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-vectorizer.h (vect_nunits_for_cost): New function.
6 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
7 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
8 (vect_analyze_slp_cost): Likewise.
9 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
10 (vect_model_load_cost): Likewise.
12 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
13 Alan Hayward <alan.hayward@arm.com>
14 David Sherwood <david.sherwood@arm.com>
16 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
17 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
18 from an unsigned int * to a poly_uint64_pod *.
19 (calculate_unrolling_factor): New function.
20 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
22 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23 Alan Hayward <alan.hayward@arm.com>
24 David Sherwood <david.sherwood@arm.com>
26 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
27 from an unsigned int to a poly_uint64.
28 (_loop_vec_info::slp_unrolling_factor): Likewise.
29 (_loop_vec_info::vectorization_factor): Change from an int
31 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
32 (vect_get_num_vectors): New function.
33 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
34 (vect_get_num_copies): Use vect_get_num_vectors.
35 (vect_analyze_data_ref_dependences): Change max_vf from an int *
37 (vect_analyze_data_refs): Change min_vf from an int * to a
39 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
40 than an unsigned HOST_WIDE_INT.
41 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
42 (vect_analyze_data_ref_dependence): Change max_vf from an int *
44 (vect_analyze_data_ref_dependences): Likewise.
45 (vect_compute_data_ref_alignment): Handle polynomial vf.
46 (vect_enhance_data_refs_alignment): Likewise.
47 (vect_prune_runtime_alias_test_list): Likewise.
48 (vect_shift_permute_load_chain): Likewise.
49 (vect_supportable_dr_alignment): Likewise.
50 (dependence_distance_ge_vf): Take the vectorization factor as a
51 poly_uint64 rather than an unsigned HOST_WIDE_INT.
52 (vect_analyze_data_refs): Change min_vf from an int * to a
54 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
55 vfm1 as a poly_uint64 rather than an int. Make the same change
56 for the returned bound_scalar.
57 (vect_gen_vector_loop_niters): Handle polynomial vf.
58 (vect_do_peeling): Likewise. Update call to
59 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
60 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
62 * tree-vect-loop.c (vect_determine_vectorization_factor)
63 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
64 (vect_get_known_peeling_cost): Likewise.
65 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
66 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
67 (vect_transform_loop): Likewise. Use the lowest possible VF when
68 updating the upper bounds of the loop.
69 (vect_min_worthwhile_factor): Make static. Return an unsigned int
71 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
72 polynomial unroll factors.
73 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
74 (vect_make_slp_decision): Likewise.
75 (vect_supported_load_permutation_p): Likewise, and polynomial
77 (vect_analyze_slp_cost): Handle polynomial vf.
78 (vect_slp_analyze_node_operations): Likewise.
79 (vect_slp_analyze_bb_1): Likewise.
80 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
81 than an unsigned HOST_WIDE_INT.
82 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
83 (vectorizable_load): Handle polynomial vf.
84 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
86 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
88 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
89 Alan Hayward <alan.hayward@arm.com>
90 David Sherwood <david.sherwood@arm.com>
92 * match.pd: Handle bit operations involving three constants
93 and try to fold one pair.
95 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
97 * tree-vect-loop-manip.c: Include gimple-fold.h.
98 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
99 niters_maybe_zero parameters. Handle other cases besides a step of 1.
100 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
101 Add a path that uses a step of VF instead of 1, but disable it
103 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
104 and niters_no_overflow parameters. Update calls to
105 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
106 Create a new SSA name if the latter choses to use a ste other
107 than zero, and return it via niters_vector_mult_vf_var.
108 * tree-vect-loop.c (vect_transform_loop): Update calls to
109 vect_do_peeling, vect_gen_vector_loop_niters and
110 slpeel_make_loop_iterate_ntimes.
111 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
112 (vect_gen_vector_loop_niters): Update declarations after above changes.
114 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
116 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
117 128-bit round to integer instructions.
118 (ceil<mode>2): Likewise.
119 (btrunc<mode>2): Likewise.
120 (round<mode>2): Likewise.
122 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
124 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
125 unaligned VSX load/store on P8/P9.
126 (expand_block_clear): Allow the use of unaligned VSX
129 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
131 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
133 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
134 swap associated with both a load and a store.
136 2018-01-02 Andrew Waterman <andrew@sifive.com>
138 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
139 * config/riscv/riscv.md (clear_cache): Use it.
141 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
143 * web.c: Remove out-of-date comment.
145 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
147 * expr.c (fixup_args_size_notes): Check that any existing
148 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
149 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
150 (emit_single_push_insn): ...here.
152 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
154 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
155 (const_vector_encoded_nelts): New function.
156 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
157 (const_vector_int_elt, const_vector_elt): Declare.
158 * emit-rtl.c (const_vector_int_elt_1): New function.
159 (const_vector_elt): Likewise.
160 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
163 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
165 * expr.c: Include rtx-vector-builder.h.
166 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
167 directly on the tree encoding.
168 (const_vector_from_tree): Likewise.
169 * optabs.c: Include rtx-vector-builder.h.
170 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
171 sequence of "u" values.
172 * vec-perm-indices.c: Include rtx-vector-builder.h.
173 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
174 directly on the vec_perm_indices encoding.
176 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
178 * doc/rtl.texi (const_vector): Describe new encoding scheme.
179 * Makefile.in (OBJS): Add rtx-vector-builder.o.
180 * rtx-vector-builder.h: New file.
181 * rtx-vector-builder.c: Likewise.
182 * rtl.h (rtx_def::u2): Add a const_vector field.
183 (CONST_VECTOR_NPATTERNS): New macro.
184 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
185 (CONST_VECTOR_DUPLICATE_P): Likewise.
186 (CONST_VECTOR_STEPPED_P): Likewise.
187 (CONST_VECTOR_ENCODED_ELT): Likewise.
188 (const_vec_duplicate_p): Check for a duplicated vector encoding.
189 (unwrap_const_vec_duplicate): Likewise.
190 (const_vec_series_p): Check for a non-duplicated vector encoding.
191 Say that the function only returns true for integer vectors.
192 * emit-rtl.c: Include rtx-vector-builder.h.
193 (gen_const_vec_duplicate_1): Delete.
194 (gen_const_vector): Call gen_const_vec_duplicate instead of
195 gen_const_vec_duplicate_1.
196 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
197 (gen_const_vec_duplicate): Use rtx_vector_builder.
198 (gen_const_vec_series): Likewise.
199 (gen_rtx_CONST_VECTOR): Likewise.
200 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
201 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
202 Build a new vector rather than modifying a CONST_VECTOR in-place.
203 (handle_special_swappables): Update call accordingly.
204 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
205 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
206 Build a new vector rather than modifying a CONST_VECTOR in-place.
207 (handle_special_swappables): Update call accordingly.
209 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
211 * simplify-rtx.c (simplify_const_binary_operation): Use
212 CONST_VECTOR_ELT instead of XVECEXP.
214 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
216 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
217 the selector elements to be different from the data elements
218 if the selector is a VECTOR_CST.
219 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
220 ssizetype for the selector.
222 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
224 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
225 before testing each element individually.
226 * tree-vect-generic.c (lower_vec_perm): Likewise.
228 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
230 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
231 * selftest-run-tests.c (selftest::run_tests): Call it.
232 * vector-builder.h (vector_builder::operator ==): New function.
233 (vector_builder::operator !=): Likewise.
234 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
235 (vec_perm_indices::all_from_input_p): New function.
236 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
237 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
238 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
239 instead of reading the VECTOR_CST directly. Detect whether both
240 vector inputs are the same before constructing the vec_perm_indices,
241 and update the number of inputs argument accordingly. Use the
242 utility functions added above. Only construct sel2 if we need to.
244 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
246 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
247 the broadcast of the low byte.
248 (expand_mult_highpart): Use an explicit encoding for the permutes.
249 * optabs-query.c (can_mult_highpart_p): Likewise.
250 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
251 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
252 (vectorizable_bswap): Likewise.
253 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
254 explicit encoding for the power-of-2 permutes.
255 (vect_permute_store_chain): Likewise.
256 (vect_grouped_load_supported): Likewise.
257 (vect_permute_load_chain): Likewise.
259 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
261 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
262 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
263 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
264 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
265 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
266 (vect_gen_perm_mask_any): Likewise.
268 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
270 * int-vector-builder.h: New file.
271 * vec-perm-indices.h: Include int-vector-builder.h.
272 (vec_perm_indices): Redefine as an int_vector_builder.
273 (auto_vec_perm_indices): Delete.
274 (vec_perm_builder): Redefine as a stand-alone class.
275 (vec_perm_indices::vec_perm_indices): New function.
276 (vec_perm_indices::clamp): Likewise.
277 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
278 (vec_perm_indices::new_vector): New function.
279 (vec_perm_indices::new_expanded_vector): Update for new
280 vec_perm_indices class.
281 (vec_perm_indices::rotate_inputs): New function.
282 (vec_perm_indices::all_in_range_p): Operate directly on the
283 encoded form, without computing elided elements.
284 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
285 encoding. Update for new vec_perm_indices class.
286 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
287 the given vec_perm_builder.
288 (expand_vec_perm_var): Update vec_perm_builder constructor.
289 (expand_mult_highpart): Use vec_perm_builder instead of
290 auto_vec_perm_indices.
291 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
292 vec_perm_indices instead of auto_vec_perm_indices. Use a single
293 or double series encoding as appropriate.
294 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
295 vec_perm_indices instead of auto_vec_perm_indices.
296 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
297 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
298 (vect_permute_store_chain): Likewise.
299 (vect_grouped_load_supported): Likewise.
300 (vect_permute_load_chain): Likewise.
301 (vect_shift_permute_load_chain): Likewise.
302 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
303 (vect_transform_slp_perm_load): Likewise.
304 (vect_schedule_slp_instance): Likewise.
305 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
306 (vectorizable_mask_load_store): Likewise.
307 (vectorizable_bswap): Likewise.
308 (vectorizable_store): Likewise.
309 (vectorizable_load): Likewise.
310 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
311 vec_perm_indices instead of auto_vec_perm_indices. Use
312 tree_to_vec_perm_builder to read the vector from a tree.
313 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
314 vec_perm_builder instead of a vec_perm_indices.
315 (have_whole_vector_shift): Use vec_perm_builder and
316 vec_perm_indices instead of auto_vec_perm_indices. Leave the
317 truncation to calc_vec_perm_mask_for_shift.
318 (vect_create_epilog_for_reduction): Likewise.
319 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
320 from auto_vec_perm_indices to vec_perm_indices.
321 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
322 instead of changing individual elements.
323 (aarch64_vectorize_vec_perm_const): Use new_vector to install
324 the vector in d.perm.
325 * config/arm/arm.c (expand_vec_perm_d::perm): Change
326 from auto_vec_perm_indices to vec_perm_indices.
327 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
328 instead of changing individual elements.
329 (arm_vectorize_vec_perm_const): Use new_vector to install
330 the vector in d.perm.
331 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
332 Update vec_perm_builder constructor.
333 (rs6000_expand_interleave): Likewise.
334 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
335 (rs6000_expand_interleave): Likewise.
337 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
339 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
340 to qimode could truncate the indices.
341 * optabs.c (expand_vec_perm_var): Likewise.
343 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
345 * Makefile.in (OBJS): Add vec-perm-indices.o.
346 * vec-perm-indices.h: New file.
347 * vec-perm-indices.c: Likewise.
348 * target.h (vec_perm_indices): Replace with a forward class
350 (auto_vec_perm_indices): Move to vec-perm-indices.h.
351 * optabs.h: Include vec-perm-indices.h.
352 (expand_vec_perm): Delete.
353 (selector_fits_mode_p, expand_vec_perm_var): Declare.
354 (expand_vec_perm_const): Declare.
355 * target.def (vec_perm_const_ok): Replace with...
356 (vec_perm_const): ...this new hook.
357 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
358 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
359 * doc/tm.texi: Regenerate.
360 * optabs.def (vec_perm_const): Delete.
361 * doc/md.texi (vec_perm_const): Likewise.
362 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
363 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
364 expand_vec_perm for constant permutation vectors. Assert that
365 the mode of variable permutation vectors is the integer equivalent
366 of the mode that is being permuted.
367 * optabs-query.h (selector_fits_mode_p): Declare.
368 * optabs-query.c: Include vec-perm-indices.h.
369 (selector_fits_mode_p): New function.
370 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
371 is defined, instead of checking whether the vec_perm_const_optab
372 exists. Use targetm.vectorize.vec_perm_const instead of
373 targetm.vectorize.vec_perm_const_ok. Check whether the indices
374 fit in the vector mode before using a variable permute.
375 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
376 vec_perm_indices instead of an rtx.
377 (expand_vec_perm): Replace with...
378 (expand_vec_perm_const): ...this new function. Take the selector
379 as a vec_perm_indices rather than an rtx. Also take the mode of
380 the selector. Update call to shift_amt_for_vec_perm_mask.
381 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
382 Use vec_perm_indices::new_expanded_vector to expand the original
383 selector into bytes. Check whether the indices fit in the vector
384 mode before using a variable permute.
385 (expand_vec_perm_var): Make global.
386 (expand_mult_highpart): Use expand_vec_perm_const.
387 * fold-const.c: Includes vec-perm-indices.h.
388 * tree-ssa-forwprop.c: Likewise.
389 * tree-vect-data-refs.c: Likewise.
390 * tree-vect-generic.c: Likewise.
391 * tree-vect-loop.c: Likewise.
392 * tree-vect-slp.c: Likewise.
393 * tree-vect-stmts.c: Likewise.
394 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
396 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
397 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
398 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
399 (aarch64_vectorize_vec_perm_const): ...this new function.
400 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
401 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
402 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
403 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
404 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
405 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
406 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
408 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
409 check for NEON modes.
410 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
411 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
412 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
413 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
415 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
416 the old VEC_PERM_CONST conditions.
417 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
418 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
419 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
420 (ia64_vectorize_vec_perm_const_ok): Merge into...
421 (ia64_vectorize_vec_perm_const): ...this new function.
422 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
423 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
424 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
425 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
426 * config/mips/mips.c (mips_expand_vec_perm_const)
427 (mips_vectorize_vec_perm_const_ok): Merge into...
428 (mips_vectorize_vec_perm_const): ...this new function.
429 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
430 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
431 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
432 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
433 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
434 (rs6000_expand_vec_perm_const): Delete.
435 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
437 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
438 (altivec_expand_vec_perm_const_le): Take each operand individually.
439 Operate on constant selectors rather than rtxes.
440 (altivec_expand_vec_perm_const): Likewise. Update call to
441 altivec_expand_vec_perm_const_le.
442 (rs6000_expand_vec_perm_const): Delete.
443 (rs6000_vectorize_vec_perm_const_ok): Delete.
444 (rs6000_vectorize_vec_perm_const): New function.
445 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
446 an element count and rtx array.
447 (rs6000_expand_extract_even): Update call accordingly.
448 (rs6000_expand_interleave): Likewise.
449 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
450 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
451 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
452 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
453 (rs6000_expand_vec_perm_const): Delete.
454 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
455 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
456 (altivec_expand_vec_perm_const_le): Take each operand individually.
457 Operate on constant selectors rather than rtxes.
458 (altivec_expand_vec_perm_const): Likewise. Update call to
459 altivec_expand_vec_perm_const_le.
460 (rs6000_expand_vec_perm_const): Delete.
461 (rs6000_vectorize_vec_perm_const_ok): Delete.
462 (rs6000_vectorize_vec_perm_const): New function. Remove stray
463 reference to the SPE evmerge intructions.
464 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
465 an element count and rtx array.
466 (rs6000_expand_extract_even): Update call accordingly.
467 (rs6000_expand_interleave): Likewise.
468 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
469 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
471 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
473 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
475 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
476 vector mode and that that mode matches the mode of the data
478 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
479 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
480 directly using expand_vec_perm_1 when forcing selectors into
482 (expand_vec_perm_var): New function, split out from expand_vec_perm.
484 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
486 * optabs-query.h (can_vec_perm_p): Delete.
487 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
488 * optabs-query.c (can_vec_perm_p): Split into...
489 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
490 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
491 particular selector is valid.
492 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
493 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
494 (vect_grouped_load_supported): Likewise.
495 (vect_shift_permute_load_chain): Likewise.
496 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
497 (vect_transform_slp_perm_load): Likewise.
498 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
499 (vectorizable_bswap): Likewise.
500 (vect_gen_perm_mask_checked): Likewise.
501 * fold-const.c (fold_ternary_loc): Likewise. Don't take
502 implementations of variable permutation vectors into account
503 when deciding which selector to use.
504 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
505 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
506 with a false third argument.
507 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
508 to test whether the constant selector is valid and can_vec_perm_var_p
509 to test whether a variable selector is valid.
511 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
513 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
514 * optabs-query.c (can_vec_perm_p): Likewise.
515 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
516 instead of vec_perm_indices.
517 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
518 (vect_gen_perm_mask_checked): Likewise,
519 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
520 (vect_gen_perm_mask_checked): Likewise,
522 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
524 * optabs-query.h (qimode_for_vec_perm): Declare.
525 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
526 (qimode_for_vec_perm): ...this new function.
527 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
529 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
531 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
532 does not have a conditional at the top.
534 2018-01-02 Richard Biener <rguenther@suse.de>
536 * ipa-inline.c (big_speedup_p): Fix expression.
538 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
541 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
544 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
548 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
549 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
550 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
551 cond_taken_branch_cost 3->4.
553 2018-01-01 Jakub Jelinek <jakub@redhat.com>
555 PR tree-optimization/83581
556 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
557 TODO_cleanup_cfg if any changes have been made.
560 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
561 convert_modes if target mode has the right side, but different mode
565 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
566 last argument when extracting from CONCAT. If either from_real or
567 from_imag is NULL, use expansion through memory. If result is not
568 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
569 the parts directly to inner mode, if even that fails, use expansion
573 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
574 check for bswap in mode rather than HImode and use that in expand_unop
577 2017-12-31 Jakub Jelinek <jakub@redhat.com>
580 * config/i386/i386.c (ix86_attribute_table): Remove excess
581 initializer for "shared" attribute.
583 2017-12-30 Tom de Vries <tom@codesourcery.com>
586 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
588 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
589 offload_funcs entries that no longer have a corresponding cgraph_node.
590 Mark the remaining ones as DECL_PRESERVE_P.
591 (output_lto): Call prune_offload_funcs.
593 2017-12-30 Jakub Jelinek <jakub@redhat.com>
595 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
596 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
597 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
598 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
599 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
600 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
601 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
602 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
603 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
604 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
605 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
606 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
607 vaesenc_<mode>, vpclmulqdq_<mode>,
608 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
610 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
612 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
613 _Float<N> and _Float<N>X variants for rounding built-in
615 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
616 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
617 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
618 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
619 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
620 * builtins.c (mathfn_built_in_2): Likewise.
621 * internal-fn.def (CEIL): Likewise.
623 (NEARBYINT): Likewise.
627 * convert.c (convert_to_integer_1): Likewise.
628 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
629 (integer_valued_real_call_p): Likewise.
630 * fold-const-call.c (fold_const_call_ss): Likewise.
631 * gencfn-macros.c (print_case_cfn): Change CFN and operator
632 printers to take a const char * suffix instead of a bool.
633 (print_define_operator_list): Likewise.
634 (fltall_suffixes): New list of suffixes, that include the
635 traditional suffixes as well as all of the _Float<N> and
637 (main): For _Float<N> and _Float<N>X functions, emit both
638 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
639 has the _Float<N> and _Float<N>X case names or operators. The
640 <name>_ALL variant has both the traditional and the
641 _Float<N>/_Float<N>X case names or operators.
642 * match.pd (COPYSIGN optimizations): Provide optimizations for
643 _Float<N> and _Float<N>X types where possible.
644 (MIN/MAX optimizations): Likewise.
645 (sqrt optimizations): Likewise.
646 (rounding optimizations): Likewise.
648 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
650 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
651 gen_int_mode rather than GEN_INT.
652 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
654 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
656 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
657 instead of CONSTANT_P.
658 (gen_vec_series): Likewise.
659 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
661 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
663 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
666 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
668 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
669 (aarch64_ld1x2<VDC:mode>): Likewise.
670 (aarch64_simd_ld1<mode>_x2): Likewise.
671 (aarch64_simd_ld1<mode>_x2): Likewise.
672 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
673 (vld1_s8_x2): Likewise.
