-address".
40373 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40374 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40375 ("*do_return"): Generate combined instructions according to key index.
40376 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40377 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40379 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40380 * config/aarch64/aarch64.opt (msign-return-address=): New.
40381 * doc/extend.texi (AArch64 Function Attributes): Documents
40382 "sign-return-address=".
40383 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40385 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40387 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40388 overall option summary.
40390 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40392 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40393 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40394 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40395 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40397 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40399 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40400 -mpower9-minmax by default for -mcpu=power9.
40401 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40402 128-bit floating point.
40404 2017-01-20 Alan Modra <amodra@gmail.com>
40406 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40407 optimizing for size.
40409 2017-01-20 Alan Modra <amodra@gmail.com>
40412 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40413 for strcmp and strncmp from corresponding builtin decl.
40415 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40417 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40418 instead of i386/rtems-64.h.
40419 * config/i386/rtems-64.h: Remove.
40421 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40425 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40427 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40429 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40431 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40432 Change int to HOST_WIDE_INT.
40433 * config/aarch64/aarch64-protos.h
40434 (aarch64_simd_gen_const_vector_dup): Likewise.
40435 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40437 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40439 * langhooks-def.h (lhd_type_for_size): New decl.
40440 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40441 * langhooks.c (lhd_type_for_size): New function, taken from
40444 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40446 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40447 define_bypass for CR latency.
40448 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40449 (power9-alu2): Add define_bypass for CR latency.
40451 (power9-mul): Update insn latency.
40452 (power9-mul-compare): Update insn latency, bypass latency and remove
40455 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40457 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40459 * config/aarch64/aarch64.md
40460 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40461 aarch64_nopcrelative_literal_loads.
40462 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40464 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40466 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40467 TARGET_LOONGSON_3A.
40468 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40470 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40473 * config.gcc (supported_defaults): Add lxc1-sxc1.
40474 (with_lxc1_sxc1): Add validation.
40475 (all_defaults): Add lxc1-sxc1.
40476 * config/mips/mips.opt (mlxc1-sxc1): New option.
40477 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40479 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40480 __mips_no_lxc1_sxc1.
40481 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40482 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40483 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40485 2017-01-19 Richard Biener <rguenther@suse.de>
40487 PR tree-optimization/72488
40488 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40489 sure to restore SSA info.
40490 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40492 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40494 PR rtl-optimization/79121
40495 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40496 of the inner type when shifting an extended value.
40498 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40501 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40502 interposable aliases.
40504 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40507 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40508 Use the evmergelohi instruction.
40509 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40510 (mov_sitf_e500_subreg8_2_be): Likewise.
40511 (mov_sitf_e500_subreg12_2_le): Likewise.
40512 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40513 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40514 (mov_sitf_e500_subreg8_2_le): Likewise.
40515 (mov_sitf_e500_subreg12_2_be): Likewise.
40517 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40519 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40520 attribute from vecsimple to vecperm.
40521 (altivec_vbpermq2): Likewise.
40523 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40526 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40528 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40529 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40530 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40531 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40532 case where N arg is SIZE_MAX.
40533 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40534 (cmpstrsi): Add pattern.
40536 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40539 __builtin_vec_revb builtins.
40540 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40541 built-in functions to support generation of the ISA 3.0 XXBR<x>
40542 vector byte reverse instructions.
40543 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40544 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40545 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40546 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40547 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40548 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40549 (P9V_BUILTIN_VEC_REVB): Likewise.
40550 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40551 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40552 (p9_xxbrq_v16qi): Likewise.
40553 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40554 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40555 (p9_xxbrh_v8hi): Likewise.
40556 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40557 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40558 vec_revb built-in functions.
40560 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40562 PR rtl-optimization/78952
40563 * config/i386/i386.md (any_extract): New code iterator.
40564 (*insvqi_2): Use any_extract for source operand.
40565 (*insvqi_3): Use any_shiftrt for source operand.
40567 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40569 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40571 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40573 2017-01-18 Matthias Klose <doko@ubuntu.com>
40575 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40577 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40579 * config/rs6000/altivec.h (vec_bperm): Change #define.
40580 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40581 (altivec_vbpermq2): New define_insn.
40582 (altivec_vbpermd): Likewise.
40583 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40584 function interface.
40585 (VBPERMD): Likewise.
40586 (VBPERM): New polymorphic function interface.
40587 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40588 Add entries for P9V_BUILTIN_VEC_VBPERM.
40589 * doc/extend.texi: Add interfaces for vec_bperm.
40591 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40593 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40594 first letter of error messages.
40595 (s390_resolve_overloaded_builtin): Likewise.
40596 * config/s390/s390.c (s390_expand_builtin): Likewise.
40597 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40598 (s390_valid_target_attribute_inner_p): Likewise.
40599 * config/s390/s390.md ("tabort"): Likewise.
40601 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40603 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40604 (ISA_AVOID_DIV_HILO): New macro.
40605 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40606 (ISA_HAS_DDIV): Likewise.
40608 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40610 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40612 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40614 * doc/invoke.texi (fabi-version): Spelling fix.
40616 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40619 * doc/invoke.texi (fabi-version): Mention mangling fix for
40622 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40625 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40627 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40630 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40633 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40635 (stack_protect_test): Ditto.
40636 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40637 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40639 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40640 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40641 -mstack-protector-guard-offset=.
40642 (RS/6000 and PowerPC Options): Ditto.
40644 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40646 * config/i386/i386.h (MASK_CLASS_P): New define.
40647 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40648 there are no registers from different register sets also when
40649 mask registers are used. Update function comment.
40650 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40651 to (*k/*r) and (*k/*km) alternatives.
40653 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40655 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40656 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40657 (EH_RETURN_HANDLER_RTX): New define.
40658 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40659 Force frame pointer in EH return functions.
40660 (aarch64_expand_epilogue): Add barrier for eh_return.
40661 (aarch64_final_eh_return_addr): Remove.
40662 (aarch64_eh_return_handler_rtx): New function.
40663 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40665 (aarch64_eh_return_handler_rtx): New prototype.
40667 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40669 * config/rs6000/altivec.h (vec_rlmi): New #define.
40670 (vec_vrlnm): Likewise.
40671 (vec_rlnm): Likewise.
40672 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40673 (UNSPEC_VRLNM): Likewise.
40674 (VIlong): New mode iterator.
40675 (altivec_vrl<VI_char>mi): New define_insn.
40676 (altivec_vrl<VI_char>nm): Likewise.
40677 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40679 (VRLDNM): Likewise.
40680 (RLNM): New polymorphic function entry.
40681 (VRLWMI): New monomorphic function entry.
40682 (VRLDMI): Likewise.
40683 (RLMI): New polymorphic function entry.
40684 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40685 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40686 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40689 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40692 * dwarf2out.c (field_byte_offset): Restore the
40693 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40694 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40695 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40698 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40701 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40703 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40706 * dwarf2out.c (add_data_member_location_attribute): For constant
40707 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40708 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40709 DW_AT_byte_size attributes.
40711 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40713 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40714 after forcing to constant memory when the code model is medium.