674 (vld1_u16_x2): Likewise.
675 (vld1_s16_x2): Likewise.
676 (vld1_u32_x2): Likewise.
677 (vld1_s32_x2): Likewise.
678 (vld1_u64_x2): Likewise.
679 (vld1_s64_x2): Likewise.
680 (vld1_f16_x2): Likewise.
681 (vld1_f32_x2): Likewise.
682 (vld1_f64_x2): Likewise.
683 (vld1_p8_x2): Likewise.
684 (vld1_p16_x2): Likewise.
685 (vld1_p64_x2): Likewise.
686 (vld1q_u8_x2): Likewise.
687 (vld1q_s8_x2): Likewise.
688 (vld1q_u16_x2): Likewise.
689 (vld1q_s16_x2): Likewise.
690 (vld1q_u32_x2): Likewise.
691 (vld1q_s32_x2): Likewise.
692 (vld1q_u64_x2): Likewise.
693 (vld1q_s64_x2): Likewise.
694 (vld1q_f16_x2): Likewise.
695 (vld1q_f32_x2): Likewise.
696 (vld1q_f64_x2): Likewise.
697 (vld1q_p8_x2): Likewise.
698 (vld1q_p16_x2): Likewise.
699 (vld1q_p64_x2): Likewise.
701 2017-12-27 Martin Liska <mliska@suse.cz>
703 PR tree-optimization/83552
704 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
705 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
707 2017-12-27 Tom de Vries <tom@codesourcery.com>
709 * config/nvptx/nvptx.c (nvptx_option_override): Disable
710 -gstatement-frontiers.
712 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
714 PR rtl-optimization/83513
715 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
716 before priority comparison.
718 2017-12-25 Jakub Jelinek <jakub@redhat.com>
721 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
722 ix86_isa_flags2 to ix86_isa_flags.
723 * config/i386/i386-c.c (ix86_target_macros_internal): Test
724 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
725 isa_flags rather than isa_flags2.
726 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
727 and -mavx512bitalg from isa2_opts to isa_opts.
728 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
729 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
730 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
732 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
733 OPTION_MASK_ISA_AVX512F ored with another option similarly to
734 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
735 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
736 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
737 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
738 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
739 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
740 builtins from bdesc_args2 to bdesc_args section.
741 (__builtin_ia32_compressstoreuqi512_mask,
742 __builtin_ia32_compressstoreuhi512_mask,
743 __builtin_ia32_compressstoreuqi256_mask,
744 __builtin_ia32_expandloadqi512_mask,
745 __builtin_ia32_expandloadqi512_maskz,
746 __builtin_ia32_expandloadhi512_mask,
747 __builtin_ia32_expandloadhi512_maskz,
748 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
749 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
750 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
751 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
752 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
753 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
754 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
755 __builtin_ia32_vpshldv_v32hi_maskz,
756 __builtin_ia32_vpopcountb_v64qi_mask,
757 __builtin_ia32_vpopcountw_v32hi_mask,
758 __builtin_ia32_vpshufbitqmb512_mask,
759 __builtin_ia32_vpshufbitqmb256_mask): Add
760 " | OPTION_MASK_ISA_AVX512BW".
761 (__builtin_ia32_expandloadqi256_mask,
762 __builtin_ia32_expandloadqi256_maskz,
763 __builtin_ia32_vpopcountb_v32qi_mask): Add
764 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
765 (__builtin_ia32_expandloadhi256_mask,
766 __builtin_ia32_expandloadhi256_maskz,
767 __builtin_ia32_expandloadqi128_mask,
768 __builtin_ia32_expandloadqi128_maskz,
769 __builtin_ia32_expandloadhi128_mask,
770 __builtin_ia32_expandloadhi128_maskz,
771 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
772 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
773 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
774 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
775 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
776 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
777 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
778 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
779 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
780 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
781 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
782 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
783 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
784 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
785 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
786 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
787 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
788 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
789 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
790 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
791 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
792 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
793 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
794 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
795 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
796 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
797 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
798 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
799 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
800 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
801 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
802 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
803 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
804 __builtin_ia32_vpopcountw_v8hi_mask): Add
805 " | OPTION_MASK_ISA_AVX512VL".
806 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
807 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
808 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
809 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
810 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
811 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
812 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
813 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
814 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
815 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
816 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
817 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
818 _mm_mask_bitshuffle_epi64_mask): Likewise.
819 * common/config/i386/i386-common.c
820 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
821 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
822 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
823 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
824 OPTION_MASK_ISA_AVX512BITALG_UNSET.
825 (OPTION_MASK_ISA2_AVX512F_UNSET,
826 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
827 (ix86_handle_option): For -mno-general-regs-only, clear from
828 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
829 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
830 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
831 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
832 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
833 for moving from ix86_isa_flags2 to ix86_isa_flags.
835 2017-12-23 Jakub Jelinek <jakub@redhat.com>
838 * fold-const.c (struct contains_label_data): New type.
839 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
840 inside of a SWITCH_BODY seen during the walk.
841 (contains_label_p): Use walk_tree instead of
842 walk_tree_without_duplicates, prepare data for contains_label_1 and
845 2017-12-22 Jakub Jelinek <jakub@redhat.com>
848 * tree-iterator.c (alloc_stmt_list): Start with cleared
849 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
853 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
854 enable avx512f explicitly in #pragma GCC target.
855 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
856 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
857 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
858 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
859 __builtin_ia32_vpdpbusds_v8si_mask,
860 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
861 __builtin_ia32_vpdpbusds_v4si_mask,
862 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
863 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
864 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
865 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
866 __builtin_ia32_vpdpwssds_v8si_mask,
867 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
868 __builtin_ia32_vpdpwssds_v4si_mask,
869 __builtin_ia32_vpdpwssds_v4si_maskz): Use
870 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
871 just OPTION_MASK_ISA_AVX512VNNI.
873 2017-12-22 Martin Jambor <mjambor@suse.cz>
876 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
879 2017-12-22 Julia Koval <julia.koval@intel.com>
880 Sebastian Peryt <sebastian.peryt@intel.com>
882 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
883 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
884 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
885 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
886 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
887 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
888 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
889 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
890 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
891 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
892 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
893 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
894 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
895 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
896 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
897 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
898 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
899 _mm256_maskz_popcnt_epi64): New intrinsics.
900 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
901 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
902 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
903 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
904 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
905 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
906 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
907 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
908 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
909 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
910 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
911 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
912 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
913 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
914 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
915 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
916 __builtin_ia32_vpshufbitqmb256_mask,
917 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
918 * config/i386/i386-c.c (__AVX512BITALG__): New.
919 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
920 (ix86_valid_target_attribute_inner_p): Ditto.
921 (ix86_expand_args_builtin): Handle new types.
922 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
923 * config/i386/i386.opt: Add -mavx512bitalg.
924 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
925 avx512bitalgintrin.h.
926 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
927 (vpopcount<mode><mask_name>): Add more types.
928 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
929 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
931 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
933 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
934 Or in OPTION_MASK_ISA_AVX512F_SET.
935 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
936 OPTION_MASK_ISA_AVX512VNNI_UNSET.
937 (ix86_handle_option): Adjust for
938 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
939 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
941 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
942 isa_flag instead of isa_flag2.
943 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
944 isa_opts2 to isa_opts.
945 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
948 2017-12-22 Mike Stump <mikestump@comcast.net>
949 Eric Botcazou <ebotcazou@adacore.com>
951 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
953 2017-12-21 Jakub Jelinek <jakub@redhat.com>
956 * config/i386/i386.c (ix86_function_arg_boundary): Return
957 PARM_BOUNDARY for TYPE_EMPTY_P types.
960 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
961 if navail is >= dir.len.
963 2017-12-21 Steve Ellcey <sellcey@cavium.com>
965 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
968 2017-12-21 Jakub Jelinek <jakub@redhat.com>
970 PR rtl-optimization/80747
971 PR rtl-optimization/83512
972 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
973 succ edge from ENTRY, copy partition from e->dest to the newly
975 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
976 ENTRY, use BB_PARTITION of its successor block as current_partition.
977 Don't copy partition when splitting succ edge from ENTRY.
979 PR tree-optimization/83523
980 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
981 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
982 (convert_mult_to_fma): Likewise.
984 PR tree-optimization/83521
985 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
986 gimple_build_assign without code on result of
987 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
990 2017-12-21 Andrew Pinski <apinski@cavium.com>
991 Steve Ellcey <sellcey@cavium.com>
993 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
994 multi-arch for ilp32.
996 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
999 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1000 constraints to operand 2.
1001 (*ashl<mode>3_mask_1): Ditto.
1002 (*<shift_insn><mode>3_mask): Ditto.
1003 (*<shift_insn><mode>3_mask_1): Ditto.
1004 (*<rotate_insn><mode>3_mask): Ditto.
1005 (*<rotate_insn><mode>3_mask_1): Ditto.
1007 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1009 * reorg.c (make_return_insns): Reemit each insn with its own location.
1011 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1013 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1014 generation for cases where splatting a value is not useful.
1015 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1016 across a vec_duplicate and a paradoxical subreg forming a vector
1017 mode to a vec_concat.
1019 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1021 * combine.c (simplify_set): Do not transform subregs to zero_extends
1022 if the destination is not a scalar int mode.
1024 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1027 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1028 shared zero if the input has overflowed.
1030 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1032 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1033 support for Cortex-A55 and Cortex-A75.
1035 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1037 * common/config/arm/arm-common.c (compare_opt_names): Add function
1038 comment. Use strcmp instead of manual loop.
1040 2017-12-21 Martin Liska <mliska@suse.cz>
1042 PR gcov-profile/83509
1043 * gcov-dump.c (dump_gcov_file): Do not read info about
1044 support_unexecuted_blocks for gcda files.
1046 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1048 PR rtl-optimization/82973
1049 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1050 (valid_for_const_vector_p): ... this.
1051 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1052 (valid_for_const_vector_p): ... this. Adjust function comment.
1053 (gen_vec_duplicate): Adjust caller.
1054 * optabs.c (expand_vector_broadcast): Likewise.
1055 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1056 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1058 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1059 instead of GET_CODE == CONST_FIXED.
1060 (simplify_subreg): Use CONST_FIXED_P macro instead of
1061 GET_CODE == CONST_FIXED.
1064 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1065 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1066 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1067 (ix86_option_override_internal): Adjust for
1068 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1069 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1070 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1071 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1072 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1073 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1074 def_builtin2 for CET builtins.
1075 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1076 up formatting in IX86_BUILTIN_RDPID code.
1077 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1078 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1079 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1080 ix86_isa_flags2 to ix86_isa_flags.
1081 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1083 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1084 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1085 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1087 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1088 Or in OPTION_MASK_ISA_AVX512F_SET.
1089 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1090 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1091 (ix86_handle_option): Adjust for
1092 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1093 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1095 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1096 Alan Hayward <alan.hayward@arm.com>
1097 David Sherwood <david.sherwood@arm.com>
1099 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1100 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1101 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1102 Track polynomial offsets.
1104 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1109 Treat drb->init as a poly_int. Fail if its misalignment wrt
1110 vector_alignment isn't known.
1112 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1113 Alan Hayward <alan.hayward@arm.com>
1114 David Sherwood <david.sherwood@arm.com>
1116 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1118 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1119 (vect_loop_versioning): Take the loop versioning threshold as a
1121 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1122 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1123 versioning_threshold.
1124 (vect_analyze_loop_2): Compute the loop versioning threshold
1125 whenever loop versioning is needed, and store it in the new
1126 field rather than combining it with the cost model threshold.
1127 (vect_transform_loop): Update call to vect_loop_versioning.
1128 Try to combine the loop versioning and cost thresholds here.
1130 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1131 Alan Hayward <alan.hayward@arm.com>
1132 David Sherwood <david.sherwood@arm.com>
1134 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1135 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1136 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1137 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1138 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1139 (group_compare_offset): Update accordingly.
1140 (split_small_address_groups_p): Likewise.
1141 (record_use): Take addr_offset as a poly_uint64 rather than
1142 an unsigned HOST_WIDE_INT.
1143 (strip_offset): Return the offset as a poly_uint64 rather than
1144 an unsigned HOST_WIDE_INT.
1145 (record_group_use, split_address_groups): Track polynomial offsets.
1146 (add_iv_candidate_for_use): Likewise.
1147 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1148 than a HOST_WIDE_INT.
1149 (strip_offset_1): Return the offset as a poly_int64 rather than
1152 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1153 Alan Hayward <alan.hayward@arm.com>
1154 David Sherwood <david.sherwood@arm.com>
1156 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1157 rather than a HOST_WIDE_INT.
1158 * tree.c (get_binfo_at_offset): Likewise.
1160 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1161 Alan Hayward <alan.hayward@arm.com>
1162 David Sherwood <david.sherwood@arm.com>
1164 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1165 rather than a HOST_WIDE_INT.
1166 * tree-sra.c (build_ref_for_offset): Likewise.
1168 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1169 Alan Hayward <alan.hayward@arm.com>
1170 David Sherwood <david.sherwood@arm.com>
1172 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1174 * tree.c (mem_ref_offset): Likewise.
1175 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1176 * builtins.c (get_object_alignment_2): Likewise.
1177 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1178 * gimple-fold.c (get_base_constructor): Likewise.
1179 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1180 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1182 * ipa-polymorphic-call.c
1183 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1184 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1185 (get_ancestor_addr_info): Likewise.
1186 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1187 * match.pd: Likewise.
1188 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1189 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1190 * tree-eh.c (tree_could_trap_p): Likewise.
1191 * tree-object-size.c (addr_object_size): Likewise.
1192 * tree-ssa-address.c (copy_ref_info): Likewise.
1193 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1194 (indirect_refs_may_alias_p): Likewise.
1195 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1196 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1197 (non_rewritable_mem_ref_base): Likewise.
1198 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1199 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1200 * varasm.c (decode_addr_const): Likewise.
1202 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1203 Alan Hayward <alan.hayward@arm.com>
1204 David Sherwood <david.sherwood@arm.com>
1206 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1207 offsets for MEM_REFs.
1209 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1210 Alan Hayward <alan.hayward@arm.com>
1211 David Sherwood <david.sherwood@arm.com>
1213 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1214 as a poly_uint64 rather than an unsigned int.
1215 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1217 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1223 (decode_addr_const): Update accordingly.
1225 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1226 Alan Hayward <alan.hayward@arm.com>
1227 David Sherwood <david.sherwood@arm.com>
1229 * tree.h (bit_field_size, bit_field_offset): New functions.
1230 * hsa-gen.c (gen_hsa_addr): Use them.
1231 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1232 (simplify_vector_constructor): Likewise.
1233 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1234 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1235 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1236 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1239 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * expr.h (emit_group_load, emit_group_load_into_temps)
1244 (emit_group_store): Take the size as a poly_int64 rather than an int.
1245 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1246 (emit_group_load_into_temp, emit_group_store): Likewise.
1248 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1249 Alan Hayward <alan.hayward@arm.com>
1250 David Sherwood <david.sherwood@arm.com>
1252 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1253 an unsigned int to a poly_uint64.
1254 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1255 sizes as poly_uint64s rather than unsigned ints.
1256 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1259 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1260 Alan Hayward <alan.hayward@arm.com>
1261 David Sherwood <david.sherwood@arm.com>
1263 * lra-constraints.c (emit_inc): Change inc_amount from an int
1266 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1267 Alan Hayward <alan.hayward@arm.com>
1268 David Sherwood <david.sherwood@arm.com>
1270 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1272 (add_stack_var, stack_var_cmp, partition_stack_vars)
1273 (dump_stack_var_partition): Update accordingly.
1274 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1275 than a HOST_WIDE_INT.
1276 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1277 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1278 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1279 rather than a HOST_WIDE_INT.
1281 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1282 Alan Hayward <alan.hayward@arm.com>
1283 David Sherwood <david.sherwood@arm.com>
1285 * target.def (return_pops_args): Treat both the input and output
1286 sizes as poly_int64s rather than HOST_WIDE_INTS.
1287 * targhooks.h (default_return_pops_args): Update accordingly.
1288 * targhooks.c (default_return_pops_args): Likewise.
1289 * doc/tm.texi: Regenerate.
1290 * emit-rtl.h (incoming_args): Change pops_args, size and
1291 outgoing_args_size from int to poly_int64_pod.
1292 * function.h (expr_status): Change x_pending_stack_adjust and
1293 x_stack_pointer_delta from int to poly_int64.
1294 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1295 (ARGS_SIZE_RTX): Update accordingly.
1296 * calls.c (highest_outgoing_arg_in_use): Change from int to
1298 (stack_usage_watermark, stored_args_watermark): New variables.
1299 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1300 (emit_call_1): Change the stack_size and rounded_stack_size
1301 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1303 (save_fixed_argument_area): Check stack_usage_watermark.
1304 (initialize_argument_information): Change old_pending_adj from
1305 a HOST_WIDE_INT * to a poly_int64_pod *.
1306 (compute_argument_block_size): Return the size as a poly_int64
1308 (finalize_must_preallocate): Track polynomial argument sizes.
1309 (compute_argument_addresses): Likewise.
1310 (internal_arg_pointer_based_exp): Track polynomial offsets.
1311 (mem_overlaps_already_clobbered_arg_p): Rename to...
1312 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1313 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1314 Check stored_args_used_watermark.
1315 (load_register_parameters): Update accordingly.
1316 (check_sibcall_argument_overlap_1): Likewise.
1317 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1318 args size as a poly_int64 rather than an int. Return a bool
1319 indicating whether the optimization was possible and return
1320 the new adjustment by reference.
1321 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1322 Update stored_args_watermark.
1323 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1324 (expand_call): Likewise. Maintain stack_usage_watermark and
1325 stored_args_watermark. Update calls to
1326 combine_pending_stack_adjustment_and_call.
1327 (emit_library_call_value_1): Handle polynomial argument sizes.
1328 Call stack_region_maybe_used_p and mark_stack_region_used.
1329 Maintain stack_usage_watermark.
1330 (store_one_arg): Likewise. Update call to
1331 mem_overlaps_already_clobbered_arg_p.
1332 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1334 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1335 * config/microblaze/microblaze.c (microblaze_function_prologue):
1337 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1338 TARGET_RETURN_POPS_ARGS interface.
1339 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1341 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1342 * config/i386/i386.c (ix86_return_pops_args): Update for new
1343 TARGET_RETURN_POPS_ARGS interface.
1344 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1345 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1346 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1347 TARGET_RETURN_POPS_ARGS interface.
1348 * config/vax/vax.c (vax_return_pops_args): Likewise.
1349 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1350 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1351 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1352 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1353 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1354 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1355 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1356 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1357 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1358 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1359 and x_stack_pointer_delta from int to poly_int64.
1360 * dojump.c (do_pending_stack_adjust): Update accordingly.
1361 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1362 stack_pointer_deltas.
1363 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1364 (pad_to_arg_alignment): Track polynomial offsets.
1365 (assign_parm_find_stack_rtl): Likewise.
1366 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1367 * toplev.c (output_stack_usage): Update reference to
1368 current_function_pushed_stack_size.
1370 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * function.c (in_arg_offset, var_offset, dynamic_offset)
1375 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1376 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1377 rather than a HOST_WIDE_INT.
1378 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1379 (instantiate_virtual_regs_in_insn): Likewise.
1381 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1382 Alan Hayward <alan.hayward@arm.com>
1383 David Sherwood <david.sherwood@arm.com>
1385 * rtl.h (get_args_size, add_args_size_note): New functions.
1386 (find_args_size_adjust): Return a poly_int64 rather than a
1388 (fixup_args_size_notes): Likewise. Make the same change to the
1389 end_args_size parameter.
1390 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1391 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1392 * calls.c (emit_call_1): Likewise.
1393 * explow.c (adjust_stack_1): Likewise.
1394 * cfgcleanup.c (old_insns_match_p): Update use of
1395 find_args_size_adjust.
1396 * combine.c (distribute_notes): Track polynomial arg sizes.
1397 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1398 end_true_args_size, beg_delay_args_size and end_delay_args_size
1399 from HOST_WIDE_INT to poly_int64.
1400 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1401 than a HOST_WIDE_INT.
1402 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1403 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1404 polynomial arg sizes.
1405 * emit-rtl.c (try_split): Use get_args_size.
1406 * recog.c (peep2_attempt): Likewise.
1407 * reload1.c (reload_as_needed): Likewise.
1408 * expr.c (find_args_size_adjust): Return the adjustment as a
1409 poly_int64 rather than a HOST_WIDE_INT.
1410 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1411 to a poly_int64 and change the return type in the same way.
1412 (emit_single_push_insn): Track polynomial arg sizes.
1414 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1415 Alan Hayward <alan.hayward@arm.com>
1416 David Sherwood <david.sherwood@arm.com>
1418 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1419 from HOST_WIDE_INT to poly_int64.
1420 * expr.c (push_block, emit_push_insn): Likewise.
1422 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1423 Alan Hayward <alan.hayward@arm.com>
1424 David Sherwood <david.sherwood@arm.com>
1426 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1428 (get_frame_size): Return the size as a poly_int64 rather than a
1430 (frame_offset_overflow): Take the offset as a poly_int64 rather
1431 than a HOST_WIDE_INT.
1432 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1433 (assign_stack_temp): Likewise for the size.
1434 * function.c (get_frame_size): Return a poly_int64 rather than
1436 (frame_offset_overflow): Take the offset as a poly_int64 rather
1437 than a HOST_WIDE_INT.
1438 (try_fit_stack_local): Take the start, length and size as poly_int64s
1439 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1440 rather than a HOST_WIDE_INT.
1441 (add_frame_space): Take the start and end as poly_int64s rather than
1443 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1444 (assign_stack_temp): Likewise for the size.
1445 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1447 (find_temp_slot_from_address): Handle polynomial offsets.
1448 (combine_temp_slots): Likewise.
1449 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1451 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1452 poly_int64 rather than a HOST_WIDE_INT.
1453 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1454 than a HOST_WIDE_INT.
1455 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1456 polynomial frame offsets.
1457 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1458 as a poly_int64 rather than an int.
1459 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1460 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1461 * config/v850/v850.c (compute_frame_size): Likewise.
1462 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1463 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1464 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1465 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1466 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1467 rather than a HOST_WIDE_INT.
1468 * explow.c (get_dynamic_stack_base): Likewise.
1469 * final.c (final_start_function): Use the constant lower bound
1470 of the frame size for -Wframe-larger-than.
1471 * ira.c (do_reload): Adjust for new get_frame_size return type.
1472 * lra.c (lra): Likewise.
1473 * reload1.c (reload): Likewise.
1474 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1475 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1476 * rtlanal.c (get_initial_register_offset): Return the offset as
1477 a poly_int64 rather than a HOST_WIDE_INT.
1479 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1480 Alan Hayward <alan.hayward@arm.com>
1481 David Sherwood <david.sherwood@arm.com>
1483 * reload1.c (elim_table): Change initial_offset, offset and
1484 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1485 (offsets_at): Change the target array's element type from
1486 HOST_WIDE_INT to poly_int64_pod.
1487 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1488 (elimination_costs_in_insn, update_eliminable_offsets)
1489 (verify_initial_elim_offsets, set_offsets_for_label)
1490 (init_eliminable_invariants): Update after above changes.
1492 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1493 Alan Hayward <alan.hayward@arm.com>
1494 David Sherwood <david.sherwood@arm.com>
1496 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1497 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1498 (decomposition): Change start and end from HOST_WIDE_INT
1500 (decompose, immune_p): Update accordingly.
1501 (find_inc_amount): Return a poly_int64 rather than an int.
1502 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1505 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * tree.h (get_inner_reference): Return the bitsize and bitpos
1510 as poly_int64_pods rather than HOST_WIDE_INT.
1511 * fold-const.h (ptr_difference_const): Return the pointer difference
1512 as a poly_int64_pod rather than a HOST_WIDE_INT.
1513 * expr.c (get_inner_reference): Return the bitsize and bitpos
1514 as poly_int64_pods rather than HOST_WIDE_INT.
1515 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1517 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1518 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1519 (optimize_bit_field_compare): Update call to get_inner_reference.
1520 (decode_field_reference): Likewise.
1521 (fold_unary_loc): Track polynomial offsets and sizes.
1522 (split_address_to_core_and_offset): Return the bitpos as a
1523 poly_int64_pod rather than a HOST_WIDE_INT.
1524 (ptr_difference_const): Likewise for the pointer difference.
1525 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1526 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1527 * dbxout.c (dbxout_expand_expr): Likewise.
1528 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1529 (loc_list_from_tree_1, fortran_common): Likewise.
1530 * gimple-laddress.c (pass_laddress::execute): Likewise.
1531 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1532 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1533 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1534 * tree-affine.c (tree_to_aff_combination): Likewise.
1535 (get_inner_reference_aff): Likewise.
1536 * tree-data-ref.c (split_constant_offset_1): Likewise.
1537 (dr_analyze_innermost): Likewise.
1538 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1539 * tree-sra.c (ipa_sra_check_caller): Likewise.
1540 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1541 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1542 (instrument_bool_enum_load, instrument_object_size): Likewise.
1543 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1544 to get_inner_reference.
1545 * hsa-gen.c (gen_hsa_addr): Likewise.
1546 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1547 * tsan.c (instrument_expr): Likewise.
1548 * match.pd: Update call to ptr_difference_const.
1550 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1551 Alan Hayward <alan.hayward@arm.com>
1552 David Sherwood <david.sherwood@arm.com>
1554 * fold-const.c (fold_comparison): Track sizes and offsets as
1555 poly_int64s rather than HOST_WIDE_INTs when folding address
1558 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * expr.h (get_bit_range): Return the bitstart and bitend as
1563 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1564 as a poly_int64 rather than a HOST_WIDE_INT.
1565 * expr.c (get_bit_range): Likewise.
1566 (expand_assignment): Update call accordingly.
1567 * fold-const.c (optimize_bit_field_compare): Likewise.
1569 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1570 Alan Hayward <alan.hayward@arm.com>
1571 David Sherwood <david.sherwood@arm.com>
1573 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1574 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1575 as an optional final parameter.
1576 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1577 (aarch64_legitimate_address_p): Likewise.
1578 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1579 instead of an rtx code.
1580 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1581 (aarch64_legitimate_address_hook_p): Likewise.
1582 (aarch64_print_ldpstp_address): Likewise.
1583 (aarch64_print_operand_address): Likewise.
1584 (aarch64_address_cost): Likewise.
1585 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1586 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1587 (aarch64_mem_pair_lanes_operand): Likewise.
1589 2017-12-20 Richard Biener <rguenther@suse.de>
1591 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1592 update_stmt_if_modified.
1594 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1596 PR tree-optimization/83491
1597 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1598 before walking uses. Improve coding style and comments.
1600 2017-12-20 Tom de Vries <tom@codesourcery.com>
1602 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1604 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1607 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1609 (pass_ipa_free_fn_summary): Add small_p private data member,
1610 initialize to false in the ctor.
1611 (pass_ipa_free_fn_summary::clone,
1612 pass_ipa_free_fn_summary::set_pass_param,
1613 pass_ipa_free_fn_summary::gate): New methods.
1614 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1615 | TODO_dump_symtab if small_p.
1616 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1617 entry and add another instance of the pass with false parm after
1619 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1620 ipa_free_fn_summary here.
1622 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1624 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1626 2017-12-20 Martin Sebor <msebor@redhat.com>
1629 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1630 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1631 (expand_builtin_strncmp): Same.
1633 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1636 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1637 allowed debug stmts before labels.
1638 (expand_gimple_basic_block): Likewise.
1639 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1640 * gimple-iterator.h (gsi_after_labels): Likewise.
1641 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1642 rename reused variable, and simplify using gsi_move_before.
1643 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1644 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1645 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1646 (gimple_verify_flow_info, gimple_block_label): Likewise.
1647 (make_blocks): Move debug markers after adjacent labels.
1648 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1649 allowed debug insns outside blocks.
1650 * df-scan.c (df_insn_delete): Likewise.
1651 * lra-constraints.c (update_ebb_live_info): Likewise.
1652 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1653 (vt_initialize, delete_vta_debug_insns): Likewise.
1654 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1656 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1657 Alan Hayward <alan.hayward@arm.com>
1658 David Sherwood <david.sherwood@arm.com>
1660 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1661 (round_up_to_byte_boundary): Likewise.
1662 * expr.h (get_bit_range): Add temporary shim.
1663 * gimple-ssa-store-merging.c (store_operand_info): Change the
1664 bitsize, bitpos, bitregion_start and bitregion_end fields from
1665 unsigned HOST_WIDE_INT to poly_uint64.
1666 (merged_store_group): Likewise load_align_base.
1667 (compatible_load_p, compatible_load_p): Update accordingly.
1668 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1669 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1670 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1671 bitregion_start and bitregion_end as poly_uint64s rather than
1672 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1673 (handled_load): Take the bitsize, bitpos,
1674 bitregion_start and bitregion_end as poly_uint64s rather than
1675 unsigned HOST_WIDE_INTs.
1676 (pass_store_merging::process_store): Update call to
1677 mem_valid_for_store_merging.
1679 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1680 Alan Hayward <alan.hayward@arm.com>
1681 David Sherwood <david.sherwood@arm.com>
1683 * builtins.c (get_object_alignment_2): Track polynomial offsets
1684 and sizes. Update the alignment handling.
1686 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1687 Alan Hayward <alan.hayward@arm.com>
1688 David Sherwood <david.sherwood@arm.com>
1690 * tree.h (get_inner_reference): Add a version that returns the
1691 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1692 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1693 the case in which bitpos is not associated with the first byte.
1695 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * tree-affine.h (get_inner_reference_aff): Return the size as a
1701 * tree-affine.c (get_inner_reference_aff): Likewise.
1702 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1703 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1705 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1706 Alan Hayward <alan.hayward@arm.com>
1707 David Sherwood <david.sherwood@arm.com>
1709 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1710 HOST_WIDE_INT rather than a poly_int64.
1712 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1713 Alan Hayward <alan.hayward@arm.com>
1714 David Sherwood <david.sherwood@arm.com>
1716 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1717 HOST_WIDE_INT to poly_int64_pod.
1718 (perform_symbolic_merge): Update accordingly.
1719 (bswap_replace): Likewise.
1721 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1722 Alan Hayward <alan.hayward@arm.com>
1723 David Sherwood <david.sherwood@arm.com>
1725 * tree-affine.h (aff_tree::offset): Change from widest_int
1727 (wide_int_ext_for_comb): Delete.
1728 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1729 constants as poly_widest_int rather than widest_int.
1730 (aff_combination_constant_multiple_p): Return the multiplier
1731 as a poly_widest_int.
1732 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1734 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1735 version static and add an overload for poly_widest_int.
1736 (aff_combination_const, aff_combination_add_cst)
1737 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1738 the constants as poly_widest_int rather than widest_int.
1739 (tree_to_aff_combination): Generalize INTEGER_CST case to
1741 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1742 (aff_combination_add_product, aff_combination_mult): Handle
1744 (aff_combination_constant_multiple_p): Return the multiplier
1745 as a poly_widest_int.
1746 * tree-predcom.c (determine_offset): Return the offset as a
1748 (split_data_refs_to_components, suitable_component_p): Update
1750 (valid_initializer_p): Update call to
1751 aff_combination_constant_multiple_p.
1752 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1753 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1754 as a poly_int64 rather than a HOST_WIDE_INT.
1755 (get_address_cost): Handle polynomial offsets.
1756 (iv_elimination_compare_lt): Likewise.
1757 (rewrite_use_nonlinear_expr): Likewise.
1759 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1760 Alan Hayward <alan.hayward@arm.com>
1761 David Sherwood <david.sherwood@arm.com>
1763 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1764 as a poly_int64_pod rather than a HOST_WIDE_INT.
1765 (get_addr_base_and_unit_offset): Likewise.
1766 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1767 (get_addr_base_and_unit_offset): Likewise.
1768 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1769 to poly_int64 in example.
1770 * fold-const.c (fold_binary_loc): Update call to
1771 get_addr_base_and_unit_offset.
1772 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1773 (maybe_canonicalize_mem_ref_addr): Likewise.
1774 (gimple_fold_stmt_to_constant_1): Likewise.
1775 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1777 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1778 * match.pd: Likewise.
1779 * omp-low.c (lower_omp_target): Likewise.
1780 * tree-sra.c (build_ref_for_offset): Likewise.
1781 (build_debug_ref_for_model): Likewise.
1782 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1783 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1784 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1785 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1786 (constant_pointer_difference): Likewise.
1787 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1788 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1789 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1790 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1791 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1792 (set_ssa_val_to): Likewise.
1793 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1794 (maybe_diag_stxncpy_trunc): Likewise.
1795 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1796 * tree.c (build_simple_mem_ref_loc): Likewise.
1797 (array_at_struct_end_p): Likewise.
1799 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1804 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1805 (get_ref_base_and_extent_hwi): Declare.
1806 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1807 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1808 (get_ref_base_and_extent_hwi): New function.
1809 * cfgexpand.c (expand_debug_expr): Update call to
1810 get_ref_base_and_extent.
1811 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1812 * gimple-fold.c (get_base_constructor): Return the offset as a
1813 poly_int64_pod rather than a HOST_WIDE_INT.
1814 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1815 * ipa-polymorphic-call.c
1816 (ipa_polymorphic_call_context::set_by_invariant)
1817 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1818 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1819 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1820 rather than get_ref_base_and_extent.
1821 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1822 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1823 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1825 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1826 call to get_ref_base_and_extent.
1827 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1828 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1829 (stmt_kills_ref_p): Likewise.
1830 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1831 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1833 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1834 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1835 when calling native_encode_expr.
1836 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1837 call to get_ref_base_and_extent.
1838 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1839 get_ref_base_and_extent.
1840 * var-tracking.c (track_expr_p): Likewise.
1842 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1843 Alan Hayward <alan.hayward@arm.com>
1844 David Sherwood <david.sherwood@arm.com>
1846 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1847 HOST_WIDE_INT to poly_int64_pod.
1848 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1849 polynomail parameter offsets.
1851 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1852 Alan Hayward <alan.hayward@arm.com>
1853 David Sherwood <david.sherwood@arm.com>
1855 * gengtype.c (main): Handle poly_int64_pod.
1856 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1857 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1858 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1859 from HOST_WIDE_INT to poly_int64_pod.
1860 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1861 (copy_cfa): New function.
1862 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1863 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1864 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1865 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1866 (initial_return_save): Treat offsets as poly_ints.
1867 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1868 offset is nonconstant.
1869 (reg_save): Take the offset as a poly_int64. Fall back to
1870 DW_CFA_expression for nonconstant offsets.
1871 (queue_reg_save): Take the offset as a poly_int64.
1872 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1874 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1875 Alan Hayward <alan.hayward@arm.com>
1876 David Sherwood <david.sherwood@arm.com>
1878 * rtl.h (operand_subword, operand_subword_force): Take the offset
1879 as a poly_uint64 an unsigned int.
1880 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1882 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1887 'p' format code. Use INT_LIST rather than SUBREG as the example of
1888 a code with an XINT and an XEXP. Remove the implication that
1889 accessing an rtx field using XINT is expected to work.
1890 * rtl.def (SUBREG): Change format from "ei" to "ep".
1891 * rtl.h (rtunion::rt_subreg): New field.
1892 (XCSUBREG): New macro.
1893 (SUBREG_BYTE): Use it.
1894 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1895 Update constructor accordingly.
1896 (subreg_shape::operator ==): Update accordingly.
1897 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1898 than an unsigned int.
1899 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1900 a poly_uint64 rather than an unsigned int.
1901 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1902 than an unsigned int.
1903 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1904 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1905 an unsigned int. Take the sizes as poly_uint64s.
1906 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1907 an unsigned int. Take the shift as a poly_uint64 rather than
1909 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1910 as a poly_uint64 rather than an unsigned int.
1911 (simplify_subreg_regno): Likewise.
1912 (byte_lowpart_offset): Return the memory offset as a poly_int64
1914 (subreg_memory_offset): Likewise. Take the subreg offset as a
1915 poly_uint64 rather than an unsigned int.
1916 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1917 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1918 poly_uint64 rather than an unsigned int.
1919 * rtl.c (rtx_format): Describe 'p' in comment.
1920 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1921 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1922 offset as a poly_uint64 rather than an unsigned int.
1923 (byte_lowpart_offset): Return the memory offset as a poly_int64
1925 (subreg_memory_offset): Likewise. Take the subreg offset as a
1926 poly_uint64 rather than an unsigned int.
1927 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1928 mode sizes as poly_uint64s rather than unsigned ints. Return a
1929 poly_uint64 rather than an unsigned int.
1930 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1931 (copy_insn_1): Handle 'p'.
1932 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1933 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1934 an unsigned int. Return the shift in the same way.
1935 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1937 (subreg_size_offset_from_lsb): Take the sizes and shift as
1938 poly_uint64s rather than unsigned ints. Return the offset as
1940 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1941 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1943 * rtlhash.c (add_rtx): Handle 'p'.
1944 * genemit.c (gen_exp): Likewise.
1945 * gengenrtl.c (type_from_format, gendef): Likewise.
1946 * gensupport.c (subst_pattern_match, get_alternatives_number)
1947 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1948 (subst_dup): Likewise.
1949 * gengtype.c (adjust_field_rtx_def): Likewise.
1950 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1951 (match_pattern_2): Likewise.
1952 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1953 (rtx_test::subreg_field): New function.
1954 (operator ==, safe_to_hoist_p, transition_parameter_type)
1955 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1956 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1957 * genpeep.c (match_rtx): Likewise.
1958 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1959 (rtx_writer::print_rtx_operand): Handle 'p'.
1960 (print_value): Handle SUBREG.
1961 * read-rtl.c (apply_int_iterator): Likewise.
1962 (rtx_reader::read_rtx_operand): Handle 'p'.
1963 * alias.c (rtx_equal_for_memref_p): Likewise.
1964 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1965 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1967 * calls.c (expand_call): Likewise.
1968 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1969 (make_extraction, gen_lowpart_for_combine): Likewise.
1970 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1972 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1973 rather than an unsigned int. Treat subreg offsets as poly_ints.
1974 (exp_equiv_p): Handle 'p'.
1975 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1976 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1977 * dse.c (find_shift_sequence): Likewise.
1978 * dwarf2out.c (rtl_for_decl_location): Likewise.
1979 * expmed.c (extract_low_bits): Likewise.
1980 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1981 (expand_expr_real_2): Likewise.
1982 * final.c (alter_subreg): Likewise.
1983 (leaf_renumber_regs_insn): Handle 'p'.
1984 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1985 Treat subreg offsets as poly_ints.
1986 * fwprop.c (forward_propagate_and_simplify): Likewise.
1987 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1988 * ira.c (get_subreg_tracking_sizes): Likewise.
1989 * ira-conflicts.c (go_through_subreg): Likewise.
1990 * ira-lives.c (process_single_reg_class_operands): Likewise.
1991 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1992 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1993 as a poly_uint64 rather than an unsigned int.
1994 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1995 subreg offsets as poly_ints.
1996 * lra-constraints.c (operands_match_p): Handle 'p'.
1997 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1998 * lra-spills.c (assign_mem_slot): Likewise.
1999 * postreload.c (move2add_valid_value_p): Likewise.
2000 * recog.c (general_operand, indirect_operand): Likewise.
2001 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2002 (copyprop_hardreg_forward_1): Likewise.
2003 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2004 (record_subregs_of_mode): Likewise.
2005 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2006 * reload.c (operands_match_p): Handle 'p'.
2007 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2008 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2009 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2010 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2011 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2012 (simplify_const_poly_int_tests<N>::run): Likewise.
2013 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2014 a poly_uint64 rather than an unsigned int.
2015 * valtrack.c (debug_lowpart_subreg): Likewise.
2016 * var-tracking.c (var_lowpart): Likewise.
2017 (loc_cmp): Handle 'p'.
2019 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2020 Alan Hayward <alan.hayward@arm.com>
2021 David Sherwood <david.sherwood@arm.com>
2023 * ira.c (get_subreg_tracking_sizes): New function.
2024 (init_live_subregs): Take an integer size rather than a register.
2025 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2026 to init_live_subregs.
2028 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * expr.c (store_constructor_field): Change bitsize from a
2033 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2034 HOST_WIDE_INT to a poly_int64.
2035 (store_constructor): Change size from a HOST_WIDE_INT to
2037 (store_field): Likewise bitsize and bitpos.
2039 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2040 Alan Hayward <alan.hayward@arm.com>
2041 David Sherwood <david.sherwood@arm.com>
2043 * expmed.h (store_bit_field): Change bitregion_start and
2044 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2045 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2046 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2047 (store_fixed_bit_field, store_split_bit_field): Likewise.