40716 2017-01-17 Julia Koval <julia.koval@intel.com>
40719 * config/i386/avx512fintrin.h
40720 (_mm512_i32gather_ps): Change __addr type to void const*.
40721 (_mm512_mask_i32gather_ps): Ditto.
40722 (_mm512_i32gather_pd): Ditto.
40723 (_mm512_mask_i32gather_pd): Ditto.
40724 (_mm512_i64gather_ps): Ditto.
40725 (_mm512_mask_i64gather_ps): Ditto.
40726 (_mm512_i64gather_pd): Ditto.
40727 (_mm512_mask_i64gather_pd): Ditto.
40728 (_mm512_i32gather_epi32): Ditto.
40729 (_mm512_mask_i32gather_epi32): Ditto.
40730 (_mm512_i32gather_epi64): Ditto.
40731 (_mm512_mask_i32gather_epi64): Ditto.
40732 (_mm512_i64gather_epi32): Ditto.
40733 (_mm512_mask_i64gather_epi32): Ditto.
40734 (_mm512_i64gather_epi64): Ditto.
40735 (_mm512_mask_i64gather_epi64): Ditto.
40736 (_mm512_i32scatter_ps): Change __addr type to void*.
40737 (_mm512_mask_i32scatter_ps): Ditto.
40738 (_mm512_i32scatter_pd): Ditto.
40739 (_mm512_mask_i32scatter_pd): Ditto.
40740 (_mm512_i64scatter_ps): Ditto.
40741 (_mm512_mask_i64scatter_ps): Ditto.
40742 (_mm512_i64scatter_pd): Ditto.
40743 (_mm512_mask_i64scatter_pd): Ditto.
40744 (_mm512_i32scatter_epi32): Ditto.
40745 (_mm512_mask_i32scatter_epi32): Ditto.
40746 (_mm512_i32scatter_epi64): Ditto.
40747 (_mm512_mask_i32scatter_epi64): Ditto.
40748 (_mm512_i64scatter_epi32): Ditto.
40749 (_mm512_mask_i64scatter_epi32): Ditto.
40750 (_mm512_i64scatter_epi64): Ditto.
40751 (_mm512_mask_i64scatter_epi64): Ditto.
40752 * config/i386/avx512pfintrin.h
40753 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40754 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40755 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40756 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40757 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40758 (_mm512_prefetch_i32scatter_ps): Ditto.
40759 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40760 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40761 (_mm512_prefetch_i64scatter_pd): Ditto.
40762 (_mm512_prefetch_i64scatter_ps): Ditto.
40763 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40764 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40765 * config/i386/avx512vlintrin.h
40766 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40767 (_mm_mmask_i32gather_ps): Ditto.
40768 (_mm256_mmask_i32gather_pd): Ditto.
40769 (_mm_mmask_i32gather_pd): Ditto.
40770 (_mm256_mmask_i64gather_ps): Ditto.
40771 (_mm_mmask_i64gather_ps): Ditto.
40772 (_mm256_mmask_i64gather_pd): Ditto.
40773 (_mm_mmask_i64gather_pd): Ditto.
40774 (_mm256_mmask_i32gather_epi32): Ditto.
40775 (_mm_mmask_i32gather_epi32): Ditto.
40776 (_mm256_mmask_i32gather_epi64): Ditto.
40777 (_mm_mmask_i32gather_epi64): Ditto.
40778 (_mm256_mmask_i64gather_epi32): Ditto.
40779 (_mm_mmask_i64gather_epi32): Ditto.
40780 (_mm256_mmask_i64gather_epi64): Ditto.
40781 (_mm_mmask_i64gather_epi64): Ditto.
40782 (_mm256_i32scatter_ps): Change __addr type to void*.
40783 (_mm256_mask_i32scatter_ps): Ditto.
40784 (_mm_i32scatter_ps): Ditto.
40785 (_mm_mask_i32scatter_ps): Ditto.
40786 (_mm256_i32scatter_pd): Ditto.
40787 (_mm256_mask_i32scatter_pd): Ditto.
40788 (_mm_i32scatter_pd): Ditto.
40789 (_mm_mask_i32scatter_pd): Ditto.
40790 (_mm256_i64scatter_ps): Ditto.
40791 (_mm256_mask_i64scatter_ps): Ditto.
40792 (_mm_i64scatter_ps): Ditto.
40793 (_mm_mask_i64scatter_ps): Ditto.
40794 (_mm256_i64scatter_pd): Ditto.
40795 (_mm256_mask_i64scatter_pd): Ditto.
40796 (_mm_i64scatter_pd): Ditto.
40797 (_mm_mask_i64scatter_pd): Ditto.
40798 (_mm256_i32scatter_epi32): Ditto.
40799 (_mm256_mask_i32scatter_epi32): Ditto.
40800 (_mm_i32scatter_epi32): Ditto.
40801 (_mm_mask_i32scatter_epi32): Ditto.
40802 (_mm256_i32scatter_epi64): Ditto.
40803 (_mm256_mask_i32scatter_epi64): Ditto.
40804 (_mm_i32scatter_epi64): Ditto.
40805 (_mm_mask_i32scatter_epi64): Ditto.
40806 (_mm256_i64scatter_epi32): Ditto.
40807 (_mm256_mask_i64scatter_epi32): Ditto.
40808 (_mm_i64scatter_epi32): Ditto.
40809 (_mm_mask_i64scatter_epi32): Ditto.
40810 (_mm256_i64scatter_epi64): Ditto.
40811 (_mm256_mask_i64scatter_epi64): Ditto.
40812 (_mm_i64scatter_epi64): Ditto.
40813 (_mm_mask_i64scatter_epi64): Ditto.
40814 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40815 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40816 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40817 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40818 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40819 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40820 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40821 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40822 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40823 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40824 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40825 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40826 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40827 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40828 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40829 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40830 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40831 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40832 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40833 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40834 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40835 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40836 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40837 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40838 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40839 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40840 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40841 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40842 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40843 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40844 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40845 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40846 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40847 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40848 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40849 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40850 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40851 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40852 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40853 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40854 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40855 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40856 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40857 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40858 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40859 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40860 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40861 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40862 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40863 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40864 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40865 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40866 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40867 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40868 definitions accordingly.
40870 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40871 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40874 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40877 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40880 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40881 hard regs for inner regno.
40883 2017-01-17 Martin Liska <mliska@suse.cz>
40886 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40887 assumption and add comment.
40889 2017-01-17 Nathan Sidwell <nathan@acm.org>
40891 * ipa-visibility.c (localize_node): New function, broken out of ...
40892 (function_and_variable_visibility): ... here. Call it.
40894 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40896 PR middle-end/77445
40897 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40898 correctly set frequency of oudgoing edge.
40899 (duplicate_thread_path): Fix profile updating.
40901 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40904 * configure.ac: Add GCC_BASE_VER.
40905 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40906 version from BASE-VER file.
40907 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40908 (gcc.o): Depend on $(BASEVER).
40909 * common.opt (dumpfullversion): New option.
40910 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40911 * doc/invoke.texi: Document -dumpfullversion.