2048 * expr.c (store_constructor_field, store_field): Likewise.
2049 (optimize_bitfield_assignment_op): Likewise. Make the same change
2050 to bitsize and bitpos.
2051 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2052 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2053 same change in the constructor arguments.
2054 (get_best_mode): Change bitregion_start and bitregion_end from
2055 unsigned HOST_WIDE_INT to poly_uint64.
2056 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2057 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2059 (bit_field_mode_iterator::next_mode): Update for new types
2060 of m_bitregion_start and m_bitregion_end.
2061 (get_best_mode): Change bitregion_start and bitregion_end from
2062 unsigned HOST_WIDE_INT to poly_uint64.
2064 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2065 Alan Hayward <alan.hayward@arm.com>
2066 David Sherwood <david.sherwood@arm.com>
2068 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2069 accepts poly_uint64 offsets.
2070 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2071 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2072 * expmed.c (lowpart_bit_field_p): Likewise.
2073 (extract_bit_field_as_subreg): New function, split out from...
2074 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2075 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2076 extractions, check that BITSIZE matches the size of the extracted
2077 value and that BITNUM is an exact multiple of that size.
2078 If all else fails, try forcing the value into memory if
2079 BITNUM is variable, and adjusting the address so that the
2080 offset is constant. Split the part that can only handle constant
2081 bitsize and bitnum out into...
2082 (extract_integral_bit_field): ...this new function.
2083 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2084 rather than unsigned HOST_WIDE_INTs.
2086 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2087 Alan Hayward <alan.hayward@arm.com>
2088 David Sherwood <david.sherwood@arm.com>
2090 * expmed.h (store_bit_field): Take bitsize and bitnum as
2091 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2092 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2093 that returns the byte size.
2094 (store_bit_field_1): Take bitsize and bitnum as
2095 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2096 to simple_mem_bitfield_p. Split the part that can only handle
2097 constant bitsize and bitnum out into...
2098 (store_integral_bit_field): ...this new function.
2099 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2100 than unsigned HOST_WIDE_INTs.
2101 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2103 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2104 Alan Hayward <alan.hayward@arm.com>
2105 David Sherwood <david.sherwood@arm.com>
2107 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2108 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2110 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2111 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2112 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2113 offset as a poly_int64 rather than an int.
2114 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2115 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2116 * lra-constraints.c (equiv_address_substitution): Track offsets
2118 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2119 (curr_insn_transform): Handle the new form of sp_offset.
2120 * lra-eliminations.c (lra_elim_table): Change previous_offset
2121 and offset from HOST_WIDE_INT to poly_int64.
2122 (print_elim_table, update_reg_eliminate): Update accordingly.
2123 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2124 (get_elimination): Update accordingly.
2125 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2126 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2127 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2128 poly_int64 offsets generally.
2129 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2130 (mark_not_eliminable, init_elimination): Update accordingly.
2131 (remove_reg_equal_offset_note): Return a bool and pass the new
2132 offset back by pointer as a poly_int64.
2133 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2134 rather than a HOST_WIDE_INT.
2135 (do_remat): Track offsets poly_int64s.
2136 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2138 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2139 Alan Hayward <alan.hayward@arm.com>
2140 David Sherwood <david.sherwood@arm.com>
2142 * rtl.h (mem_attrs): Add a default constructor. Change size and
2143 offset from HOST_WIDE_INT to poly_int64.
2144 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2145 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2146 (widen_memory_access): Take the sizes and offsets as poly_int64s
2147 rather than HOST_WIDE_INTs.
2148 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2149 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2151 (adjust_offset_for_component_ref): Change the offset from a
2152 HOST_WIDE_INT to a poly_int64.
2153 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2154 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2155 * dce.c (find_call_stack_args): Likewise.
2156 * dse.c (record_store): Likewise.
2157 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2158 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2159 * read-rtl-function.c (test_loading_mem): Likewise.
2160 * rtlanal.c (may_trap_p_1): Likewise.
2161 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2162 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2163 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2164 (mem_attrs::mem_attrs): New function.
2165 (set_mem_attributes_minus_bitpos): Change bitpos from a
2166 HOST_WIDE_INT to poly_int64.
2167 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2168 (clear_mem_offset, clear_mem_size, change_address)
2169 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2170 initialize mem_attrs.
2171 (set_mem_offset, set_mem_size, adjust_address_1)
2172 (adjust_automodify_address_1, offset_address, widen_memory_access):
2173 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2175 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2176 Alan Hayward <alan.hayward@arm.com>
2177 David Sherwood <david.sherwood@arm.com>
2179 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2180 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2181 rather than 0 to represent an unknown size. Assert that the size
2182 is known when the mode isn't BLKmode.
2183 (may_trap_p_1): Use -1 for unknown sizes.
2184 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2186 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2187 Alan Hayward <alan.hayward@arm.com>
2188 David Sherwood <david.sherwood@arm.com>
2190 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2191 to poly_int64. Update commentary for positions_needed.large.
2192 (read_info_type): Change offset and width from HOST_WIDE_INT
2194 (set_usage_bits): Likewise.
2195 (canon_address): Return the offset as a poly_int64 rather than
2196 a HOST_WIDE_INT. Use strip_offset_and_add.
2197 (set_all_positions_unneeded, any_positions_needed_p): Use
2198 positions_needed.large to track stores with non-constant widths.
2199 (all_positions_needed_p): Likewise. Take the offset and width
2200 as poly_int64s rather than ints. Assert that rhs is nonnull.
2201 (record_store): Cope with non-constant offsets and widths.
2202 Nullify the rhs of an earlier store if we can't tell which bytes
2204 (find_shift_sequence): Take the access_size and shift as poly_int64s
2206 (get_stored_val): Take the read_offset and read_width as poly_int64s
2207 rather than HOST_WIDE_INTs.
2208 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2209 non-constant offsets and widths.
2211 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2212 Alan Hayward <alan.hayward@arm.com>
2213 David Sherwood <david.sherwood@arm.com>
2215 * inchash.h (inchash::hash::add_poly_int): New function.
2216 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2217 Use poly_int64 rather than HOST_WIDE_INT.
2218 (ao_ref::max_size_known_p): New function.
2219 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2220 rather than HOST_WIDE_INT.
2221 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2222 to temporaries until its interface is adjusted to match.
2223 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2224 (aliasing_component_refs_p, decl_refs_may_alias_p)
2225 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2226 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2227 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2229 * alias.c (ao_ref_from_mem): Likewise.
2230 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2231 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2232 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2233 (maybe_trim_complex_store, maybe_trim_constructor_store)
2234 (live_bytes_read, dse_classify_store): Likewise.
2235 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2236 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2237 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2238 (vn_reference_lookup_3): Likewise.
2239 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2241 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2242 Alan Hayward <alan.hayward@arm.com>
2243 David Sherwood <david.sherwood@arm.com>
2245 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2246 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2247 instead of ranges_overlap_p.
2249 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2250 Alan Hayward <alan.hayward@arm.com>
2251 David Sherwood <david.sherwood@arm.com>
2253 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2254 sizes as poly_int64s rather than HOST_WIDE_INTs.
2256 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2257 Alan Hayward <alan.hayward@arm.com>
2258 David Sherwood <david.sherwood@arm.com>
2260 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2261 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2262 * gimple-fold.c (fold_ctor_reference): Likewise.
2264 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2265 Alan Hayward <alan.hayward@arm.com>
2266 David Sherwood <david.sherwood@arm.com>
2268 * target.def (dwarf_poly_indeterminate_value): New hook.
2269 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2270 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2271 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2272 * doc/tm.texi: Regenerate.
2273 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2274 offset as a poly_int64.
2275 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2276 offset as a poly_int64.
2277 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2278 Take the offset as a poly_int64.
2279 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2280 (frame_pointer_fb_offset): Change to a poly_int64.
2281 (int_loc_descriptor): Take the offset as a poly_int64. Use
2282 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2283 (based_loc_descr): Take the offset as a poly_int64.
2284 Use strip_offset_and_add to handle (plus X (const)).
2285 Use new_reg_loc_descr instead of an open-coded version of the
2286 previous implementation.
2287 (mem_loc_descriptor): Handle CONST_POLY_INT.
2288 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2289 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2291 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2292 Alan Hayward <alan.hayward@arm.com>
2293 David Sherwood <david.sherwood@arm.com>
2295 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2296 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2297 * inchash.h (inchash::hash::add_poly_hwi): New function.
2298 * gengtype.c (main): Register poly_int64.
2299 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2300 offset as a poly_int.
2301 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2302 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2303 offset as a poly_int64.
2304 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2305 * print-rtl.c (print_poly_int): New function.
2306 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2308 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2310 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2311 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2312 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2313 Enforce track_offset_p for parts of a PARALLEL.
2314 (vt_add_function_parameter): Use const_offset for the final
2315 offset to track. Use get_tracked_reg_offset for the parts
2318 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2323 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2324 * doc/tm.texi: Regenerate.
2325 * hooks.h (hook_bool_uint_uint_true): Delete.
2326 (hook_bool_puint64_puint64_true): Declare.
2327 * hooks.c (hook_bool_uint_uint_true): Delete.
2328 (hook_bool_puint64_puint64_true): New function.
2329 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2330 instead of unsigned ints.
2331 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2332 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2334 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2335 Alan Hayward <alan.hayward@arm.com>
2336 David Sherwood <david.sherwood@arm.com>
2338 * optabs.h (expand_operand): Add an int_value field.
2339 (create_expand_operand): Add an int_value parameter and use it
2340 to initialize the new expand_operand field.
2341 (create_integer_operand): Replace with a declaration of a function
2342 that accepts poly_int64s. Move the implementation to...
2343 * optabs.c (create_integer_operand): ...here.
2344 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2345 the mode preserves the value of int_value, instead of calling
2346 const_int_operand on the rtx. Use gen_int_mode to generate
2349 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2350 Alan Hayward <alan.hayward@arm.com>
2351 David Sherwood <david.sherwood@arm.com>
2353 * dumpfile.h (dump_dec): Declare.
2354 * dumpfile.c (dump_dec): New function.
2355 * pretty-print.h (pp_wide_integer): Turn into a function and
2356 declare a poly_int version.
2357 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2359 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayward <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * doc/generic.texi (POLY_INT_CST): Document.
2364 * tree.def (POLY_INT_CST): New tree code.
2365 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2366 * tree-core.h (tree_poly_int_cst): New struct.
2367 (tree_node): Add a poly_int_cst field.
2368 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2369 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2370 instead of a wide_int_ref.
2371 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2373 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2374 instead of an unsigned HOST_WIDE_INT.
2375 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2376 (ptrdiff_tree_p): Declare.
2377 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2378 extern inline implementations if the target doesn't use POLY_INT_CST.
2379 (poly_int_tree_p): New function.
2380 (wi::unextended_tree): New class.
2381 (wi::int_traits <unextended_tree>): New override.
2382 (wi::extended_tree): Add a default constructor.
2383 (wi::extended_tree::get_tree): New function.
2384 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2385 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2386 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2387 (wi::tree_to_poly_wide_ref): New typedefs.
2388 (wi::ints_for): Provide overloads for extended_tree and
2390 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2391 (wi::to_wide): New functions.
2392 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2393 * tree.c (poly_int_cst_hasher): New struct.
2394 (poly_int_cst_hash_table): New variable.
2395 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2396 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2398 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2399 (init_ttree): Initialize poly_int_cst_hash_table.
2400 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2401 a poly_int64 instead of a HOST_WIDE_INT.
2402 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2403 instead of an unsigned HOST_WIDE_INT.
2404 (wide_int_to_tree): Rename to...
2405 (wide_int_to_tree_1): ...this.
2406 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2407 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2408 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2409 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2410 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2411 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2413 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2414 (streamer_read_tree_body): Likewise.
2415 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2416 (streamer_write_tree_body): Likewise.
2417 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2418 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2419 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2420 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2421 * gimple-expr.h (is_gimple_constant): Likewise.
2422 * gimplify.c (maybe_with_size_expr): Likewise.
2423 * print-tree.c (print_node): Likewise.
2424 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2425 * tree-pretty-print.c (dump_generic_node): Likewise.
2426 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2427 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2428 * tree-vrp.c (compare_values_warnv): Likewise.
2429 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2430 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2431 (force_expr_to_var_cost): Likewise.
2432 * tree-ssa-loop.c (for_each_index): Likewise.
2433 * fold-const.h (build_invariant_address, size_int_kind): Take a
2434 poly_int64 instead of a HOST_WIDE_INT.
2435 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2436 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2438 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2439 (int_const_binop_2): New function, split out from...
2440 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2441 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2442 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2443 * gimple-ssa-strength-reduction.c (slsr_process_add)
2444 (slsr_process_mul): Check for INTEGER_CSTs before using them
2446 * stor-layout.c (bits_from_bytes): New function.
2447 (bit_from_pos): Use it.
2448 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2449 by BITS_PER_UNIT to get the TYPE_SIZE.
2450 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2451 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2453 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2454 Alan Hayward <alan.hayward@arm.com>
2455 David Sherwood <david.sherwood@arm.com>
2457 * doc/rtl.texi (const_poly_int): Document. Also document the
2458 rtl sharing behavior.
2459 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2460 * rtl.h (const_poly_int_def): New struct.
2461 (rtx_def::u): Add a cpi field.
2462 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2463 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2464 (wi::rtx_to_poly_wide_ref): New typedef
2465 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2466 (poly_int_rtx_p): New functions.
2467 (trunc_int_for_mode): Declare a poly_int64 version.
2468 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2469 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2471 (strip_offset): Declare.
2472 (strip_offset_and_add): New function.
2473 * rtl.def (CONST_POLY_INT): New rtx code.
2474 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2475 (shared_const_p): Use poly_int_rtx_p.
2476 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2478 (gen_int_shift_amount): Likewise.
2479 * emit-rtl.c (const_poly_int_hasher): New class.
2480 (const_poly_int_htab): New variable.
2481 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2482 (const_poly_int_hasher::hash): New function.
2483 (const_poly_int_hasher::equal): Likewise.
2484 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2485 (immed_wide_int_const): Rename to...
2486 (immed_wide_int_const_1): ...this and make static.
2487 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2488 instead of a wide_int_ref.
2489 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2490 (gen_lowpart_common): Handle CONST_POLY_INT.
2491 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2492 * cselib.c (cselib_hash_rtx): Likewise.
2493 * dwarf2out.c (const_ok_for_output_1): Likewise.
2494 * expr.c (convert_modes): Likewise.
2495 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2496 * rtlhash.c (add_rtx): Likewise.
2497 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2498 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2499 Handle existing CONST_POLY_INT rtxes.
2500 * expmed.h (expand_shift): Take a poly_int64 instead of a
2502 * expmed.c (expand_shift): Likewise.
2503 * rtlanal.c (strip_offset): New function.
2504 (commutative_operand_precedence): Give CONST_POLY_INT the same
2505 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2507 * rtl-tests.c (const_poly_int_tests): New struct.
2508 (rtl_tests_c_tests): Use it.
2509 * simplify-rtx.c (simplify_const_unary_operation): Handle
2511 (simplify_const_binary_operation): Likewise.
2512 (simplify_binary_operation_1): Fold additions of symbolic constants
2513 and CONST_POLY_INTs.
2514 (simplify_subreg): Handle extensions and truncations of
2516 (simplify_const_poly_int_tests): New struct.
2517 (simplify_rtx_c_tests): Use it.
2518 * wide-int.h (storage_ref): Add default constructor.
2519 (wide_int_ref_storage): Likewise.
2520 (trailing_wide_ints): Use GTY((user)).
2521 (trailing_wide_ints::operator[]): Add a const version.
2522 (trailing_wide_ints::get_precision): New function.
2523 (trailing_wide_ints::extra_size): Likewise.
2525 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2526 Alan Hayward <alan.hayward@arm.com>
2527 David Sherwood <david.sherwood@arm.com>
2529 * emit-rtl.h (gen_int_shift_amount): Declare.
2530 * emit-rtl.c (gen_int_shift_amount): New function.
2531 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2533 * calls.c (shift_return_value): Likewise.
2534 * cse.c (fold_rtx): Likewise.
2535 * dse.c (find_shift_sequence): Likewise.
2536 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2537 (expand_shift, expand_smod_pow2): Likewise.
2538 * lower-subreg.c (shift_cost): Likewise.
2539 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2540 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2541 (expand_vec_perm_var): Likewise.
2542 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2543 (simplify_binary_operation_1): Likewise.
2544 * combine.c (try_combine, find_split_point, force_int_to_mode)
2545 (simplify_shift_const_1, simplify_shift_const): Likewise.
2546 (change_zero_ext): Likewise. Use simplify_gen_binary.
2548 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2550 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2552 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2554 * doc/invoke.texi (ARM Options): Document accepted extension options
2555 for -march=armv8.3-a.
2557 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2560 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2561 or --with-float=softfp, set the default CPU to arm10e.
2563 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2565 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2566 * config/visium/predicates.md (const_shift_operand): Likewise.
2567 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2568 (visium_legitimize_reload_address): Likewise.
2570 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2573 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2575 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2578 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2580 2017-12-20 Julia Koval <julia.koval@intel.com>
2582 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2583 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2584 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2585 * config.gcc: Include vpclmulqdqintrin.h.
2586 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2587 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2589 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2590 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2591 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2592 * config/i386/i386.c (isa2_opts): Add -mcx16.
2593 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2594 (ix86_option_override_internal): Move mcx16 to flags2.
2595 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2596 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2597 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2598 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2599 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2600 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2601 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2602 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2603 * doc/invoke.texi: Add -mvpclmulqdq.
2605 2017-12-20 Tom de Vries <tom@codesourcery.com>
2608 * config/i386/i386.c (ix86_static_chain): Move
2609 DECL_STATIC_CHAIN test ...
2610 * calls.c (rtx_for_static_chain): ... here. New function.
2611 * calls.h (rtx_for_static_chain): Declare.
2612 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2613 instead of targetm.calls.static_chain.
2614 * df-scan.c (df_get_entry_block_def_set): Same.
2616 2017-12-19 Tom de Vries <tom@codesourcery.com>
2618 PR tree-optimization/83493
2619 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2622 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2624 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2625 inform with hardcoded english plural handling.
2627 2017-12-18 Jeff Law <law@redhat.com>
2629 PR tree-optimization/83477
2630 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2631 a non-virtual PHI, always push a new range.
2633 2017-12-19 Martin Sebor <msebor@redhat.com>
2636 * builtins.c (compute_objsize): Handle non-constant offsets.
2638 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2640 PR tree-optimization/83444
2641 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2642 character load case, if get_stridx on MEM_REF's operand doesn't
2643 look usable, retry with get_addr_stridx.
2645 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2648 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2649 (variable_tracking_main_1): Keep markers even when VTA fails.
2652 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2653 even if there are markers before it.
2654 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2656 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2658 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2659 typical order conditions.
2660 * sel-sched.c (extract_new_fences_from): Likewise.
2661 * config/visium/constraints.md (J, K, L): Likewise.
2662 * config/visium/predicates.md (const_shift_operand): Likewise.
2663 * config/visium/visium.c (visium_legitimize_address,
2664 visium_legitimize_reload_address): Likewise.
2665 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2666 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2667 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2668 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2669 SET_DUMP_DETAIL): Likewise.
2670 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2671 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2672 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2673 avr_set_core_architecture, avr_set_current_function,
2674 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2675 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2676 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2677 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2678 avr_map_decompose, avr_fold_builtin): Likewise.
2679 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2680 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2681 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2682 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2683 * config/m32c/m32c.c (m32c_conditional_register_usage,
2684 m32c_address_cost): Likewise.
2685 * config/m32c/predicates.md (shiftcount_operand,
2686 longshiftcount_operand): Likewise.
2687 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2688 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2689 can_use_cdx_ldstw): Likewise.
2690 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2691 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2693 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2694 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2695 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2696 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2697 * config/vax/vax.c (adjacent_operands_p): Likewise.
2698 * config/ft32/constraints.md (L, b, KA): Likewise.
2699 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2701 * cfgexpand.c (expand_stack_alignment): Likewise.
2702 * gcse.c (insert_expr_in_table): Likewise.
2703 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2704 * cgraphunit.c (cgraph_node::expand): Likewise.
2705 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2706 * emit-rtl.c (add_insn): Likewise.
2707 * input.c (dump_location_info): Likewise.
2708 * passes.c (NEXT_PASS): Likewise.
2709 * read-rtl-function.c (parse_note_insn_name,
2710 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2712 * sched-rgn.c (sched_rgn_init): Likewise.
2713 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2714 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2715 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2717 * explow.c (eliminate_constant_term): Likewise.
2718 * final.c (leaf_renumber_regs_insn): Likewise.
2719 * cfgrtl.c (print_rtl_with_bb): Likewise.
2720 * genhooks.c (emit_init_macros): Likewise.
2721 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2722 * tree-data-ref.c (conflict_fn): Likewise.
2723 * selftest.c (assert_streq): Likewise.
2724 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2725 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2726 fold_binary_loc, multiple_of_p): Likewise.
2727 * reload.c (push_reload, find_equiv_reg): Likewise.
2728 * et-forest.c (et_nca, et_below): Likewise.
2729 * dbxout.c (dbxout_symbol_location): Likewise.
2730 * reorg.c (relax_delay_slots): Likewise.
2731 * dojump.c (do_compare_rtx_and_jump): Likewise.
2732 * gengtype-parse.c (type): Likewise.
2733 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2734 simplify_const_relational_operation): Likewise.
2735 * reload1.c (do_output_reload): Likewise.
2736 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2737 * gengtype.c (type_for_name): Likewise.
2738 * gimple-ssa-sprintf.c (format_directive): Likewise.
2740 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2743 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2744 accessing it. Adjust comment.
2746 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2749 * predict.c (zero_one_minusone): New function.
2750 (apply_return_prediction): Avoid return prediction for functions
2751 returning only -1, 0 and 1 values, unless they only return -1 and 0
2754 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2756 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2757 fall-through warning.
2758 (arc_compute_frame_size): Remove unused variables.
2759 (arc_print_operand): Fix fprintif format.
2760 (arc_can_follow_jump): Clean fall-through warning.
2762 2017-12-19 Marek Polacek <polacek@redhat.com>
2765 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2768 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2770 * config/arc/arc.c (overriderregs): New variable.
2771 (arc_override_options): Track fixed/call saved/call options.
2772 (arc_conditional_register_usage): Check against overrideregs
2773 variable whenever we change register properties.
2775 2017-12-19 Nathan Sidwell <nathan@acm.org>
2777 * opts.c (finish_options): Don't prefix dump_base_name if it
2778 already contains directories.
2780 2017-12-19 Martin Liska <mliska@suse.cz>
2782 PR rtl-optimization/82675
2783 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2784 more element in sbitmap.
2786 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2788 * gimplify.c (gimplify_expr): Use error_operand_p.
2790 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2793 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2796 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2798 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2799 the jump is to a label.
2801 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2803 PR tree-optimization/83444
2804 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2809 * ipa-inline.c (flatten_remove_node_hook): New function.
2810 (ipa_inline): Keep only nodes with flatten attribute at the end of
2811 the array in the order from ipa_reverse_postorder, only walk that
2812 portion of array for flattening, if there is more than one such
2813 node, temporarily register a removal hook and ignore removed nodes.
2815 PR tree-optimization/80631
2816 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2817 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2818 IFN_REDUC_MAX or IFN_REDUC_MIN.
2820 2017-12-18 Jeff Law <law@redhat.com>
2822 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2823 record symbolic equivalences from backedges in the CFG.
2826 2017-11-19 Jeff Law <law@redhat.com>
2828 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2829 of degenerates resulting from ignoring an edge.
2831 2017-12-18 Martin Sebor <msebor@redhat.com>
2834 PR tree-optimization/78450
2835 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2836 (handle_builtin_strlen): Call it.
2838 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2840 PR rtl-optimization/83424
2841 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2843 2017-12-18 Marek Polacek <polacek@redhat.com>
2846 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2847 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2850 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2852 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2853 (umaddsidi4, umaddsidi_split): Likewise.
2855 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2857 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2860 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2863 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2865 2017-12-18 Richard Biener <rguenther@suse.de>
2867 PR tree-optimization/81877
2868 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2869 (outermost_indep_loop): Adjust.
2870 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2871 (can_sm_ref_p): Adjust.
2873 2017-12-18 Richard Biener <rguenther@suse.de>
2876 * tree.c (array_at_struct_end_p): Return true if the underlying
2877 object has space for at least one element in excess of what
2878 the array domain specifies.
2880 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2882 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2883 example to avoid overfull hbox.
2884 * doc/invoke.texi (Option Summary): Add missing @gol.
2885 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2888 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2889 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2891 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2892 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2893 (AArch64 Options): Document -mverbose-cost-dump.
2894 (ARM Options): Likewise, plus -mflip-thumb.
2896 2017-12-17 Martin Sebor <msebor@redhat.com>
2899 * gimple-ssa-warn-restrict.c
2900 (builtin_memref::offset_out_of_bounds): Correct the handling of
2903 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2905 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2907 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2909 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2911 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2913 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2916 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2918 * doc/invoke.texi: Fix some typos.
2920 2017-12-16 Martin Sebor <msebor@redhat.com>
2922 PR tree-optimization/78918
2923 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2924 * builtins.c (check_sizes): Rename...
2925 (check_access): ...to this. Rename function arguments for clarity.
2926 (check_memop_sizes): Adjust names.
2927 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2928 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2929 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2930 (check_strncat_sizes, expand_builtin_strncat): Same.
2931 (expand_builtin_strncpy, expand_builtin_memset): Same.
2932 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2933 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2934 (maybe_emit_sprintf_chk_warning): Same.
2935 (expand_builtin_strcpy): Adjust.
2936 (expand_builtin_stpcpy): Same.
2937 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2938 in pointer-checking forms of memcpy, memmove, and mempcpy.
2939 (gcall_to_tree_minimal, max_object_size): Define new functions.
2940 * builtins.h (max_object_size): Declare.
2941 * calls.c (alloc_max_size): Call max_object_size instead of
2942 hardcoding ssizetype limit.
2943 (get_size_range): Handle new argument.
2944 * calls.h (get_size_range): Add a new argument.
2945 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2946 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2947 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2949 (gimple_fold_builtin_memory_chk): Same.
2950 (gimple_fold_builtin_stxcpy_chk): New function.
2951 * gimple-ssa-warn-restrict.c: New source.
2952 * gimple-ssa-warn-restrict.h: New header.
2953 * gimple.c (gimple_build_call_from_tree): Propagate location.
2954 * passes.def (pass_warn_restrict): Add new pass.
2955 * tree-pass.h (make_pass_warn_restrict): Declare.
2956 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2958 (handle_builtin_strcat): Same.
2959 (strlen_optimize_stmt): Rename...
2960 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2961 stpncpy, strncpy, and their checking forms.
2963 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2965 PR rtl-optimization/82849
2966 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2967 and get_max_loop_iterations_int.
2969 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2970 Alan Hayward <alan.hayward@arm.com>
2971 David Sherwood <david.sherwood@arm.com>
2973 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2974 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2975 size as a poly_uint64.
2976 (mode_for_vector, mode_for_int_vector): Take the number of vector
2977 elements as a poly_uint64.
2978 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2979 size as a poly_uint64.
2980 (mode_for_vector, mode_for_int_vector): Take the number of vector
2981 elements as a poly_uint64.
2983 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2984 Alan Hayward <alan.hayward@arm.com>
2985 David Sherwood <david.sherwood@arm.com>
2987 * machmode.h (MACRO_MODE): New macro.
2988 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2989 * targhooks.c (default_libcall_value, default_secondary_reload)
2990 (default_memory_move_cost, default_register_move_cost)
2991 (default_class_max_nregs): Likewise.
2993 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2994 Alan Hayward <alan.hayward@arm.com>
2995 David Sherwood <david.sherwood@arm.com>
2997 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2998 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3000 * genautomata.c (main): Likewise.
3001 * genconditions.c (write_header): Likewise.
3002 * genemit.c (main): Likewise.
3003 * genextract.c (print_header): Likewise.
3004 * genopinit.c (main): Likewise.
3005 * genoutput.c (output_prologue): Likewise.
3006 * genpeep.c (main): Likewise.
3007 * genpreds.c (write_insn_preds_c): Likewise.
3008 * genrecog.c (writer_header): Likewise.
3009 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3010 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3011 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3012 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3013 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3014 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3015 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3016 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3017 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3018 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3019 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3020 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3021 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3022 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3023 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3024 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3025 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3026 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3027 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3028 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3029 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3030 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3031 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3032 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3033 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3034 * config/darwin.c (IN_TARGET_CODE): Likewise.
3035 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3036 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3037 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3038 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3039 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3040 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3041 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3042 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3043 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3044 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3045 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3046 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3047 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3048 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3049 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3050 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3051 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3052 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3053 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3054 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3055 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3056 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3057 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3058 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3059 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3060 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3061 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3062 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3063 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3064 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3065 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3066 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3067 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3068 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3069 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3070 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3071 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3072 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3073 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3074 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3075 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3076 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3077 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3078 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3079 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3080 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3081 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3082 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3083 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3084 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3085 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3086 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3087 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3088 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3089 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3090 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3091 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3092 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3093 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3094 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3095 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3096 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3097 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3098 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3099 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3100 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3101 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3102 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3103 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3104 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3105 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3106 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3107 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3108 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3109 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3110 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3111 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3112 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3113 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3114 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3115 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3116 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3117 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3118 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3119 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3120 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3121 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3122 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3123 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3124 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3125 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3126 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3127 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3128 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3129 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3130 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3131 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3132 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3133 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3134 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3135 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3136 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3137 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3138 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3139 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3140 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3142 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3143 Alan Hayward <alan.hayward@arm.com>
3144 David Sherwood <david.sherwood@arm.com>
3146 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3149 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3150 Alan Hayward <alan.hayward@arm.com>
3151 David Sherwood <david.sherwood@arm.com>
3153 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3154 * doc/md.texi (vec_series@var{m}): Document.
3155 * tree.def (VEC_SERIES_EXPR): New tree code.
3156 * tree.h (build_vec_series): Declare.
3157 * tree.c (build_vec_series): New function.
3158 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3159 * tree-pretty-print.c (dump_generic_node): Likewise.
3160 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3161 * tree-inline.c (estimate_operator_cost): Likewise.
3162 * expr.c (expand_expr_real_2): Likewise.
3163 * optabs-tree.c (optab_for_tree_code): Likewise.
3164 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3165 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3166 * expmed.c (make_tree): Handle VEC_SERIES.
3167 * optabs.def (vec_series_optab): New optab.
3168 * optabs.h (expand_vec_series_expr): Declare.
3169 * optabs.c (expand_vec_series_expr): New function.
3170 * tree-vect-generic.c (expand_vector_operations_1): Check that
3171 the operands also have vector type.
3173 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3174 Alan Hayward <alan.hawyard@arm.com>
3175 David Sherwood <david.sherwood@arm.com>
3177 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3178 (VEC_COND_EXPR): Add missing @tindex.
3179 * doc/md.texi (vec_duplicate@var{m}): Document.
3180 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3181 * tree.c (build_vector_from_val): Add stubbed-out handling of
3182 variable-length vectors, using VEC_DUPLICATE_EXPR.
3183 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3184 * cfgexpand.c (expand_debug_expr): Likewise.
3185 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3186 * tree-inline.c (estimate_operator_cost): Likewise.
3187 * tree-pretty-print.c (dump_generic_node): Likewise.
3188 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3189 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3190 (test_vec_duplicate_folding): New function.
3191 (fold_const_c_tests): Call it.
3192 * optabs.def (vec_duplicate_optab): New optab.
3193 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3194 * optabs.h (expand_vector_broadcast): Declare.
3195 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3196 vec_duplicate_optab.
3197 * expr.c (store_constructor): Try using vec_duplicate_optab for
3199 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3201 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3204 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3205 div/mod latencies a bit.
3207 2017-12-15 Jeff Law <law@redhat.com>
3209 PR tree-optimization/36550
3210 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3211 (mark_threaded_blocks): Rewrite code to avoid block copying when
3212 optimizing for size. Don't pessimize blocks which will be
3213 copied, but all the statements will be dead.
3215 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3217 PR tree-optimization/81165
3218 * tree-ssa-threadupdate.c (uses_in_bb): New.
3219 (estimate_threading_killed_stmts): New.
3220 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3221 * tree-ssa-threadedge.c
3222 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3225 2017-12-15 Jeff Law <law@redhat.com>
3227 PR tree-optimization/83410
3228 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3229 threads when parallelizing loops.
3231 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3233 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3235 * config/alpha/alpha.c (vms_attribute_table): Swap
3236 affects_type_identity and handler fields, adjust comments.
3237 * config/mips/mips.c (mips_attribute_table): Likewise.
3238 * config/visium/visium.c (visium_attribute_table): Likewise.
3239 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3240 * config/microblaze/microblaze.c (microblaze_attribute_table):
3242 * config/spu/spu.c (spu_attribute_table): Likewise.
3243 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3244 * config/arc/arc.c (arc_attribute_table): Likewise.
3245 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3246 * config/v850/v850.c (v850_handle_interrupt_attribute,
3247 v850_handle_data_area_attribute): Formatting fixes.
3248 (v850_attribute_table): Swap affects_type_identity and handler
3249 fields, adjust comments.
3250 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3251 * config/arm/arm.c (arm_attribute_table): Likewise.
3252 * config/avr/avr.c (avr_attribute_table): Likewise.
3253 * config/s390/s390.c (s390_attribute_table): Likewise.
3254 * config/sh/sh.c (sh_attribute_table): Likewise.
3255 * config/i386/i386.c (ix86_handle_cconv_attribute,
3256 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3257 (ix86_attribute_table): Swap affects_type_identity and handler
3258 fields, adjust comments.
3259 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3260 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3261 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3262 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3263 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3264 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3265 * config/rx/rx.c (rx_attribute_table): Likewise.
3266 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3267 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3268 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3269 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3270 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3271 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3272 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3273 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3274 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3275 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3276 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3278 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3280 (register_scoped_attributes, decl_attributes): Formatting fixes.
3282 PR tree-optimization/83269
3283 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3284 subtraction in arg0's type if type is signed and arg0 is unsigned.
3288 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3289 convert? on @0 instead of convert. Check type of @1, not @0.
3290 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3291 convert? on @0 instead of convert. Check type of @1, not @0.
3292 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3293 only optimize if either both @1 and @2 types are narrower
3294 precision, or both are wider or equal precision, and in the former
3295 case only if both have undefined overflow.
3297 2017-12-15 Richard Biener <rguenther@suse.de>
3300 * internal-fn.def (IFN_NOP): Add.
3301 * internal-fn.c (expand_NOP): Do nothing.
3302 * lto-streamer-in.c (input_function): Instead of removing
3303 sanitizer calls replace them with IFN_NOP calls.
3305 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3306 Alan Hayward <alan.hayward@arm.com>
3307 David Sherwood <david.sherwood@arm.com>
3309 * dse.c (store_info, read_info_type): Replace begin and end with
3311 (print_range): New function.
3312 (set_all_positions_unneeded, any_positions_needed_p)
3313 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3315 (record_store): Likewise. Optimize the case in which all positions
3317 (get_stored_val): Replace read_begin and read_end with read_offset
3319 (replace_read): Update call accordingly.
3321 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3323 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3324 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3325 (loop_cand::loop_cand): Initialize above members.
3326 (loop_cand::supported_operations): Delete.
3327 (loop_cand::can_interchange_p): Inline above function.
3328 (loop_cand::classify_simple_reduction): Record number of constant
3329 initialized simple reductions.
3330 (should_interchange_loops): New parameters. Check stmt cost of loops
3332 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3333 Update call to should_interchange_loops.
3334 (should_interchange_loop_nest): Update call to
3335 should_interchange_loops.
3337 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3340 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3341 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3342 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3344 2017-12-15 Julia Koval <julia.koval@intel.com>
3346 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3347 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3349 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3350 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3351 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3353 2017-12-15 Julia Koval <julia.koval@intel.com>
3355 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3356 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3357 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3358 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3359 _mm_aesenc_epi128): New intrinsics.
3361 2017-12-15 Julia Koval <julia.koval@intel.com>
3363 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3364 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3366 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3367 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3368 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3370 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3372 * gimple-ssa-strength-reduction.c (analyze_increments):
3373 Distinguish replacement costs for constant strides from those for
3376 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3378 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3380 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3382 * doc/invoke.texi: Document -Wcast-function-type.
3383 * recog.h (stored_funcptr): Change signature.
3384 * tree-dump.c (dump_node): Avoid warning.
3385 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3387 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3390 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3393 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3395 PR tree-optimization/83312
3396 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3397 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3398 "val" param, and to cope with arbitrary basic blocks.
3399 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3400 handle NULL_TREE for "val", dropping "bb" param.
3401 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3402 drop "bb" param. Handle NULL_TREE for "val".
3403 (find_case_label_for_value): Make "switch_stmt" param const.
3404 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3406 (vrp_prop::check_all_array_refs): Reimplement as...
3407 (check_array_bounds_dom_walker::before_dom_children): ...this new
3408 vfunc. Replace linear search through BB block list, excluding
3409 those with non-executable in-edges via dominator walk.
3411 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3413 * config/arm/arm.opt (mverbose-cost-dump): New option.
3414 * config/arm/arm.c (arm_rtx_costs): Use it.
3416 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3419 * reload1.c (emit_input_reload_insns): Skip debug markers.
3421 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3423 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3427 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3428 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3429 * config/sh/sh.c: Skip debug insns besides notes.
3430 * config/sh/sh.md: Likewise.
3431 * config/sh/sh_treg_combine.cc: Likewise.
3432 * config/sh/sync.md: Likewise.
3434 2017-12-14 Tom de Vries <tom@codesourcery.com>
3436 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3437 Add item for weak_undefined.
3439 2017-12-14 Richard Biener <rguenther@suse.de>
3441 PR tree-optimization/67842
3442 * sese.h (bb_in_region): Remove #if 0'ed code.
3444 2017-12-14 Richard Biener <rguenther@suse.de>
3446 PR tree-optimization/83326
3447 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3448 may_be_zero parameter and handle it by not marking the first
3449 peeled copy as not exiting the loop.
3450 (try_peel_loop): Likewise.
3451 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3452 to handle the case of constant or zero iterations and perform
3453 loop header copying on-the-fly.
3455 2017-12-14 Richard Biener <rguenther@suse.de>
3457 PR tree-optimization/83418
3458 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3459 Instead of asserting we don't get unfolded comparisons deal with them.
3461 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3464 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3465 basic blocks. Assert debug bind insns don't appear outside of bbs,
3466 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3469 PR tree-optimization/83198
3470 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3471 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3472 value if arg is a REAL_CST with incompatible type.
3474 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3475 Bin Cheng <bin.cheng@arm.com>
3478 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3480 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3483 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3485 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3486 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3488 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3489 Alan Hayward <alan.hayward@arm.com>
3490 David Sherwood <david.sherwood@arm.com>
3492 * poly-int.h: New file.
3493 * poly-int-types.h: Likewise.
3494 * coretypes.h: Include them.
3495 (POLY_INT_CONVERSION): Define.
3496 * target.def (estimated_poly_value): New hook.
3497 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3498 * doc/tm.texi: Regenerate.
3499 * doc/poly-int.texi: New file.
3500 * doc/gccint.texi: Include it.
3501 * doc/rtl.texi: Describe restrictions on subreg modes.
3502 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3503 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3504 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3505 * targhooks.h (default_estimated_poly_value): Declare.
3506 * targhooks.c (default_estimated_poly_value): New function.
3507 * target.h (estimated_poly_value): Likewise.
3508 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3509 (wi::unary_traits): Delete.
3510 (wi::binary_traits::signed_shift_result_type): Define for
3511 offset_int << HOST_WIDE_INT, etc.
3512 (generic_wide_int::operator <<=): Define for all types and use
3513 wi::lshift instead of <<.
3514 (wi::hwi_with_prec): Add a default constructor.
3515 (wi::ints_for): New class.
3516 (operator <<): Define for all wide-int types.
3517 (operator /): New function.
3518 (operator %): Likewise.
3519 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3520 (ASSERT_MAYBE_NE_AT): New macros.
3522 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3523 Dominik Vogt <vogt@linux.vnet.ibm.com>
3526 * emit-rtl.c (init_emit): Remove ??? comment.
3527 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3528 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3529 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3530 alignment of 3 virtual registers to BITS_PER_WORD.