40912 * doc/install.texi: Document --with-gcc-major-version-only.
40913 * configure: Regenerated.
40915 2017-01-17 Richard Biener <rguenther@suse.de>
40917 PR tree-optimization/71433
40918 * tree-vrp.c (register_new_assert_for): Merge same asserts
40919 on all incoming edges.
40920 (process_assert_insertions_for): Handle insertions at the
40923 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40925 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40926 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40928 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40931 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40934 2017-01-17 Alan Modra <amodra@gmail.com>
40937 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40938 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40939 symbolic stack limit when pic.
40941 2017-01-16 Martin Sebor <msebor@redhat.com>
40943 PR tree-optimization/78608
40944 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40946 2017-01-16 Jeff Law <law@redhat.com>
40949 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40950 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40951 for several include directories that may be relative to sysroot.
40952 * config/i386/x-mingw32 (gplus_includedir): Define.
40953 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40954 (native_system_includedir): Likewise.
40955 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40956 override if TARGET_SYSTEM_ROOT is defined.
40957 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40959 PR tree-optimization/79090
40960 PR tree-optimization/33562
40961 PR tree-optimization/61912
40962 PR tree-optimization/77485
40963 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40964 and computed trims into the dump file.
40966 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40968 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40970 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40973 * gimplify.c (gimplify_init_constructor): If want_value and
40974 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40978 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40979 sequence. Formatting fixes.
40980 (doloop_optimize): Formatting fixes.
40983 * gcc.c (debug_level_greater_than_spec_func): New function.
40984 (static_spec_functions): Add debug-level-gt spec function.
40985 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40987 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40988 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40989 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40990 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40991 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40992 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40994 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40996 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40997 QImode fixups to general and mask registers only.
40999 2017-01-16 Carl Love <cel@us.ibm.com>
41001 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41002 for built-in functions
41003 vector signed char vec_nabs (vector signed char)
41004 vector signed short vec_nabs (vector signed short)
41005 vector signed int vec_nabs (vector signed int)
41006 vector signed long long vec_nabs (vector signed long long)
41007 vector float vec_nabs (vector float)
41008 vector double vec_nabs (vector double)
41009 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41011 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41012 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41013 * doc/extend.texi: Update the documentation file for the new built-in
41016 2017-01-16 Martin Sebor <msebor@redhat.com>
41018 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41021 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41023 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41024 UNSPEC_VSX__XXSPLTD to require special splat handling.
41026 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41029 * system.h: Poison strndup.
41031 2017-01-16 Alan Modra <amodra@gmail.com>
41034 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41037 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41039 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41041 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41043 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41044 call recog here. Assert that INSN_CODE (insn) is non-negative.
41046 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41049 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41051 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41052 in the currently scheduled RTL fragment.
41054 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41056 PR rtl-optimization/78751
41057 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41060 2017-01-14 Jeff Law <law@redhat.com>
41062 PR tree-optimization/79090
41063 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41064 variable length stores.
41065 (compute_trims): Delete dead assignment to *trim_tail.
41066 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41069 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41071 PR rtl-optimization/78626
41072 PR rtl-optimization/78727
41073 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41074 of a block, and split such blocks after everything else is finished.
41076 2017-01-14 Alan Modra <amodra@gmail.com>
41079 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41080 target legitimate_combined_insn.
41081 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41082 (rs6000_legitimate_combined_insn): New function.
41083 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41085 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41086 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41087 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41089 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41091 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41093 2017-01-13 Jeff Law <law@redhat.com>
41095 PR tree-optimization/33562
41096 PR tree-optimization/61912
41097 PR tree-optimization/77485
41098 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41100 (delete_dead_assignment): Likewise.
41101 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41102 statement to delete_dead_call and delete_dead_assignment.
41104 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41107 * substring-locations.c (format_warning_va): Strengthen case 1 so
41108 that both endpoints of the substring must be within the format
41109 range for just the substring to be printed.
41111 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41113 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41114 * config/i386/i386.c (ix86_target_string): Add missing options
41115 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41116 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41117 flags_other and ix86_target_other to flags2_other. Display unknown
41119 (ix86_valid_target_attribute_inner_p): Add missing options and
41120 reorder options by implied ISAs, as in ix86_target_string.
41122 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41124 * hash-table.h (hash_table::too_empty_p): New function.
41125 (hash_table::expand): Use it.
41126 (hash_table::traverse): Likewise.
41127 (hash_table::empty_slot): Use sizeof (value_type) instead of
41128 sizeof (PTR) to convert bytes to elements. Shrink the table
41129 if the current size is excessive for the current number of
41132 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41134 * ira-costs.c (record_reg_classes): Break from the inner loop
41135 early once alt_fail is known to be true. Update outer loop
41136 handling accordingly.
41138 2017-01-13 Jeff Law <law@redhat.com>
41140 * tree-ssa-dse.c (decrement_count): New function.
41141 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41142 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41143 when we know the partially dead statement is a mem* function.
41145 PR tree-optimization/61912
41146 PR tree-optimization/77485
41147 * tree-ssa-dse.c: Include expr.h.
41148 (maybe_trim_constructor_store): New function.
41149 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41151 PR tree-optimization/33562
41152 PR tree-optimization/61912
41153 PR tree-optimization/77485
41154 * doc/invoke.texi: Document new dse-max-object-size param.
41155 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41156 * tree-ssa-dse.c: Include params.h.
41157 (dse_store_status): New enum.
41158 (initialize_ao_ref_for_dse): New, partially extracted from
41160 (valid_ao_ref_for_dse, normalize_ref): New.
41161 (setup_live_bytes_from_ref, compute_trims): Likewise.
41162 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41163 (maybe_trim_partially_dead_store): Likewise.
41164 (maybe_trim_complex_store): Likewise.
41165 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41166 Track what bytes live from the original store. Return tri-state
41167 for dead, partially dead or live.
41168 (dse_dom_walker): Add constructor, destructor and new private members.
41169 (delete_dead_call, delete_dead_assignment): New extracted from
41171 (dse_optimize_stmt): Make a member of dse_dom_walker.
41172 Use initialize_ao_ref_for_dse.
41174 PR tree-optimization/33562
41175 PR tree-optimization/61912
41176 PR tree-optimization/77485
41177 * sbitmap.h (bitmap_count_bits): Prototype.
41178 (bitmap_clear_range, bitmap_set_range): Likewise.
41179 * sbitmap.c (bitmap_clear_range): New function.
41180 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41182 2017-01-13 Martin Liska <mliska@suse.cz>
41185 * function.c (set_cfun): Add new argument force.
41186 * function.h (set_cfun): Likewise.
41187 * ipa-inline-transform.c (inline_call): Use the function when
41188 strict alising from is dropped for function we inline to.
41190 2017-01-13 Richard Biener <rguenther@suse.de>
41192 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41193 for dumping GIMPLE INTEGER_CSTs.
41195 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41197 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41198 to 201112L since C++17.
41200 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41203 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41204 if -fsanitize=kernel-address is present.