3532 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3534 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3536 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3537 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3539 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3542 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3544 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3546 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3547 with the exception of debug begin stmt markers appear before
3551 * final.c (rest_of_handle_final): Call variable_tracking_main only
3552 if !flag_var_tracking.
3554 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3555 Jakub Jelinek <jakub@redhat.com>
3559 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3560 labels debug stmts that can only appear after labels.
3562 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3564 PR rtl-optimization/82398
3565 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3566 EXPR_USEFULNESS in priority comparison.
3568 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3570 PR rtl-optimization/83393
3571 * combine.c (move_deaths): If reg_stat points to a too new insn in
3572 last_death, do not use it: find the proper insn instead.
3574 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3577 * config/rs6000/rs6000-string.c (expand_block_compare,
3578 expand_strn_compare): Fix set_mem_size() calls.
3580 2017-12-12 Jeff Law <law@redhat.com>
3582 PR tree-optimization/83298
3583 PR tree-optimization/83362
3584 PR tree-optimization/83383
3585 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3586 push_value_range a public interface. Add new argument to
3587 record_ranges_from_stmt.
3588 * gimple-ssa-evrp-analyze.c
3589 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3590 Update comments. Handle recording temporary equivalences.
3591 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3592 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3593 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3594 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3595 gimple-ssa-evrp-analyze.h.
3596 (record_temporary_equivalences_from_phis): Add new argument. When
3597 the PHI arg is an SSA_NAME, set the result's range to the range
3599 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3600 from statements too.
3601 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3602 Pass it down to children as needed.
3603 (thread_outgoing_edges): Likewise.
3604 (thread_across_edge): Likewise. Push/pop range state as needed.
3605 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3607 2017-12-12 Julia Koval <julia.koval@intel.com>
3609 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3610 (PTA_CANNONLAKE): Remove PTA_CLWB.
3612 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3615 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3616 (vcondv2div2df): Likewise.
3617 (vconduv2dfv2di): Likewise.
3619 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3621 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3622 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3623 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3624 ATTR_MATHFN_FPROUNDING_ERRNO.
3626 2017-12-12 Richard Biener <rguenther@suse.de>
3628 PR tree-optimization/83385
3629 * graphite-scop-detection.c (get_order, order): Remove.
3630 (bb_to_rpo): New global.
3632 (build_scops): Sort pbbs in RPO order.
3634 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3636 * combine.c (simplify_set): Do not transform subregs to zero_extends
3637 if the destination mode is a vector mode.
3639 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3641 PR tree-optimization/83359
3642 * tree-cfg.h (fold_loop_internal_call): Declare.
3643 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3644 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3645 (find_loop_dist_alias): New function.
3646 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3647 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3650 PR tree-optimization/80631
3651 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3652 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3653 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3654 hardcoding zero as the value if COND_EXPR is never true. For
3655 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3656 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3657 hardcoding MAX_EXPR as the reduction operation.
3658 (is_nonwrapping_integer_induction): Allow negative step.
3659 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3660 vect_create_epilog_for_reduction, if no value is suitable, don't
3661 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3663 2017-12-12 Richard Biener <rguenther@suse.de>
3665 PR tree-optimization/81889
3666 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3667 range info from the non-wrapping IV instead of just the range
3670 2017-12-12 Julia Koval <julia.koval@intel.com>
3672 * config.gcc: Add vaesintrin.h.
3673 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3675 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3676 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3678 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3679 * config/i386/immintrin.h: Include vaesintrin.h.
3680 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3681 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3682 _mm_aesdec_epi128): New intrinsics.
3684 2017-12-12 Julia Koval <julia.koval@intel.com>
3686 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3687 OPTION_MASK_ISA_VAES_UNSET): New.
3688 (ix86_handle_option): Handle -mvaes.
3689 * config/i386/cpuid.h: Define bit_VAES.
3690 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3691 * config/i386/i386-c.c (__VAES__): New.
3692 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3693 (ix86_valid_target_attribute_inner_p): Ditto.
3694 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3695 * config/i386/i386.opt: Add -mvaes.
3696 * doc/invoke.texi: Ditto.
3698 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3700 * debug.h (gcc_debug_hooks): Add inline_entry.
3701 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3702 * debug.c (do_nothing_debug_hooks): Likewise.
3703 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3704 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3705 (dwarf2_lineno_debug_hooks): Likewise.
3707 * common.opt (gstatement-frontiers): New, setting
3708 debug_nonbind_markers_p.
3709 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3710 * toplev.c (process_options): Autodetect value for debug statement
3712 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3713 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3715 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3716 markers. Integrate source bind into debug stmt expand loop.
3717 (pass_expand::execute): Check debug marker limit. Avoid deep
3718 TER and expand debug locations for debug bind insns only.
3719 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3721 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3722 * final.c (reemit_insn_block_notes): Take current block from
3723 nonbind markers. Declare note where it's first set.
3724 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3725 begin stmt markers if enabled.
3726 (notice_source_line): Handle nonbind markers. Fail if their
3727 location is unknown or that of builtins.
3728 (rest_of_handle_final): Convert begin stmt markers to notes if
3729 var-tracking didn't run.
3730 (rest_of_clean_state): Skip begin stmt markers.
3731 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3733 * function.c (allocate_struct_function): Set begin_stmt_markers.
3734 * function.h (struct function): Add debug_marker_count counter
3735 and debug_nonbind_markers flag.
3736 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3737 * gimple-low.c (lower_function_body): Adjust
3738 debug_nonbind_markers.
3739 (lower_stmt): Drop or skip gimple debug stmts.
3740 (lower_try_catch): Skip debug stmts.
3741 * gimple.c (gimple_build_debug_begin_stmt): New.
3742 (gimple_copy): Increment debug_marker_count if copying one.
3743 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3744 * gimplify.c (rexpr_location): New.
3745 (rexpr_has_location): New.
3746 (warn_switch_unreachable_r): Handle gimple debug stmts.
3747 (shortcut_cond_r): Call expr_location.
3749 (find_goto_label): New.
3750 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3752 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3754 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3755 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3756 (LANG_HOOKS_INITIALIZER): ... this.
3757 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3758 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3760 (update_ebb_live_info): Skip debug insn markers.
3761 * lra.c (debug_insn_static_data): Rename to...
3762 (debug_bind_static_data): ... this.
3763 (debug_marker_static_data): New.
3764 (lra_set_insn_recog_data): Select one of the above depending
3766 (lra_update_isn_regno_info): Don't assume debug insns have
3768 (push_insns): Skip debug insns.
3769 * lto-streamer-in.c (input_function): Drop debug stmts
3770 depending on active options. Adjust debug_nonbind_markers.
3771 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3772 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3773 begin stmt marker notes.
3774 (print_insn): Likewise.
3775 * recog.c (extract_insn): Recognize rtl for debug markers.
3776 * rtl.def (DEBUG_MARKER): New.
3777 * tree-inline.c: Include params.h.
3778 (remap_gimple_stmt): Handle nonbind markers.
3779 (maybe_move_debug_stmts_to_successors): Likewise.
3780 (copy_debug_stmt): Likewise.
3781 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3782 markers regardless of no side effects.
3783 (tsi_link_before): Don't update container's side effects when adding
3784 a begin stmt marker.
3785 (tsi_link_after): Likewise.
3786 (expr_first): Skip begin stmt markers.
3787 (expr_last): Likewise.
3788 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3789 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3790 Disregard nonbind markers.
3791 * tree.c (make_node_stat): Don't set side effects for begin stmt
3793 (build1_stat): Likewise.
3794 * tree.def (DEBUG_BEGIN_STMT): New.
3795 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3796 * var-tracking.c (delete_debug_insns): Renamed to...
3797 (delete_vta_debug_insns): ... this.
3798 (reemit_marker_as_note): New.
3799 (vt_initialize): Reemit markers.
3800 (delete_vta_debug_insns): Likewise.
3801 (vt_debug_insns_local): Reemit or delete markers.
3802 (variable_tracking_main_1): Likewise.
3803 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3804 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3805 (gimple_debug_nonbind_marker_p): New.
3806 (gimple_build_debug_bind): Adjust.
3807 (gimple_build_debug_begin_stmt): New.
3808 * doc/invoke.texi (max-debug-marker-count): New param.
3809 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3810 (debug_parameter_ref, debug_marker): New.
3811 (NOTE_INSN_BEGIN_STMT): New.
3812 (DEBUG_INSN): Describe begin stmt markers.
3814 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3815 without debug insns, we wouldn't, but clean up debug insns
3816 after a control flow insn nevertheless.
3818 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3819 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3820 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3821 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3822 (rtl_tidy_fallthru_edge): Likewise.
3823 (rtl_verify_fallthru): Likewise.
3824 (rtl_verify_bb_layout): Likewise.
3825 (skip_insns_after_block): Likewise.
3826 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3827 * dwarf2out.c: Include print-rtl.h.
3828 (dwarf2out_next_real_insn): New.
3829 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3830 Dump debug binds in asm comments.
3831 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3832 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3833 callers to use gsi_start_nondebug_bb instead.
3834 (gsi_after_labels): Skip gimple debug stmts.
3835 (gsi_start_nondebug): New.
3836 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3837 (proper_loop_form_for_interchange): Adjust.
3838 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3839 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3840 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3841 (collect_fallthrough_labels): Likewise.
3842 (should_warn_for_implicit_fallthrough): Likewise.
3843 (warn_implicit_fallthrough_r): Likewise.
3844 (expand_FALLTHROUGH_r): Likewise.
3845 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3846 (graphite_copy_stmts_from_block): Skip nonbind markers.
3847 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3848 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3849 * jump.c (clean_barriers): Skip debug insns.
3850 * omp-expand.c (expand_parallel_call): Skip debug insns.
3851 (expand_task_call): Likewise.
3852 (remove_exit_barrier): Likewise.
3853 (expand_omp_taskreg): Likewise.
3854 (expand_omp_for_init_counts): Likewise.
3855 (expand_omp_for_generic): Likewise.
3856 (expand_omp_for_static_nochunk): Likewise.
3857 (expand_omp_for_static_chunk): Likewise.
3858 (expand_omp_simd): Likewise.
3859 (expand_omp_taskloop_for_outer): Likewise.
3860 (expand_omp_taskloop_for_inner): Likewise.
3861 (expand_oacc_for): Likewise.
3862 (expand_omp_sections): Likewise.
3863 (expand_omp_single): Likewise.
3864 (expand_omp_synch): Likewise.
3865 (expand_omp_atomic_load): Likewise.
3866 (expand_omp_atomic_store): Likewise.
3867 (expand_omp_atomic_fetch_op): Likewise.
3868 (expand_omp_atomic_pipeline): Likewise.
3869 (expand_omp_atomic_mutex): Likewise.
3870 (expand_omp_target): Likewise.
3871 (grid_expand_omp_for_loop): Likewise.
3872 (grid_expand_target_grid_body): Likewise.
3873 (build_omp_regions_1): Likewise.
3874 * omp-low.c (check_combined_parallel): Skip debug stmts.
3875 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3876 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3877 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3879 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3880 (make_edges): Likewise.
3881 (cleanup_dead_labels): Likewise.
3882 (gimple_can_merge_blocks_p): Likewise.
3883 (stmt_starts_bb_p): Likewise.
3884 (gimple_block_label): Likewise.
3885 (gimple_redirect_edge_and_branch): Likewise.
3886 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3888 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3890 * tree-pretty-print (print_declaration): Omit initializer in slim
3892 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3894 (eliminate_unnecessary_stmts): Stabilize block removal order.
3895 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3896 * var-tracking.c (get_first_insn): New.
3897 (vt_emit_notes): Call it.
3898 (vt_initialize): Walk any insns before the first BB.
3899 (delete_debug_insns): Likewise.
3901 * gimple.h (enum gimple_debug_subcode): Add
3902 GIMPLE_DEBUG_BEGIN_STMT.
3903 (gimple_debug_begin_stmt_p): New.
3904 (gimple_debug_nonbind_marker_p): New.
3905 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3906 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3907 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3908 * insn-notes.def (BEGIN_STMT): New.
3909 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3910 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3911 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3912 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3913 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3914 (INSN_DEBUG_MARKER_KIND): New.
3915 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3916 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3917 (INSN_VAR_LOCATION_PTR): New.
3918 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3920 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3921 deps only if debug bind insns are enabled.
3922 (pass_expand::execute): Avoid deep TER and expand
3923 debug locations for debug bind insns only.
3924 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3925 debug stmts special handling down to debug bind stmts.
3926 * combine.c (try_combine): Narrow debug insns special handling
3927 down to debug bind insns.
3928 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3929 Narrow debug insns preexisting special handling down to debug
3931 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3932 handling down to debug bind insns.
3933 * function.c (instantiate_virtual_regs): Skip debug markers,
3934 adjust handling of debug binds.
3935 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3936 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3937 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3938 insns to debug bind insns.
3939 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3940 special handling of debug stmts to debug bind stmts.
3941 * ipa-split.c (split_function): Likewise.
3942 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3943 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3945 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3946 * regrename.c (build_def_use): Likewise.
3947 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3948 (pass_cprop_hardreg): Narrow special casing of debug insns to
3950 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3951 * reload1.c (reload): Likewise.
3952 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3953 casing of debug stmts to debug bind stmts.
3954 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3955 * ssa-iterators.h (num_imm_uses): Likewise.
3956 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3957 debug stmts to debug bind stmts.
3958 * tree-inline.c (tree_function_versioning): Narrow special casing
3959 of debug stmts to debug bind stmts.
3960 * tree-loop-distribution.c (generate_loops_for_partition):
3961 Narrow special casing of debug stmts to debug bind stmts.
3962 * tree-sra.c (analyze_access_subtree): Narrow special casing
3963 of debug stmts to debug bind stmts.
3964 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3965 stmts to debug bind stmts.
3966 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3967 casing of debug stmts to debug bind stmts.
3968 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3969 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3970 casing of debug stmts to debug bind stmts.
3971 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3973 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3974 debug stmts to debug bind stmts.
3975 (gimple_replace_ssa_lhs): Likewise.
3976 (insert_debug_temp_for_var_def): Likewise.
3977 (insert_debug_temps_for_defs): Likewise.
3978 (reset_debug_uses): Likewise.
3979 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3980 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3981 (adjust_debug_stmts): Likewise.
3982 (adjust_phi_and_debug_stmts): Likewise.
3983 (vect_do_peeling): Likewise.
3984 * tree-vect-loop.c (vect_transform_loop): Likewise.
3985 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3986 * var-tracking.c (adjust_mems): Narrow special casing of debug
3987 insns to debug bind insns.
3988 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3989 (compute_bb_dataflow, vt_find_locations): Likewise.
3990 (vt_expand_loc, emit_notes_for_changes): Likewise.
3991 (vt_init_cfa_base): Likewise.
3992 (vt_emit_notes): Likewise.
3993 (vt_initialize): Likewise.
3994 (vt_finalize): Likewise.
3996 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3997 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3998 (next_nonnote_nondebug_insn_bb): New.
3999 (prev_nonnote_nondebug_insn_bb): New.
4000 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4001 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4002 (prev_nonnote_nondebug_insn_bb): Declare.
4003 (next_nonnote_nondebug_insn_bb): Declare.
4004 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4005 * cfgrtl.c (get_last_bb_insn): Likewise.
4006 * lra.c (push_insns): Likewise.
4008 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4011 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4013 * selftest.c (selftest::test_runner::test_runner): New ctor.
4014 (selftest::test_runner::~test_runner): New dtor.
4015 * selftest.h (class selftest::test_runner): New class.
4017 2017-12-11 Carl Love <cel@us.ibm.com>
4019 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4020 vec_extract_fp32_from_shortl]): Add #defines.
4021 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4022 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4023 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4024 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4025 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4026 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4027 * doc/extend.texi: Add documentation for the added builtins.
4029 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4031 PR rtl-optimization/80693
4032 PR rtl-optimization/81019
4033 PR rtl-optimization/81020
4034 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4035 are not mentioned in i3. Place the REG_UNUSED note on i2,
4036 possibly modified to REG_DEAD, if it did not originate in i3.
4038 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4040 * recog.c (store_data_bypass_p_1): New function.
4041 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4042 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4044 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4046 PR rtl-optimization/83361
4047 * ifcvt.c (if_convert): Call fixup_partitions.
4049 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4051 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4052 early folding of splat_u{8,16,32}.
4054 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4056 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4057 output_operand_lossage first argument with capital letter.
4058 (aarch64_override_options): Don't start error and sorry first argument
4059 with capital letter.
4061 2017-12-11 Andi Kleen <ak@linux.intel.com>
4063 PR gcov-profile/83355
4064 * auto-profile.c (string_table::get_index_by_decl): Don't
4065 recurse when abstract origin points to itself.
4067 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4069 PR tree-optimization/83320
4070 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4071 (prune_datarefs_not_in_loop): Ditto.
4073 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4075 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4077 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4079 PR tree-optimization/83337
4080 * gimple-loop-interchange.cc (compute_access_stride): Handle
4081 bitfield DRs properly.
4083 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4085 PR tree-optimization/83338
4086 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4087 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4088 vector of unsigned integers to vector of signed integers.
4090 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4092 PR rtl-optimization/83317
4093 * lra-constraints.c (process_address_1): Add insn code check.
4095 2017-12-08 Michael Matz <matz@suse.de>
4097 Fix PR tree-optimization/83323
4098 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4099 head-controlled loops and loop BBs.
4100 * common.opt (funroll-and-jam): Remove, instead ...
4101 (floop-unroll-and-jam): ... reuse this option.
4102 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4103 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4104 (-floop-unroll-and-jam): ... this option.
4106 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4108 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4109 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4110 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4112 2017-12-08 Richard Biener <rguenther@suse.de>
4115 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4116 handle accesses outside of zero-sized vars.
4118 2017-12-08 Martin Jambor <mjambor@suse.cz>
4120 PR tree-optimization/83141
4121 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4122 test for MEM_REFs implicitely changing types with padding. Remove
4124 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4126 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4128 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4130 * config/avr/avr.c (avr_attribute_table): Likewise.
4131 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4132 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4133 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4134 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4135 of struct attribute_spec.
4136 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4138 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4141 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4143 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4146 * profile-count.c (profile_count::from_gcov_type): Move from
4147 profile-count.h; handle overflow.
4148 * profile-count.h (profile_count::from_gcov_type): Move offline.
4150 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4152 PR rtl-optimization/83304
4153 * combine.c (move_deaths): If we do not know where a register died,
4156 2017-12-08 Richard Biener <rguenther@suse.de>
4158 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4159 Provide -fopt-info-loop feedback when we interchange in a nest.
4161 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4163 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4164 for armv6 ARM CPU IDs.
4166 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4168 * common/config/arm/arm-common.c: Include <algorithm>.
4169 (INCLUDE_VECTOR): Define.
4170 (compare_opt_names): New function.
4171 (arm_rewrite_selected_arch): Only strip out extensions that can be
4172 expressed through -mfpu. Sort the remaining extensions
4175 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4177 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4178 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4179 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4180 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4182 2017-06-08 Tristan Gingold <gindold@adacore.com>
4185 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4186 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4188 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4190 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4191 of struct attribute_spec.
4193 2017-12-08 Julia Koval <julia.koval@intel.com>
4195 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4196 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4197 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4198 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4199 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4200 _mm_maskz_dpwssds_epi32): Ditto.
4202 2017-12-08 Richard Biener <rguenther@suse.de>
4204 PR tree-optimization/81303
4205 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4206 conditions try to create a comparison vector type matching
4207 the data vector type.
4208 (vectorizable_condition): Adjust.
4209 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4210 Leave invariant conditions alone in case we can vectorize those.
4212 2017-12-08 Julia Koval <julia.koval@intel.com>
4214 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4215 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4216 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4217 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4218 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4220 2017-12-08 Julia Koval <julia.koval@intel.com>
4222 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4223 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4224 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4225 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4226 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4227 _mm_maskz_dpbusds_epi32): New intrinsics.
4229 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4231 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4233 (ld<bh>io_signed): Likewise.
4234 (st<bhw>io): Likewise.
4235 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4236 SMALL_INT12 constant integer operand.
4238 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4240 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4242 (nios2_large_constant_allowed): ...this. Adjust uses.
4243 (nios2_plus_symbolic_constant_p): Rename to...
4244 (nios2_plus_large_constant_p): ...this. Adjust uses.
4245 (nios2_legitimate_address_p): Correct CONST_INT handling.
4246 (nios2_symbolic_memory_operand_p): Rename to...
4247 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4248 (nios2_large_constant_p): Check for large constant integers too.
4249 (nios2_split_large_constant): Handle constant integers.
4250 (nios2_split_symbolic_memory_operand): Rename to...
4251 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4252 (nios2_legitimize_constant_address): Handle constant integers.
4253 (r0rel_constant_p): Handle small constant integers.
4254 (nios2_print_operand_address): Handle r0-relative integer addresses.
4255 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4256 * config/nios2/nios2.md: Adjust for renamed functions.
4258 2017-12-07 Andrew Waterman <andrew@sifive.com>
4260 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4261 (TARGET_HAVE_SRODATA_SECTION): New define.
4262 (riscv_select_section): New function.
4264 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4265 Alexander Monakov <amonakov@ispras.ru>
4266 Jakub Jelinek <jakub@redhat.com>
4269 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4271 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4273 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4274 a bool success value. Don't call output_operand_lossage here.
4275 (aarch64_print_ldpstp_address): Return a bool success value.
4276 (aarch64_print_operand_address): Call output_addr_const if
4277 aarch64_print_address_internal fails.
4278 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4279 'y'; call output_operand_lossage instead. Call output_operand_lossage
4280 if aarch64_print_ldpstp_address fails.
4282 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4284 * tree-vector-builder.h
4285 (tree_vector_builder::binary_encoded_nelts): Declare.
4286 * tree-vector-builder.c
4287 (tree_vector_builder::binary_encoded_nelts): New function.
4288 * fold-const.c (negate_expr_p): Likewise.
4289 (operand_equal_p, fold_checksum_tree): Likewise.
4290 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4291 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4292 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4293 (uniform_vector_p): Likewise.
4294 * varasm.c (const_hash_1, compare_constant): Likewise.
4295 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4296 (valid_lattice_transition): Operate directly on the VECTOR_CST
4298 * ipa-icf.c: Include tree-vector-builder.h.
4299 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4300 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4302 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4304 * tree.c (build_vector): Delete.
4305 * tree.h (build_vector): Make static and move into the self-testing
4308 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4310 * vector-builder.h (vector_builder::derived): New const overload.
4311 (vector_builder::elt): New function.
4312 * tree-vector-builder.h (tree_vector_builder::type): New function.
4313 (tree_vector_builder::apply_step): Declare.
4314 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4316 * gimple-fold.h (tree_vector_builder): Declare.
4317 (gimple_build_vector): Take a tree_vector_builder instead of a
4318 type and vector of elements.
4319 * gimple-fold.c (gimple_build_vector): Likewise.
4320 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4322 (get_initial_defs_for_reduction): Likewise.
4323 (vectorizable_induction): Likewise.
4325 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4327 * tree-vector-builder.h
4328 (tree_vector_builder::new_binary_operation): Declare.
4329 * tree-vector-builder.c
4330 (tree_vector_builder::new_binary_operation): New function.
4331 * fold-const.c (fold_relational_const): Use it.
4332 (const_binop): Likewise. Check that both input vectors have
4333 the same number of elements, thus excluding things like WIDEN_SUM.
4334 Check whether it is possible to operate directly on the encodings
4337 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4339 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4340 new_unary_operation, operating only on the encoded elements.
4341 (const_unop): Likewise.
4342 (exact_inverse): Likewise.
4343 (distributes_over_addition_p): New function.
4344 (const_binop): Use tree_vector_builder and new_unary_operation
4345 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4346 on the encoded elements unless the encoding is strided and the
4347 operation does not distribute over addition.
4348 (fold_convert_const): Use tree_vector_builder and
4349 new_unary_operation. Operate only on the encoded elements
4350 for truncating integer conversions, or for non-stepped encodings.
4352 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4354 * config/sparc/sparc.c: Include tree-vector-builder.h.
4355 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4356 * expmed.c: Include tree-vector-builder.h.
4357 (make_tree): Use tree_vector_builder instead of build_vector.
4358 * fold-const.c: Include tree-vector-builder.h.
4359 (const_binop): Use tree_vector_builder instead of build_vector.
4360 (const_unop): Likewise.
4361 (native_interpret_vector): Likewise.
4362 (fold_vec_perm): Likewise.
4363 (fold_ternary_loc): Likewise.
4364 * gimple-fold.c: Include tree-vector-builder.h.
4365 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4367 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4368 (simplify_vector_constructor): Use tree_vector_builder instead
4370 * tree-vect-generic.c: Include tree-vector-builder.h.
4371 (add_rshift): Use tree_vector_builder instead of build_vector.
4372 (expand_vector_divmod): Likewise.
4373 (optimize_vector_constructor): Likewise.
4374 * tree-vect-loop.c: Include tree-vector-builder.h.
4375 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4376 of build_vector. Explicitly use a stepped encoding for
4378 * tree-vect-slp.c: Include tree-vector-builder.h.
4379 (vect_get_constant_vectors): Use tree_vector_builder instead
4381 (vect_transform_slp_perm_load): Likewise.
4382 (vect_schedule_slp_instance): Likewise.
4383 * tree-vect-stmts.c: Include tree-vector-builder.h.
4384 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4385 (vect_gen_perm_mask_any): Likewise.
4386 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4387 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4389 (build_vector_from_val): Likewise. Explicitly use a duplicate
4392 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4394 * doc/generic.texi (VECTOR_CST): Describe new representation of
4396 * vector-builder.h: New file.
4397 * tree-vector-builder.h: Likewise.
4398 * tree-vector-builder.c: Likewise.
4399 * Makefile.in (OBJS): Add tree-vector-builder.o.
4400 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4401 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4402 (tree_vector): Change the number of elements to
4403 vector_cst_encoded_nelts.
4404 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4405 (VECTOR_CST_ELTS): Delete.
4406 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4407 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4408 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4409 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4410 (VECTOR_CST_ENCODED_ELT): Likewise.
4411 (vector_cst_encoded_nelts): New function.
4412 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4413 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4414 (vector_cst_int_elt, vector_cst_elt): Declare.
4415 * tree.c: Include tree-vector-builder.h.
4416 (tree_code_size): Abort if passed VECTOR_CST.
4417 (tree_size): Update for new VECTOR_CST layout.
4418 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4419 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4420 (build_vector): Use tree_vector_builder.
4421 (vector_cst_int_elt, vector_cst_elt): New functions.
4422 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4423 encoded elements and then create the vector in the canonical form.
4424 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4425 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4426 (tree_c_tests): Call test_vector_cst_patterns.
4427 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4429 (hash_tree): Likewise.
4430 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4431 (streamer_write_tree_header): Likewise.
4432 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4433 (streamer_alloc_tree): Likewise. Update call to make_vector.
4434 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4436 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4438 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4439 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4440 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4442 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4443 Richard Biener <rguenther@suse.de>
4445 PR tree-optimization/81303
4446 * Makefile.in (gimple-loop-interchange.o): New object file.
4447 * common.opt (floop-interchange): Reuse the option from graphite.
4448 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4449 -floop-interchange and mention it for -O3.
4450 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4451 * gimple-loop-interchange.cc: New file.
4452 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4453 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4454 * passes.def (pass_linterchange): New pass.
4455 * timevar.def (TV_LINTERCHANGE): New time var.
4456 * tree-pass.h (make_pass_linterchange): New declaration.
4457 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4458 interchange. Record IV before/after increment in new parameters.
4459 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4460 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4462 (check_reduction_path): ...New function here.
4463 * tree-vectorizer.h (check_reduction_path): New declaration.
4465 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4468 PR rtl-optimization/80818
4469 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4470 always early clobbered.
4471 * lra-lives.c (process_bb_lives): Check input hard regs for early
4472 clobbered non-operand hard reg.
4474 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4477 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4478 types_compatible_p, just that TYPE_MODE is the same.
4480 2017-12-07 Martin Sebor <msebor@redhat.com>
4483 * attribs.c (empty_attribute_table): Initialize new member of
4484 struct attribute_spec.
4485 (decl_attributes): Add argument. Handle mutually exclusive
4486 combinations of attributes.
4487 (selftests::test_attribute_exclusions): New function.
4488 (selftests::attribute_c_tests): Ditto.
4489 * attribs.h (decl_attributes): Add default argument.
4490 * selftest.h (attribute_c_tests): Declare.
4491 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4492 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4494 * doc/extend.texi (Common Function Attributes): Update const and pure.
4495 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4496 of struct attribute_spec.
4497 * config/arc/arc.c (arc_attribute_table): Same.
4498 * config/arm/arm.c (arm_attribute_table): Same.
4499 * config/avr/avr.c ( avr_attribute_table): Same.
4500 * config/bfin/bfin.c (bfin_attribute_table): Same.
4501 * config/cr16/cr16.c (cr16_attribute_table): Same.
4502 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4503 * config/h8300/h8300.c (h8300_attribute_table): Same.
4504 * config/i386/i386.c (ix86_attribute_table): Same.
4505 * config/ia64/ia64.c (ia64_attribute_table): Same.
4506 * config/m32c/m32c.c (m32c_attribute_table): Same.
4507 * config/m32r/m32r.c (m32r_attribute_table): Same.
4508 * config/m68k/m68k.c (m68k_attribute_table): Same.
4509 * config/mcore/mcore.c (mcore_attribute_table): Same.
4510 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4511 * config/mips/mips.c (mips_attribute_table): Same.
4512 * config/msp430/msp430.c (msp430_attribute_table): Same.
4513 * config/nds32/nds32.c (nds32_attribute_table): Same.
4514 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4515 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4516 * config/rl78/rl78.c (rl78__attribute_table): Same.
4517 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4518 * onfig/rx/rx.c (rx_attribute_table): Same.
4519 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4520 * config/sh/sh.c (sh_attribute_table): Same.
4521 * config/sparc/sparc.c (sparc_attribute_table): Same.
4522 * config/spu/spu.c (spu_attribute_table): Same.
4523 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4524 * config/v850/v850.c (v850_attribute_table): Same.
4525 * config/visium/visium.c (visium_attribute_table): Same.
4527 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4530 * config/arm/arm.c (INCLUDE_STRING): Define.
4531 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4532 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4535 2017-12-07 Michael Matz <matz@suse.de>
4537 Add unroll and jam pass
4539 * gimple-loop-jam.c: New file.
4540 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4541 * common.opt (funroll-and-jam): New option.
4542 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4543 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4544 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4545 * passes.def: Add pass_loop_jam.
4546 * timevar.def (TV_LOOP_JAM): Add.
4547 * tree-pass.h (make_pass_loop_jam): Declare.
4548 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4549 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4550 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4551 to flow_loop_tree_node_add.
4552 (duplicate_subloops, copy_loops_to): Append to sibling list.
4553 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4554 * doc/invoke.texi (-funroll-and-jam): Document new option.
4555 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4557 2017-12-07 Richard Biener <rguenther@suse.de>
4559 PR tree-optimization/83296
4560 PR tree-optimization/67769
4561 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4562 flow sensitive info in an unrelated BB.
4563 (value_replacement): Use reset_flow_sensitive_info.
4564 (minmax_replacement): Reset flow sensitive info on the def
4565 we move. Do not reset flow sensitive info in the whole BB
4566 we move the stmt to.
4567 (abs_replacement): Likewise.
4569 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4572 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4573 rs6000_cpu to the given -mcpu=, or to the default processor.
4575 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4577 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4578 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4579 instead of rs6000_cpu_attr.
4580 (rs6000_adjust_cost): Ditto.
4581 (is_microcoded_insn): Ditto.
4582 (rs6000_adjust_priority): Ditto.
4583 (rs6000_issue_rate): Ditto.
4584 (rs6000_use_sched_lookahead): Ditto.
4585 (rs6000_use_sched_lookahead_guard): Ditto.
4586 (rs6000_sched_reorder): Ditto.
4587 (force_new_group): Ditto.
4588 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4589 (group_ending_nop): Ditto.
4591 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4593 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4595 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4596 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4598 (rs6000_reassociation_width): Ditto.
4599 (rs6000_emit_epilogue): Ditto.
4600 (rs6000_adjust_cost): Ditto.
4601 (is_microcoded_insn): Ditto.
4602 (is_cracked_insn): Ditto.
4603 (rs6000_adjust_priority): Ditto.
4604 (rs6000_sched_reorder): Ditto.
4605 (rs6000_sched_reorder2): Ditto.
4606 (insn_must_be_first_in_group): Ditto.
4607 (insn_must_be_last_in_group): Ditto.
4608 (rs6000_register_move_cost): Ditto.
4609 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4612 2017-12-07 Julia Koval <julia.koval@intel.com>
4614 * config.gcc: Add vaesintrin.h.
4615 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4617 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4618 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4620 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4621 * config/i386/immintrin.h: Include vaesintrin.h.
4622 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4623 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4624 _mm_aesdec_epi128): New intrinsics.
4626 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4628 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4629 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4630 c-family/c-spellcheck.cc.
4631 (best_macro_match::best_macro_match): Likewise.
4633 (struct edit_distance_traits<cpp_hashnode *>): Move to
4634 c-family/c-spellcheck.h.
4635 (class best_macro_match): Likewise.
4637 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4639 PR tree-optimization/83293
4640 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4641 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4642 might insert into empty bb.
4645 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4646 simplify for plus with :c added, and pointer_plus without that.
4647 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4648 with undefined overflow and the conversion is not widening,
4649 perform negation in utype and only convert to type afterwards.
4650 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4651 simplify for plus with :c added, and pointer_plus without that.
4652 If type is integral with undefined overflow and the conversion is
4653 not widening, perform minus in utype and only convert to type
4654 afterwards. Move the last pointer_diff_expr simplify into the
4657 2017-12-06 Martin Sebor <msebor@redhat.com>
4659 PR tree-optimization/82646
4660 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4661 strncpy, not maxlen.
4663 2017-12-06 Martin Sebor <msebor@redhat.com>
4665 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4668 PR tree-optimization/83075
4669 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4670 strncat/strncpy don't change length of source string.
4672 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4675 2017-11-29 Martin Aberg <maberg@gaisler.com>
4677 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4678 to prevent b2bst errata sequence.
4679 (sqrtdf2_fix): Likewise.
4681 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4683 PR tree-optimization/81945
4684 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4685 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4686 to dest_cfun has orig_loop_num set, either remap it to the new
4687 loop number if the loop got moved too, or clear it.
4689 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4691 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4693 (thunderx2t99_nothing): New insn reservation.
4694 (thunderx2t99_mrs): New insn reservation.
4695 (thunderx2t99_multiple): New insn reservation.
4696 (thunderx2t99_alu_basi): Add bfx to reservation.
4697 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4699 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4702 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4704 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4706 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4707 tree-ssa-pre.c::remove_dead_inserted_code.
4708 * tree-ssa-dce.h: New file.
4709 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4710 (remove_dead_inserted_code): Move and rename to function
4711 tree-ssa-dce.c::simple_dce_from_worklist.
4712 (pass_pre::execute): Update use.
4714 2017-12-05 Richard Biener <rguenther@suse.de>
4716 PR tree-optimization/83277
4717 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4718 to code-gen liveout vars.
4720 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4722 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4723 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4724 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4727 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4729 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4731 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4732 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4735 2017-12-05 Richard Biener <rguenther@suse.de>
4737 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4738 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4739 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4740 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4742 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4744 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4746 2017-12-05 Martin Liska <mliska@suse.cz>
4747 Jakub Jelinek <jakub@redhat.com>
4749 * doc/invoke.texi: Document the options.
4750 * flag-types.h (enum sanitize_code): Add
4751 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4752 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4753 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4754 * opts.c: Define new sanitizer options.
4755 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4756 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4758 2017-12-05 Julia Koval <julia.koval@intel.com>
4760 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4761 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4762 (ix86_handle_option): Handle -mavx512vnni.
4763 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4764 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4765 * config/i386/i386-c (__AVX512VNNI__): New.
4766 * config/i386/i386.c (ix86_target_string): Handle new option.
4767 (ix86_valid_target_attribute_inner_p): Handle new option.
4768 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4769 * config/i386/i386.opt (mavx512vnni): New option.
4771 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4774 * athlon.md: Disable for generic.
4775 * haswell.md: Enable for generic.
4776 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4777 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4779 (ix86_adjust_cost): Move generic to haswell path.
4781 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4783 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4784 instead of MEM_P in a couple more places. Fix formatting issues.
4786 2017-12-04 Jim Wilson <jimw@sifive.com>
4788 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4789 instead of GP_REG_LAST-1.
4790 (riscv_adjust_libcall_cfi_prologue): Likewise.
4791 (riscv_adjust_libcall_cri_epilogue): Likewise.
4792 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4795 2017-12-04 Luis Machado <luis.machado@linaro.org>
4797 * ipa-pure-const.c (check_decl): Add missing newline.
4798 (state_from_flags): Likewise.
4800 2017-12-04 Jeff Law <law@redhat.com>
4802 PR tree-optimizatin/78496
4803 * gimple-ssa-evrp-analyze.h
4804 (evrp_range_analyzer::get_vr_values): Simplify.
4805 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4806 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4807 and gimple-ssa-evrp-analyze.h.
4808 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4809 (simplify_stmt_for_jump_threading): Copy a blob of code from
4810 tree-vrp.c to use ranges to simplify statements.
4811 (dom_opt_dom_walker::before_dom_children): Call
4812 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4813 (dom_opt_dom_walker::after_dom_children): Similarly for
4814 evrp_range_analyzer::leave.
4815 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4818 * gimple-ssa-evrp-analyze.c
4819 (evrp_range_analyzer::extract_range_from_stmt): Always use
4820 vr_values::update_value_range so preexisting range info is
4821 medged with new range info, even if the new range is VR_VARYING.
4823 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4825 * combine.c: Adjust comment.
4826 (use_crosses_set_p): Delete.
4827 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4828 (try_combine): Ditto.
4830 2017-12-04 Richard Biener <rguenther@suse.de>
4832 PR tree-optimization/83255
4833 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4834 Re-add zero-iteration check.
4836 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4838 PR rtl-optimization/83245
4839 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4840 hard registers as earlyclobber, also if not in an asm.
4842 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4846 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4850 rs6000_cpu based on cpu_index, not tune_index.
4852 2017-12-04 Richard Biener <rguenther@suse.de>
4854 PR tree-optimization/83238
4855 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4856 code match comment, rejecting invalid SESE regions.
4858 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4860 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4861 require base operand is a REG_POINTER prior to reload on targets
4862 with non-equivalent space registers.
4864 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4866 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4867 (update_specialized_profile): Fix updating of counts.
4868 (perhaps_add_new_callers): Likewise.
4870 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4873 * x86-tune.def: Remove obsolette FIXMEs.
4874 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4875 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4876 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4878 (X86_TUNE_PAD_RETURNS): Disable for generic.
4880 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4882 PR tree-optimization/83170
4883 PR tree-optimization/83241
4884 * gimple-ssa-store-merging.c
4885 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4886 gimple_vuse (ins_stmt) in case it has changed.
4887 (imm_store_chain_info::output_merged_store): Likewise.
4889 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4893 * tree-cfg.c (pass_warn_function_return::execute): Handle
4894 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4895 with BUILTINS_LOCATION.
4899 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4900 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4901 instead of DECL_MODE.
4903 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4904 last argument to rtx pointer.
4905 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4906 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4907 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4908 depending on the chosen ISAs.
4909 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4910 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4911 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4913 * config/i386/sse.md (mov<mode>_internal): Likewise.
4914 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4916 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4918 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4920 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4922 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4923 parameters from prototype.
4924 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4925 parameters. Don't print a comment.
4926 (emit_fusion_gpr_load): Adjust.
4927 (emit_fusion_load_store): Adjust.
4928 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4929 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4930 comment on the second line.
4932 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4935 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4936 rs6000_cpu based on cpu_index, not tune_index.
4938 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4940 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4941 which_alternative instead of which_alternative + 1.
4942 (output_asm_insn): Print an extra tab if the template is short.
4944 2017-12-01 Jim Wilson <jimw@sifive.com>
4946 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4948 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4949 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4951 * doc/tm.texi: Regenerate.
4953 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4956 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4957 whether we can allocate pseudos before trying to fix an address.
4958 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4959 memory address is indexed or indirect.
4960 (floatuns_<mode>si2_hw2): Likewise.