41206 2017-01-13 Richard Biener <rguenther@suse.de>
41208 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41209 as _Literal ( type ) number in case usual suffixes do not
41210 preserve all information.
41212 2017-01-13 Richard Biener <rguenther@suse.de>
41214 PR tree-optimization/77283
41215 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41216 and ssa-iterators.h.
41217 (is_feasible_trace): Implement a cost model based on joiner
41220 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41223 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41224 char or short to __float128/_Float128 directly.
41226 2017-01-12 Martin Sebor <msebor@redhat.com>
41228 to -Wformat-overflow.
41229 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41230 (min_bytes_remaining): Same.
41231 (get_string_length): Same.
41232 (format_string): Same.
41233 (format_directive): Same.
41235 (pass_sprintf_length::handle_gimple_call): Same.
41237 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41239 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41240 info.nowrite calls with no lhs that can't throw. Return bool
41241 whether gsi_remove has been called or not.
41242 (pass_sprintf_length::handle_gimple_call): Return bool whether
41243 try_substitute_return_value called gsi_remove. Formatting fix.
41244 (pass_sprintf_length::execute): Don't use gsi_remove if
41245 handle_gimple_call returned true.
41248 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41249 be removed due to side-effects, don't remove following barrier nor
41250 turn the successor edge into fallthru edge.
41252 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41255 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41256 element-reversing loads and stores as not swappable.
41258 2017-01-12 Nathan Sidwell <nathan@acm.org>
41259 Nicolai Stange <nicstange@gmail.com>
41261 * combine.c (try_combine): Don't ignore result of overlap checking
41262 loop. Combine overlap & asm check into single loop.
41264 2017-01-12 Richard Biener <rguenther@suse.de>
41266 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41267 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41269 2017-01-12 Richard Biener <rguenther@suse.de>
41271 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41272 and TS_TARGET_OPTION directly derive from TS_BASE.
41273 * tree-core.h (tree_optimization_option): Derive from tree_base.
41274 (tree_target_option): Likewise.
41276 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41278 * config/i386/i386.c (memory_address_length): Increase len
41279 only when rip_relative_addr_p returns false.
41281 2017-01-11 Julia Koval <julia.koval@intel.com>
41283 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41284 (OPTION_MASK_ISA_SGX_SET): New.
41285 (ix86_handle_option): Handle OPT_msgx.
41286 * config.gcc: Added sgxintrin.h.
41287 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41288 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41289 * config/i386/i386.c (ix86_target_string): Add -msgx.
41291 (ix86_option_override_internal): Handle new options.
41292 (ix86_valid_target_attribute_inner_p): Add sgx.
41293 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41294 * config/i386/i386.opt: Add msgx.
41295 * config/i386/sgxintrin.h: New file.
41296 * config/i386/x86intrin.h: Add sgxintrin.h.
41298 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41301 * fold-const.c (maybe_nonzero_address): Return 1 for function
41303 (tree_single_nonzero_warnv_p): Don't handle function local objects
41307 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41310 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41313 * opts.c: Include "spellcheck.h"
41314 (struct string_fragment): New struct.
41315 (struct edit_distance_traits<const string_fragment &>): New
41317 (get_closest_sanitizer_option): New function.
41318 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41320 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41322 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41324 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41325 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41326 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41327 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41328 for initial die_offset if dwarf_split_debug_info.
41329 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41330 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41332 (output_skeleton_debug_sections): Formatting fix. Use
41333 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41334 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41336 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41338 * config/arm/cortex-a53.md: Add bypasses for
41339 cortex_a53_r2f_cvt.
41340 (cortex_a53_r2f): Only use for transfers.
41341 (cortex_a53_f2r): Likewise.
41342 (cortex_a53_r2f_cvt): Add reservation for conversions.
41343 (cortex_a53_f2r_cvt): Likewise.
41345 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41347 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41348 to all inlined functions, change static to extern.
41350 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41353 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41355 (arm_assemble_integer): Likewise.
41357 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41359 * config.gcc: Use new awk script to check CPU, FPU and architecture
41360 parameters for --with-... options.
41361 * config/arm/parsecpu.awk: New file
41362 * config/arm/arm-cpus.in: New file.
41363 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41365 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41367 * config/arm/t-arm: Update dependency rules.
41368 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41369 of processing .def files.
41370 * config/arm/genopt.sh: Deleted.
41371 * config/arm/gentune.sh: Deleted.
41372 * config/arm/arm-cores.def: Deleted.
41373 * config/arm/arm-arches.def: Deleted.
41374 * config/arm/arm-fpus.def: Deleted.
41375 * config/arm/arm-tune.md: Regenerated.
41376 * config/arm/arm-tables.opt: Regenerated.
41377 * config/arm/arm-cpu.h: New generated file.
41378 * config/arm/arm-cpu-data.h: New generated file.
41379 * config/arm/arm-cpu-cdata.h: New generated file.
41381 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41384 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41386 (input_varpool_node): Unpack dynamically_initialized bit.
41388 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41390 PR rtl-optimization/79032
41391 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41392 the alignment of the adjusted memory reference against that of MODE,
41393 instead of the alignment of the original memory reference.
41395 2017-01-11 Martin Jambor <mjambor@suse.cz>
41397 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41399 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41400 decorated functions.
41402 2017-01-11 Richard Biener <rguenther@suse.de>
41404 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41405 set range/nonnull info for PHI results. Do not set it on
41406 stmts marked for removal.
41408 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41410 * expr.c (store_field): In the bitfield case, fetch the return value
41411 from the registers before applying a single big-endian adjustment.
41412 Always do a final load for a BLKmode value not larger than a word.
41414 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41417 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41418 that we correctly handle column numbers greater than
41419 LINE_MAP_MAX_COLUMN_NUMBER.
41421 2017-01-10 Martin Sebor <msebor@redhat.com>
41423 PR middle-end/78245
41424 * gimple-ssa-sprintf.c (get_destination_size): Call
41425 {init,fini}object_sizes.
41426 * tree-object-size.c (addr_object_size): Adjust.
41427 (pass_through_call): Adjust.
41428 (pass_object_sizes::execute): Adjust.
41429 * tree-object-size.h (fini_object_sizes): Declare.
41431 2017-01-10 Martin Sebor <msebor@redhat.com>
41433 PR tree-optimization/78775
41434 * builtins.c (get_size_range): Move...
41435 * calls.c: ...to here.
41436 (alloc_max_size): Accept zero argument.
41437 (operand_signed_p): Remove.
41438 (maybe_warn_alloc_args_overflow): Call get_size_range.
41439 * calls.h (get_size_range): Declare.
41441 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41443 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41444 from TI's devices.csv file as of September 2016.
41445 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41447 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41449 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41450 * doc/invoke.texi: Likewise.
41451 * doc/md.texi: Likewise.
41452 * doc/objc.texi: Likewise.
41454 2017-01-10 Joshua Conner <joshconner@google.com>
41456 * config/arm/fuchsia-elf.h: New file.
41457 * config/fuchsia.h: New file.
41458 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41459 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41461 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41463 2016-01-10 Richard Biener <rguenther@suse.de>
41465 PR tree-optimization/79034
41466 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41467 Propagate out degenerate PHIs in the joiner.
41469 2017-01-10 Martin Liska <mliska@suse.cz>
41471 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41472 (sort_congruence_classes_by_decl_uid): Likewise.
41473 (sort_congruence_class_groups_by_decl_uid): Likewise.
41474 (sem_item_optimizer::merge_classes): Sort class, groups in these
41475 classes and members in the groups by DECL_UID of declarations.
41476 This would make merge operations stable.
41478 2017-01-10 Martin Liska <mliska@suse.cz>
41480 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41481 usage of m_classes_vec.
41482 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41483 (sem_item_optimizer::get_group_by_hash): Likewise.
41484 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41485 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41486 (sem_item_optimizer::verify_classes): Likewise.
41487 (sem_item_optimizer::process_cong_reduction): Likewise.
41488 (sem_item_optimizer::dump_cong_classes): Likewise.
41489 (sem_item_optimizer::merge_classes): Likewise.
41490 * ipa-icf.h (congruence_class_hash): Rename from
41491 congruence_class_group_hash. Remove declaration of m_classes_vec.
41493 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41495 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41496 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41497 * config.gcc: Add avx512vpopcntdqintrin.h.
41498 * config/i386/avx512vpopcntdqintrin.h: New.
41499 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41500 * config/i386/i386-builtin-types.def: Add new types.
41501 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41502 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41503 __builtin_ia32_vpopcountq_v8di_mask): New.
41504 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41505 __AVX512VPOPCNTDQ__.
41506 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41507 (PTA_AVX512VPOPCNTDQ): Define.
41508 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41509 TARGET_AVX512VPOPCNTDQ_P): Define.
41510 * config/i386/i386.opt: Add mavx512vpopcntdq.
41511 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41512 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41514 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41516 PR middle-end/77484
41517 * predict.def (PRED_CALL): Set to 67.
41519 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41521 * expr.c (store_field): In the bitfield case, if the value comes from
41522 a function call and is of an aggregate type returned in registers, do
41523 not modify the field mode; extract the value in all cases if the mode
41524 is BLKmode and the size is not larger than a word.
41526 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41529 * config/i386/cpuid.h: Fix undefined behavior.
41531 2017-01-04 Jeff Law <law@redhat.com>
41533 PR tree-optimization/79007
41534 PR tree-optimization/67955
41535 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41536 conservative for pt.null when flag_non_call_exceptions is on.
41538 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41540 PR translation/79019
41541 PR translation/79020
41542 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41543 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41544 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41546 * config/avr/avr.opt (maccumulate-args): Likewise.
41547 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41548 * common.opt (freport-bug): Likewise.
41549 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41550 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41551 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41552 translatable string.
41553 * config/i386/i386.c (function_value_32): Likewise.
41554 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41555 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41557 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41558 * common/config/msp430/msp430-common.c (msp430_handle_option):
41560 * symtab.c (symtab_node::verify_base): Likewise.
41561 * opts.c (set_debug_level): Likewise.
41562 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41563 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41564 missing whitespace to translatable strings.
41565 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41566 * config/sh/superh.h: Likewise.
41567 * config/i386/xopintrin.h: Likewise.
41568 * config/i386/znver1.md: Likewise.
41569 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41570 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41571 * double-int.h (struct double_int): Likewise.
41572 * double-int.c (div_and_round_double): Likewise.
41573 * wide-int.cc: Likewise.
41574 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41575 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41576 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41577 (crossjumps_occurred): ... this.
41578 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41581 PR tree-optimization/78899
41582 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41583 returning bool return struct loop *, NULL for failure and the new
41585 (versionable_outer_loop_p): Don't version outer loop if it has
41586 dont_vectorized bit set.
41587 (tree_if_conversion): When versioning outer loop, ensure
41588 tree_if_conversion is performed also on the inner loop of the
41589 non-vectorizable outer loop copy.
41590 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41591 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41592 prevent vectorization of it.
41593 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41594 the outer loop vectorization of the non-scalar version is attempted
41595 before vectorization of the inner loop in scalar version. If
41596 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41597 vectorization of its inner loop.
41598 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41599 has 2 inner loops, rename also on edges from bb whose single pred
41600 is outer_loop->header. Fix typo in function comment.
41602 2017-01-09 Martin Sebor <msebor@redhat.com>
41605 * asan.c (asan_emit_stack_protection): Increase local buffer size
41606 to avoid snprintf truncation warning.
41608 2017-01-09 Andrew Pinski <apinski@cavium.com>
41610 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41611 to reference thunderx2t99 for the tuning structure
41612 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41614 (thunderx2t99_extra_costs): This.
41615 * config/aarch64/aarch64-tune.md: Regenerate.
41616 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41617 (vulcan_addrcost_table): This.
41618 (vulcan_regmove_cost): Rename to ...
41619 (thunderx2t99_regmove_cost): This.
41620 (vulcan_vector_cost): Rename to ...
41621 (thunderx2t99_vector_cost): this.
41622 (vulcan_branch_cost): Rename to ...
41623 (thunderx2t99_branch_cost): This.
41624 (vulcan_tunings): Rename to ...
41625 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41626 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41628 2017-01-09 Martin Jambor <mjambor@suse.cz>
41632 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41633 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41634 (propagate_vr_accross_jump_function): Use the above function for all
41635 value range computations for pass-through jump functions and type
41636 converasion from explicit value range values.
41637 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41638 parameters from TYPE_ARG_TYPES.
41639 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41640 (ipa_write_node_info): Stream type of the actual argument.
41641 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41643 2017-01-09 Martin Liska <mliska@suse.cz>
41646 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41647 (lookup_compiler): Do not show error message with have_E.
41649 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41651 PR tree-optimization/78938
41652 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41653 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41654 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41655 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41658 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41660 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41663 2017-01-09 Richard Biener <rguenther@suse.de>
41665 PR tree-optimization/78997
41666 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41667 name condition properly.
41669 2017-01-09 Richard Biener <rguenther@suse.de>
41672 * dwarf2out.c (is_cxx): New overload with context.
41673 (is_naming_typedef_decl): Use it.
41675 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41677 * invoke.texi (Option Summary): Correct spacing in option lists
41678 and add line breaks to fix over-long lines.
41680 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41682 PR middle-end/17660
41684 * extend.texi (Common Variable Attributes): Add xref to GCC
41685 Internals manual to explain mode attribute keywords.
41687 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41690 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41691 and Preprocessor Options.
41692 (Options for Linking): Document -pthread here....
41693 (RS/6000 and PowerPC Options): ...not here.
41694 (Solaris 2 Options): ...or here.
41695 * doc/cppopts.texi: Document -pthread.
41697 2017-01-08 Martin Sebor <msebor@redhat.com>
41699 PR middle-end/77708
41700 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41701 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41702 New member functions.
41703 (format_directive): Used them.
41705 (pass_sprintf_length::handle_gimple_call): Same.
41706 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41707 to avoid truncation for any argument.
41708 (extract_affine_mul): Same.
41709 * tree.c (get_file_function_name): Same.
41711 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41713 PR middle-end/77484
41714 * predict.def (PRED_INDIR_CALL): Set to 86.
41716 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41718 PR preprocessor/54124
41719 * doc/cppopts.texi: Reformat -d subtable to list the full name
41720 of the options. Add cross-reference to the docs for the general
41721 compiler -d options.
41722 * doc/invoke.texi (Developer Options): Add cross-reference to the
41723 preprocessor-specific -d option documentation.
41725 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41727 PR preprocessor/13498
41728 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41729 redudant material, and reflect new command-line options.
41730 (System Headers): Likewise.
41732 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41734 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41735 -isystem, and -idirafter. Copy-edit.
41736 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41737 default for -ftrack-macro-expansion. Delete obsolete and
41738 badly-formatted implementation details about -fdebug-cpp output.
41739 * doc/cppwarnopts.texi: Copy-edit.
41741 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41744 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41745 that the transition from a max line width >= 1<<10 to narrower
41746 lines works correctly.
41748 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41750 * doc/options.texi (PerFunction): New.
41751 * opt-functions.awk (switch_flags): Map both Optimization and
41752 PerFunction to CL_OPTIMIZATION.
41753 * opth-gen.awk: Test for PerFunction flag along with
41755 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41756 it only when the latter is present. Skip those that don't in
41757 the hash function generator.
41758 * common.opt (fvar-tracking): Mark as PerFunction instead of
41760 (fvar-tracking-assignments): Likewise.
41761 (fvar-tracking-assignments-toggle): Likewise.
41762 (fvar-tracking-uninit): Likewise.
41764 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41766 PR translation/79018
41767 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41770 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41773 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41774 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41775 TARGET_LONG_JUMP_TABLE_OFFSETS.
41776 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41777 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41778 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41779 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41780 * config/m68k/m68k.md (tablejump expander): Likewise.
41781 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41782 TARGET_LONG_JUMP_TABLE_OFFSETS.
41783 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41784 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41786 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41787 David Holsgrove <david.holsgrove@xilinx.com>
41789 * common/config/microblaze/microblaze-common.c
41790 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41791 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41793 * config/microblaze/microblaze.c (microblaze_must_save_register)
41794 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41796 (microblaze_eh_return): New function.
41797 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41798 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41799 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41800 * config/microblaze/microblaze.md (eh_return): New pattern.
41802 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41804 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41805 GCC_DIAGNOSTIC_STRINGIFY): Define.
41807 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41809 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41811 * config/arm/arm.md (<mcrr>): New.
41813 * config/arm/arm.c (arm_arch5te): New.
41814 (arm_option_override): Set arm_arch5te.
41815 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41817 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41818 (arm_mcrr_qualifiers): ... this. New.
41819 (MRRC_QUALIFIERS): Define to...
41820 (arm_mrrc_qualifiers): ... this. New.
41821 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41823 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41824 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41825 (MRRCI, mrrc, MRRC): New.
41826 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41827 VUNSPEC_MRRC2): New.
41829 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41831 * config/arm/arm.md (<mcr>): New.
41833 * config/arm/arm.c (arm_coproc_builtin_available): Add
41834 support for mcr, mrc, mcr2 and mrc2.
41835 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41836 (arm_mcr_qualifiers): ... this. New.
41837 (MRC_QUALIFIERS): Define to ...
41838 (arm_mrc_qualifiers): ... this. New.
41839 (MCR_QUALIFIERS): Define to ...
41840 (arm_mcr_qualifiers): ... this. New.
41841 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41843 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41844 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41845 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41846 VUNSPEC_MRC2): New.
41848 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41850 * config/arm/arm.md (*ldc): New.
41854 * config/arm/arm.c (arm_coproc_builtin_available): Add
41855 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41856 (arm_coproc_ldc_stc_legitimate_address): New.
41857 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41858 'qualifier_const_pointer'.
41859 (LDC_QUALIFIERS): Define to...
41860 (arm_ldc_qualifiers): ... this. New.
41861 (STC_QUALIFIERS): Define to...
41862 (arm_stc_qualifiers): ... this. New.
41863 * config/arm/arm-protos.h
41864 (arm_coproc_ldc_stc_legitimate_address): New.
41865 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41866 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41867 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41868 stc2, stcl, stc2l): New.
41869 * config/arm/constraints.md (Uz): New.
41870 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41871 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41872 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41873 VUNSPEC_STC2L): New.
41875 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41877 * config/arm/arm.md (<cdp>): New.
41878 * config/arm/arm.c (neon_const_bounds): Rename this ...
41879 (arm_const_bounds): ... this.
41880 (arm_coproc_builtin_available): New.
41881 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41882 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41883 (CDP_QUALIFIERS): Define to...
41884 (arm_cdp_qualifiers): ... this. New.
41886 (arm_expand_builtin_args): Add case for 6 arguments.
41887 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41888 (arm_const_bounds): ... this.
41889 (arm_coproc_builtin_available): New.
41890 * config/arm/arm_acle.h (__arm_cdp): New.
41892 * config/arm/arm_acle_builtins.def (cdp): New.
41894 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41895 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41896 'arm_const_bounds'.
41897 * config/arm/types.md (coproc): New.
41898 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41899 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41900 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41901 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41903 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41905 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41906 (UBINOP_QUALIFIERS): New.
41908 (acle_builtin_data): New. Change comment.
41909 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41910 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41911 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41912 arm_acle_builtins.def.
41913 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41914 (arm_init_acle_builtins): New.
41915 (CRC32_BUILTIN): Remove.
41916 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41917 crc32cb, crc32ch and crc32cw.
41918 (arm_init_crc32_builtins): Remove.
41919 (arm_init_builtins): Use arm_init_acle_builtins rather
41920 than arm_init_crc32_builtins.
41921 (arm_expand_acle_builtin): New.
41922 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41923 * config/arm/arm_acle_builtins.def: New.
41925 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41927 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41928 (arm_builtin_datum): ... this.
41929 (arm_init_neon_builtin): Rename to ...
41930 (arm_init_builtin): ... this. Add a new parameters PREFIX
41931 and USE_SIG_IN_NAME.
41932 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41933 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41934 'arm_builtin_datum'.
41935 (arm_init_vfp_builtins): Likewise.
41936 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41937 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41938 (arm_expand_neon_args): Rename to ...
41939 (arm_expand_builtin_args): ... this. Rename builtin_arg
41940 enum values and differentiate between ARG_BUILTIN_MEMORY
41941 and ARG_BUILTIN_NEON_MEMORY.
41942 (arm_expand_neon_builtin_1): Rename to ...
41943 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41944 values, arm_expand_builtin_args and add bool parameter NEON.
41945 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41946 (arm_expand_vfp_builtin): Likewise.
41947 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41949 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41951 PR middle-end/77484
41952 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41953 * predict.c (tree_estimate_probability_bb): Reverse direction of
41954 polymorphic call predictor.
41956 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41958 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41959 (determine_pass_name_match): ...this new function and...
41960 (should_skip_pass_p): ...this new function.
41962 2017-01-06 Nathan Sidwell <nathan@acm.org>
41964 * ipa-visibility.c (function_and_variable_visibility): Reformat
41965 comments and long lines. Remove extrneous if.
41966 * symtab.c (symtab_node::make_decl_local): Fix code format.
41967 (symtab_node::set_section_for_node): Fix comment typo.
41969 2017-01-06 Martin Liska <mliska@suse.cz>
41972 * lra-constraints.c: Rename invariant to lra_invariant.
41973 * predict.c (set_even_probabilities): Initialize e to NULL.
41975 2017-01-05 Martin Sebor <msebor@redhat.com>
41977 PR tree-optimization/78910
41978 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41979 (format_integer): Correct off-by-one error in the handling
41980 of precision with negative numbers in signed conversions..
41982 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41984 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41986 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41988 PR tree-optimization/71016
41989 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41990 factor_out_conditional_conversion. Formatting fix.
41991 (factor_out_conditional_conversion): Add cond_stmt argument.
41992 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41993 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41996 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41998 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41999 read-rtl-function.o, and selftest-rtl.o.
42000 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42001 (selftest::aarch64_test_loading_full_dump): New function.
42002 (selftest::aarch64_run_selftests): New function.
42003 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42004 selftest::aarch64_run_selftests.
42005 * config/i386/i386.c
42006 (selftest::ix86_test_loading_dump_fragment_1): New function.
42007 (selftest::ix86_test_loading_call_insn): New function.
42008 (selftest::ix86_test_loading_full_dump): New function.
42009 (selftest::ix86_test_loading_unspec): New function.
42010 (selftest::ix86_run_selftests): Call the new functions.
42011 * emit-rtl.c (maybe_set_max_label_num): New function.
42012 * emit-rtl.h (maybe_set_max_label_num): New decl.
42013 * function.c (instantiate_decls): Guard call to
42014 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42015 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42017 * gensupport.c (gen_reader::gen_reader): Pass "false"
42018 for new "compact" param of rtx_reader.
42019 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42020 rather than an empty string for NULL strings.
42021 * read-md.c: Potentially include config.h rather than bconfig.h.
42022 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42023 (have_error): New global, copied from errors.c.
42024 (md_reader::read_name): Rename to...
42025 (md_reader::read_name_1): ...this, adding "out_loc" param,
42026 and converting "missing name or number" to returning false, rather
42028 (md_reader::read_name): Reimplement in terms of read_name_1.
42029 (md_reader::read_name_or_nil): New function.
42030 (md_reader::read_string): Handle "(nil)" by returning NULL.
42031 (md_reader::md_reader): Add new param "compact".
42032 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42033 (md_reader::read_file): New method.
42034 * read-md.h (md_reader::md_reader): Add new param "compact".
42035 (md_reader::read_file): New method.
42036 (md_reader::is_compact): New accessor.
42037 (md_reader::read_name): Convert return type from void to file_location.
42038 (md_reader::read_name_or_nil): New decl.
42039 (md_reader::read_name_1): New decl.
42040 (md_reader::m_compact): New field.
42041 (noop_reader::noop_reader): Pass "false" for new "compact" param
42043 (rtx_reader::rtx_reader): Add new "compact" param.
42044 (rtx_reader::read_rtx_operand): Make virtual and convert return
42045 type from void to rtx.
42046 (rtx_reader::read_until): New decl.
42047 (rtx_reader::handle_any_trailing_information): New virtual function.
42048 (rtx_reader::postprocess): New virtual function.
42049 (rtx_reader::finalize_string): New virtual function.
42050 (rtx_reader::m_in_call_function_usage): New field.
42051 (rtx_reader::m_reuse_rtx_by_id): New field.
42052 * read-rtl-function.c: New file.
42053 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42054 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42055 (selftest::verify_three_block_rtl_cfg): New decl.
42056 * read-rtl-function.h: New file.
42057 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42058 For host, include function.h, memmodel.h, and emit-rtl.h.
42059 (one_time_initialization): New function.
42060 (struct compact_insn_name): New struct.
42061 (compact_insn_names): New array.
42062 (find_code): Handle insn codes in compact dumps.
42063 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42064 (bind_subst_iter_and_attr): Likewise.
42065 (add_condition_to_string): Likewise.
42066 (add_condition_to_rtx): Likewise.
42067 (apply_attribute_uses): Likewise.
42068 (add_current_iterators): Likewise.
42069 (apply_iterators): Likewise.
42070 (initialize_iterators): Guard usage of apply_subst_iterator with
42071 #ifdef GENERATOR_FILE.
42072 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42073 (md_reader::read_mapping): Likewise.
42074 (add_define_attr_for_define_subst): Likewise.
42075 (add_define_subst_attr): Likewise.
42076 (read_subst_mapping): Likewise.
42077 (check_code_iterator): Likewise.
42078 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42080 (one_time_initialization): New function.
42081 (rtx_reader::read_until): New method.
42082 (read_flags): New function.
42083 (parse_reg_note_name): New function.
42084 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42085 Handle reuse_rtx ids.
42086 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42087 Add parsing support for RTL dumps, mirroring the special-cases in
42088 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42089 values, and calling handle_any_trailing_information.
42090 (rtx_reader::read_rtx_operand): Convert return type from void
42091 to rtx, returning return_rtx. Handle case 'e'. Call
42092 finalize_string on XSTR and XTMPL fields.
42093 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42094 "(nil)" values were omitted. Call the postprocess vfunc on the
42096 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42097 class ctor. Initialize m_in_call_function_usage. Call
42098 one_time_initialization.
42099 * rtl-tests.c (selftest::test_uncond_jump): Call
42100 set_new_first_and_last_insn.
42101 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42102 * selftest-rtl.c: New file.
42103 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42104 (selftest::get_insn_by_uid): New decl.
42105 * selftest-run-tests.c (selftest::run_tests): Call
42106 read_rtl_function_c_tests.
42107 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42108 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42111 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42113 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42114 operands in a special way. Assert that pos+len <= mode precision.
42116 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42118 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42119 3 argument Alias with unlimited for the negative form.
42120 (fno-vect-cost-model): Removed.
42122 2017-01-05 Martin Liska <mliska@suse.cz>
42124 * hsa-gen.c (gen_hsa_divmod): New function.
42125 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42127 2017-01-05 Martin Liska <mliska@suse.cz>
42130 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42133 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42135 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42136 small constant length operands.
42138 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42140 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42141 between loop iterations.
42143 2017-01-05 Martin Liska <mliska@suse.cz>
42146 * gimplify.c (gimplify_decl_expr): Compare to
42147 asan_poisoned_variables instread of checking flags.
42148 (gimplify_target_expr): Likewise.
42149 (gimplify_expr): Likewise.
42150 (gimplify_function_tree): Conditionally initialize
42151 asan_poisoned_variables.
42153 2017-01-04 Jeff Law <law@redhat.com>
42155 PR tree-optimizatin/78812
42156 * rtl.h (contains_mem_rtx_p): Prototype.
42157 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42158 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42159 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42160 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42161 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42163 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42165 * input.c (assert_char_at_range): Default-initialize actual_range.
42167 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42169 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42170 to match the caller.
42172 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42174 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42175 insns after final jump in test to emit dummy move.
42177 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42179 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42180 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42182 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42184 * multiple_target.c (create_dispatcher_calls): Init e_next.
42185 * tree-ssa-loop-split.c (split_loop): Init border.
42186 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42189 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42194 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42195 (altivec_register_operand): Do not return true if the operand
42196 contains a SUBREG mixing SImode and SFmode.
42197 (vsx_register_operand): Likewise.
42198 (vsx_reg_sfsubreg_ok): New predicate.
42199 (vfloat_operand): Do not return true if the operand contains a
42200 SUBREG mixing SImode and SFmode.
42201 (vint_operand): Likewise.
42202 (vlogical_operand): Likewise.
42203 (gpc_reg_operand): Likewise.
42204 (int_reg_operand): Likewise.
42205 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42206 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42207 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42209 (rs6000_emit_move_si_sf_subreg): New helper function.
42210 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42211 fixup SUBREGs involving SImode and SFmode.
42212 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42213 numbers for the new peephole2 optimization.
42214 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42215 the GLIBC math library that do AND/IOR/XOR operations on single
42216 precision floating point.
42217 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42218 target macros to say whether we need to avoid SUBREGs mixing
42220 (TARGET_ALLOW_SF_SUBREG): Likewise.
42221 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42222 (UNSPEC_SI_FROM_SF): Likewise.
42223 (iorxor): Change spacing.
42224 (and_ior_xor): New iterator for AND, IOR, and XOR.
42225 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42226 (movdi_from_sf_zero_ext): Likewise.
42227 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42228 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42229 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42230 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42231 (fms<mode>4): Likewise.
42232 (fnma<mode>4): Likewise.
42233 (fnms<mode>4): Likewise.
42234 (nfma<mode>4): Likewise.
42235 (nfms<mode>4): Likewise.
42237 2017-01-04 Marek Polacek <polacek@redhat.com>
42240 * doc/invoke.texi: Document -Wpointer-compare.
42242 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42244 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42247 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42248 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42251 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42254 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42255 documentation of the powerpc_popcntb_ok attribute.
42256 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42257 code to issue warning messages if a requested CPU configuration is
42258 not supported by the binary (assembler and loader) toolchain.
42259 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42260 made to define a built-in function that has been disabled.
42261 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42262 made to define a built-in function that has been disabled.
42263 (altivec_init_builtins): Add comment explaining why definition
42264 of the DST built-in functions is not preceded by an assertion
42265 check. Add assertions to prevent ICE if attempts are made to
42266 define an altivec predicate or an abs* built-in function that has
42268 (htm_init_builtins): Add comment explaining why definition of the
42269 htm built-in functions is not preceded by an assertion check.
42271 2017-01-04 Jeff Law <law@redhat.com>
42273 PR tree-optimizatin/67955
42274 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42275 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42276 the points-to solution does not include pt_null. Use DECL_PT_UID
42279 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42281 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42282 Use gen_int_mode instead of gen_lopwart for const_int operands.
42284 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42286 PR tree-optimization/71563
42287 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42288 out of range value - has low bits known to be zero.
42290 2017-01-04 Alan Modra <amodra@gmail.com>
42292 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42293 * configure: Regenerate.
42294 * config.in: Regenerate.
42296 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42299 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42300 a substring of the message, but strcmp with the whole message. Ifdef
42301 ENABLE_NLS, translate the message first using dgettext.
42303 2017-01-03 Jeff Law <law@redhat.com>
42305 PR tree-optimizatin/78856
42306 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42307 (mark_threaded_blocks): Remove code to truncate thread paths that
42308 cross multiple loop headers. Instead invalidate the cached loop
42309 iteration information and handle case of a thread path walking
42310 into an irreducible region.
42312 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42315 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42316 assertions. Add support for doing the signbit if the IEEE 128-bit
42317 floating point value is in a GPR.
42318 * config/rs6000/rs6000.md (Fsignbit): Delete.
42319 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42320 Update the length attribute if the value is in a GPR.
42321 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42322 the sign or zero extension instruction, since the value is always 0/1.
42323 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42326 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42327 extracting SImode to a GPR register so that we can generate a
42328 store, limit the vector to be in a traditional Altivec register
42329 for the vextuwrx instruction.
42331 2017-01-03 Ian Lance Taylor <iant@google.com>
42333 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42335 2017-01-03 Martin Sebor <msebor@redhat.com>
42337 PR tree-optimization/78696
42338 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42339 precision. Use MPFR for %f for greater fidelity. Correct handling
42341 (pass_sprintf_length::compute_format_length): Set width and precision
42342 specified by asrerisk to void_node for vararg functions.
42343 (try_substitute_return_value): Adjust dump output.
42345 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42347 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42349 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42351 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42352 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42353 -mlra/-mno-lra was passed to the compiler.
42355 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42357 PR rtl-optimization/65618
42358 * emit-rtl.c (try_split): Move initialization of "before" and
42359 "after" to just before the call to emit_insn_after_setloc.
42361 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42363 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42365 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42367 * dwarf2out.c (gen_enumeration_type_die): When
42368 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42370 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42372 PR tree-optimization/78965
42373 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42374 Change first argument from const call_info & to call_info &. For %n
42375 set info.nowrite to false.
42377 PR middle-end/78901
42378 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42379 possibly throwing calls.
42381 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42382 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42383 and fns handling, rather than in a separate case SSA_NAME.
42385 2017-01-02 Jeff Law <law@redhat.com>
42387 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42388 fixes for first_period and second_period variables.
42390 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42393 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42394 (*insvqi_1): New insn pattern.
42395 (*insvqi_1_mem_rex64): Ditto.
42396 (*insvqi_2): Ditto.
42397 (*insvqi_3): Rename from *insvqi.
42399 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42401 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42403 * doc/cfg.texi (Edges): Remove reference to Java.
42404 (Maintaining the CFG): Ditto.
42406 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42408 PR middle-end/77674
42409 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42410 transparent aliases.
42412 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42414 PR middle-end/77484
42415 * predict.def (PRED_CALL): Update hitrate.
42416 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42417 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42418 into direct/indirect/polymorphic variants.
42420 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42422 Update copyright years.
42424 * gcc.c (process_command): Update copyright notice dates.
42425 * gcov-dump.c (print_version): Ditto.
42426 * gcov.c (print_version): Ditto.
42427 * gcov-tool.c (print_version): Ditto.
42428 * gengtype.c (create_file): Ditto.
42429 * doc/cpp.texi: Bump @copying's copyright year.
42430 * doc/cppinternals.texi: Ditto.
42431 * doc/gcc.texi: Ditto.
42432 * doc/gccint.texi: Ditto.
42433 * doc/gcov.texi: Ditto.
42434 * doc/install.texi: Ditto.
42435 * doc/invoke.texi: Ditto.
42437 Copyright (C) 2017 Free Software Foundation, Inc.
42439 Copying and distribution of this file, with or without modification,
42440 are permitted in any medium without royalty provided the copyright
42441 notice and this notice are preserved.