4962 2017-12-01 Jason Merrill <jason@redhat.com>
4964 * Makefile.in (TAGS): Add c-family/*.cc.
4966 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4968 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4969 (call_value_insn): Likewise.
4970 (sibcall_insn): Likewise.
4971 (sibcall_value_insn): Likewise.
4972 (movsi_aarch64): Likewise.
4973 (movdi_aarch64): Likewise.
4974 (add_losym_): Likewise.
4975 (ldr_got_small_): Likewise.
4976 (ldr_got_small_sidi): Likewise.
4977 (ldr_got_small_28k_): Likewise.
4978 (ldr_got_small_28k_sidi): Likewise.
4979 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4980 Move output_addr_const to symbolic case. Add error check.
4982 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4984 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4985 * config/s390/s390.md: Disable MVC merging peephole if it would
4986 disable operand forwarding.
4987 (new peephole2): Split MVCs if it would turn them into up to 2
4990 2017-12-01 Richard Biener <rguenther@suse.de>
4992 PR tree-optimization/83232
4993 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4994 detection of same access. Instead of breaking the group here
4995 do not consider the duplicate. Add comment explaining real fix.
4997 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4999 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5001 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5003 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5005 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5007 * function.h (struct function): Remove cilk_frame_decl,
5008 is_cilk_function and calls_cilk_spawn fields.
5009 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5011 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5012 cilk_elemental field.
5013 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5014 * target.def: Adjust comment.
5015 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5016 Don't test cilk_elemental.
5018 PR tree-optimization/83233
5019 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5020 bswap_stat name for the struct.
5023 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5024 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5025 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5026 SWITCH_BREAK_LABEL_P set on the label.
5027 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5028 added for default case if it was missing and not all cases covered.
5029 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5030 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5032 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5033 to avoid -Wimplicit-fallthrough warning.
5034 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5035 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5037 PR tree-optimization/83221
5038 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5040 (init_reassoc): Formatting fix.
5043 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5044 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5048 * tree-cfg.c: Include asan.h.
5049 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5051 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5053 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5055 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5057 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5058 vpcmpeqd instruction.
5060 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5062 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5063 registers type for 128bit mode.
5065 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5067 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5069 * gengtype-state.c (read_state_pair): Likewise.
5070 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5071 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5072 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5073 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5074 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5075 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5076 * builtins.c (fold_builtin_3): Likewise.
5077 * graphite-scop-detection.c
5078 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5079 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5082 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5085 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5086 parameter. Return true if ignore_decl_rtl_set_p is true and other
5087 conditions are satisfied.
5088 * asan.h (asan_protect_global): Add new parameter.
5089 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5090 to asan_protect_global calls.
5092 2017-11-30 Jim Wilson <jimw@sifive.com>
5094 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5095 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5096 -mstrict-align, add info on default value. Delete redundant lines for
5097 -mabi. Add missing -mexplicit-relocs docs.
5099 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5101 * config/arc/arc.md (trap): New pattern.
5103 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5105 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5106 instruction to end into a delay slot.
5107 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5108 can be placed into a delay slot against reg_note.
5110 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5112 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5113 labels number of usages.
5115 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5117 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5119 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5121 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5124 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5125 multiplication by power of 2 constant into two shifts + comparison.
5127 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5130 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5132 2017-11-30 Richard Biener <rguenther@suse.de>
5134 PR tree-optimization/83202
5135 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5136 allow_peel argument and guard peeling.
5137 (canonicalize_loop_induction_variables): Likewise.
5138 (canonicalize_induction_variables): Pass false.
5139 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5140 peeling from cunrolli.
5142 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5144 * combine.c (try_combine): Print a message to dump file whenever
5145 I0, I1, or I2 cannot be combined into I3.
5147 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5149 PR rtl-optimization/83156
5150 PR rtl-optimization/82621
5151 * combine.c (try_combine): Don't split an I2 if one of the dests is
5152 set again before I3. Allow unused dests.
5154 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5156 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5158 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5160 PR rtl-optimization/80818
5161 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5162 recursively. Use insn code for clobber.
5163 (lra_set_insn_recog_data): Pass the new arg to
5164 collect_non_operand_hard_regs.
5165 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5167 (lra_update_insn_regno_info): Pass insn to
5168 add_regs_to_insn_regno_info.
5170 2017-11-29 Jim Wilson <jimw@sifive.com>
5171 Andrew Waterman <andrew@sifive.com>
5173 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5174 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5175 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5176 use SINGLE_SHIFT_COST.
5177 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5178 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5180 2017-11-29 Julia Koval <julia.koval@intel.com>
5182 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5183 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5184 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5185 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5186 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5187 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5188 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5189 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5190 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5191 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5192 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5193 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5194 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5195 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5196 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5197 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5198 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5199 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5200 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5201 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5202 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5203 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5204 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5205 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5206 __builtin_ia32_vpshldv_v2di_mask,
5207 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5208 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5209 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5211 2017-11-29 Julia Koval <julia.koval@intel.com>
5213 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5214 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5215 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5216 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5217 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5218 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5219 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5220 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5221 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5222 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5223 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5224 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5225 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5226 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5227 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5228 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5229 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5230 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5231 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5232 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5233 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5235 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5237 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5238 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5241 2017-11-29 Martin Aberg <maberg@gaisler.com>
5243 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5244 to prevent b2bst errata sequence.
5245 (sqrtdf2_fix): Likewise.
5247 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5249 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5250 (div_sqrt_insn_p): New function.
5251 (sparc_do_work_around_errata): Insert NOP instructions to
5252 prevent sequences that could trigger the TN-0013 errata for
5253 certain LEON3 processors.
5254 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5255 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5256 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5257 (in_branch_delay): Prevent div and sqrt in delay slot if
5259 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5261 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5263 * config/sparc/sparc.c (atomic_insn_p): New function.
5264 (sparc_do_work_around_errata): Insert NOP instructions to
5265 prevent sequences that could trigger the TN-0010 errata for
5267 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5268 instruction referable in atomic_insns_p.
5270 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5272 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5273 (atomic_compare_and_swap_leon3_1): Likewise.
5276 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5278 * config/sparc/sparc.c (fpop_insn_p): New function.
5279 (sparc_do_work_around_errata): Insert NOP instructions to
5280 prevent sequences that could trigger the TN-0012 errata for
5282 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5283 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5284 (in_branch_annul_delay): Prevent floating-point instructions
5285 in delay slot of annulled integer branch.
5287 2017-11-29 Richard Biener <rguenther@suse.de>
5289 PR tree-optimization/83202
5290 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5292 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5293 nodes vectorized to the same stmts multiple times.
5294 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5295 (vect_analyze_slp_instance): Adjust.
5296 (scalar_stmts_to_slp_tree_map_t): New typedef.
5297 (vect_schedule_slp_instance): Add a map recording the SLP node
5298 representing the vectorized stmts for a set of scalar stmts.
5299 Avoid code-generating redundancies.
5300 (vect_schedule_slp): Allocate map and pass it down.
5302 2017-11-29 Nathan Sidwell <nathan@acm.org>
5305 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5308 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5310 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5311 cases for printing LDP/STP memory addresses.
5312 (aarch64_print_address_internal): Renamed from
5313 aarch64_print_operand_address, added parameter, add Pmode check.
5314 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5315 (aarch64_print_operand_address): Indirect to
5316 aarch64_print_address_internal.
5317 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5318 'y' operand output specifier.
5320 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5323 * tree.c (build_simple_mem_ref_loc): Handle
5324 get_addr_base_and_unit_offset returning a MEM_REF.
5327 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5330 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5333 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5335 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5337 (layout::move_to_column): Likewise.
5339 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5342 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5343 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5345 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5346 Martin Jambor <mjambor@suse.cz>
5349 * tree.h (expr_type_first_operand_type_p): Declare
5350 * tree.c (expr_type_first_operand_type_p): New function.
5351 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5352 (ipa_value_from_jfunc): Adjust declaration.
5353 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5354 Use it as result type for arithmetics, unless it is NULL in which case
5355 be more conservative.
5356 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5357 ipa_get_jf_pass_through_result.
5358 (propagate_vals_across_pass_through): Likewise.
5359 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5360 is to propagate_vals_across_pass_through.
5361 (propagate_constants_across_call): Pass PARM_TYPE to
5362 propagate_scalar_across_jump_function.
5363 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5364 ipa_value_from_jfunc.
5365 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5366 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5367 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5368 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5369 target_type, pass it to ipa_value_from_jfunc.
5370 (update_indirect_edges_after_inlining): Pass parameter type to
5371 try_make_edge_direct_simple_call.
5373 2017-11-28 Jeff Law <law@redhat.com>
5375 * gimple-ssa-evrp-analyze.c
5376 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5377 refine ranges if scev_initialized_p returns true.
5378 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5380 2017-11-28 Julia Koval <julia.koval@intel.com>
5382 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5383 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5384 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5385 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5386 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5387 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5388 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5389 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5390 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5391 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5392 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5393 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5394 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5395 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5396 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5397 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5398 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5399 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5400 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5401 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5402 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5403 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5405 2017-11-28 Julia Koval <julia.koval@intel.com>
5407 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5408 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5409 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5410 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5411 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5412 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5413 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5414 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5415 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5416 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5417 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5418 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5419 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5420 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5421 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5422 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5423 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5424 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5425 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5426 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5427 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5428 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5429 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5430 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5431 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5432 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5433 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5434 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5435 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5436 __builtin_ia32_vpshld_v2di_mask): New builtins.
5437 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5438 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5440 2017-11-28 Richard Biener <rguenther@suse.de>
5442 PR tree-optimization/80776
5443 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5445 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5447 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5448 If the incoming edge is an effective fallthru because the other
5449 edge only reaches a __builtin_unreachable () then record ranges
5450 derived from the controlling condition in SSA info.
5451 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5452 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5454 2017-11-28 Olivier Hainque <hainque@adacore.com>
5456 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5457 on mingw build hosts.
5459 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5462 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5464 2017-11-28 Richard Biener <rguenther@suse.de>
5467 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5468 copies generated from memcpy use a character array as reference
5471 2017-11-28 Julia Koval <julia.koval@intel.com>
5472 Sebastian Peryt <sebastian.peryt@intel.com>
5474 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5475 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5476 cilk-common.o, cilk.h, cilk-common.c): Remove.
5478 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5479 * builtins.c (is_builtin_name): Remove cilkplus condition.
5480 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5481 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5482 cilk-builtins.def, cilkplus.def): Remove.
5483 * cif-code.def (CILK_SPAWN): Remove.
5484 * cilk-builtins.def: Delete.
5485 * cilk-common.c: Ditto.
5487 * cilkplus.def: Ditto.
5488 * config/darwin.h (fcilkplus): Delete.
5489 * cppbuiltin.c: Ditto.
5490 * doc/extend.texi: Remove cilkplus doc.
5491 * doc/generic.texi: Ditto.
5492 * doc/invoke.texi: Ditto.
5493 * doc/passes.texi: Ditto.
5494 * gcc.c (fcilkplus): Remove.
5495 * gengtype.c (cilk.h): Remove.
5496 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5498 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5500 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5502 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5503 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5504 cilkplus conditions.
5505 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5506 inline_read_section): Ditto.
5507 * ipa-inline-analysis.c (cilk.h): Remove.
5508 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5509 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5510 append_linker_options): Remove condition for fcilkplus.
5511 * lto/lto-lang.c (cilk.h): Remove.
5512 (lto_init): Remove condition for fcilkplus.
5513 * omp-expand.c (expand_cilk_for_call): Delete.
5514 (expand_omp_taskreg, expand_omp_for_static_chunk,
5515 expand_omp_for): Remove cilkplus
5517 (expand_cilk_for): Delete.
5518 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5519 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5520 execute_lower_omp, diagnose_sb_0): Ditto.
5521 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5522 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5523 * tree-nested.c: Ditto.
5524 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5525 (dump_generic_node): Ditto.
5526 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5527 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5529 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5531 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5533 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5534 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5535 for vector mode and !TARGET_SIMD.
5537 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5539 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5541 * tree.h (SWITCH_LABELS): Remove.
5542 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5543 assert SWITCH_BODY is non-NULL.
5544 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5546 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5548 PR tree-optimization/80788
5549 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5550 has TREE_OVERFLOW set, call drop_tree_overflow.
5552 2017-11-28 Richard Biener <rguenther@suse.de>
5554 PR tree-optimization/83158
5555 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5557 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5560 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5561 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5563 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5565 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5566 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5568 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5571 * varasm.c (compare_constant): Take the mode of the constants into
5572 account when comparing floating point constants.
5574 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5576 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5577 from explicit instantiation of debug_helper.
5578 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5580 2017-11-27 Richard Biener <rguenther@suse.de>
5582 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5585 2017-11-27 Richard Biener <rguenther@suse.de>
5587 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5588 (type_hash_canon): Also clear int_cst_hash_table entry for
5590 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5592 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5594 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5595 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5596 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5598 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5600 * hash-map.h (gt_cleare_cache): Avoid UB.
5602 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5604 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5605 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5606 (decide_unroll_runtime_iterations): Likewise.
5607 (decide_unroll_stupid): Likewise.
5609 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5612 * config/i386/i386.md: Add a loop with incssp.
5614 2017-11-27 Martin Jambor <mjambor@suse.cz>
5616 PR tree-optimization/81248
5617 * tree-sra.c (splice_param_accesses): Remove size check.
5618 (decide_one_param_reduction): Fix size check.
5619 * gimple-pretty-print.c (dump_profile): Silence warning.
5620 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5622 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5625 * dbxout.c (lastlineno): New variable.
5626 (dbx_debug_hooks): Use dbxout_switch_text_section as
5627 switch_text_section debug hook.
5628 (dbxout_function_end): Switch to current_function_section
5629 rather than function_section. If crtl->has_bb_partition,
5630 output just one N_FUN, depending on in_cold_section_p.
5631 (dbxout_source_line): Remember last lineno in lastlineno.
5632 (dbxout_switch_text_section): New function.
5633 (dbxout_function_decl): Adjust dbxout_block caller.
5634 (dbx_block_with_cold_children): New function.
5635 (dbxout_block): Return true if any LBRAC/RBRAC have been
5636 emitted. Use dbx_block_with_cold_children at depth == 0
5637 in second partition. Add PARENT_BLOCKNUM argument, pass
5638 it optionally adjusted to children. Output LBRAC/RBRAC
5639 around recursive call only if the block is in the current
5640 partition, if not and anything was output, emit empty
5642 * final.c (final_scan_insn): Compute cold_function_name
5643 before calling switch_text_section debug hook. Call
5644 that hook even if dwarf2out_do_frame if not emitting
5648 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5649 TREE_READONLY decls.
5651 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5653 PR rtl-optimization/82488
5654 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5656 2017-11-26 Julia Koval <julia.koval@intel.com>
5658 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5660 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5663 2017-11-26 Julia Koval <julia.koval@intel.com>
5665 * config/i386/driver-i386.c (host_detect_local_cpu):
5666 Detect skylake-avx512.
5668 2017-11-26 Julia Koval <julia.koval@intel.com>
5670 * config.gcc: Add -march=cannonlake.
5671 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5672 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5673 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5674 (PTA_CANNONLAKE): New.
5675 (processor_target_table): Add cannonlake.
5676 (ix86_option_override_internal): Ditto.
5677 (fold_builtin_cpu): Ditto.
5678 (get_builtin_code_for_version): Handle cannonlake.
5679 (M_INTEL_COREI7_CANNONLAKE): New.
5680 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5681 * doc/invoke.texi: Add -march=cannonlake.
5683 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5685 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5686 (try_init_one_plugin): Alternative implementation for MinGW.
5687 * Makefile.in (plugin_implib): New.
5688 (gengtype-lex.c): Fix broken AIX workaround.
5689 * configure: Regenerate.
5690 * doc/plugins.texi: Document support for MinGW.
5692 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5694 PR rtl-optimization/81553
5695 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5696 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5697 is a shift where C1 has different mode than the whole shift, use C1's
5698 mode for MULT rather than the shift's mode.
5701 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5702 builtins not enabled in the currently selected ISA.
5704 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5706 PR tree-optimization/71026
5707 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5708 (insert_reciprocals): Change to insert reciprocals before a division
5709 by a square and to insert the square of a reciprocal.
5710 (execute_cse_reciprocals_1): Change to consider division by a square.
5711 (register_division_in): Add importance parameter.
5713 2017-11-24 Richard Biener <rguenther@suse.de>
5715 PR tree-optimization/82402
5716 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5717 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5719 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5721 * match.pd (0-ptr): New transformation.
5723 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5726 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5727 when skipping checking.
5729 2017-11-24 Richard Biener <rguenther@suse.de>
5731 PR tree-optimization/83128
5732 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5733 (vn_reference_lookup_3): Likewise.
5735 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5738 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5739 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5742 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5743 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5744 check for the latter with gimple_call_builtin_p. Do not handle
5745 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5747 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5749 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5752 2017-11-23 Julia Koval <julia.koval@intel.com>
5754 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5755 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5756 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5757 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5758 _mm512_maskz_expandloadu_epi16): New intrinsics.
5759 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5760 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5761 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5762 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5763 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5764 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5765 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5766 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5767 _mm256_maskz_expandloadu_epi8): New intrinsics.
5768 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5769 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5770 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5771 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5772 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5773 * config/i386/sse.md (VI248_VLBW): New iterator.
5774 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5776 2017-11-23 Julia Koval <julia.koval@intel.com>
5778 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5779 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5780 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5781 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5782 _mm512_mask_compressstoreu_epi16): New.
5783 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5784 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5785 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5786 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5787 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5788 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5789 _mm256_mask_compressstoreu_epi8): New.
5790 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5791 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5792 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5793 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5794 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5795 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5796 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5797 __builtin_ia32_compresshi128_mask,
5798 __builtin_ia32_compressstoreuqi512_mask,
5799 __builtin_ia32_compressstoreuhi512_mask,
5800 __builtin_ia32_compressstoreuqi256_mask,
5801 __builtin_ia32_compressstoreuqi128_mask,
5802 __builtin_ia32_compressstoreuhi256_mask,
5803 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5804 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5806 (ix86_expand_special_args_builtin): Handle new types.
5807 (s4fma_expand): Handle new builtin array.
5808 * config/i386/immintrin.h: Include new headers.
5809 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5810 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5812 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5814 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5815 fixes. Declare temp and g variables at the top in order to avoid
5816 {} in most of the cases.
5818 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5820 * match.pd (ptr-0): New transformation.
5822 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5824 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5825 (struct addr_mode_cost_table): New.
5826 (struct tune_params): Add field addr_mode_costs.
5827 * config/arm/arm.c (generic_addr_mode_costs): New.
5828 (arm_slowmul_tune): Initialise addr_mode_costs field.
5829 (arm_fastmul_tune): Likewise.
5830 (arm_strongarm_tune): Likewise.
5831 (arm_xscale_tune): Likewise.
5832 (arm_9e_tune): Likewise.
5833 (arm_marvell_pj4_tune): Likewise.
5834 (arm_v6t2_tune): Likewise.
5835 (arm_cortex_tune): Likewise.
5836 (arm_cortex_a8_tune): Likewise.
5837 (arm_cortex_a7_tune): Likewise.
5838 (arm_cortex_a15_tune): Likewise.
5839 (arm_cortex_a35_tune): Likewise.
5840 (arm_cortex_a53_tune): Likewise.
5841 (arm_cortex_a57_tune): Likewise.
5842 (arm_exynosm1_tune): Likewise.
5843 (arm_xgene1_tune): Likewise.
5844 (arm_cortex_a5_tune): Likewise.
5845 (arm_cortex_a9_tune): Likewise.
5846 (arm_cortex_a12_tune): Likewise.
5847 (arm_cortex_a73_tune): Likewise.
5848 (arm_v7m_tune): Likewise.
5849 (arm_cortex_m7_tune): Likewise.
5850 (arm_v6m_tune): Likewise.
5851 (arm_fa726te_tune): Likewise.
5852 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5854 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5856 * config/arm/arm.c (arm_mem_costs): New function.
5857 (arm_rtx_costs_internal): Use arm_mem_costs.
5859 2017-11-23 Mark Wielaard <mark@klomp.org>
5861 * dwarf2out.c (init_sections_and_labels): Use generation to create
5862 unique ranges_section_label and ranges_base_label. Return generation.
5863 (output_rnglists): Add generation argument. Use generation to create
5864 unique ranges labels.
5865 (dwarf2out_finish): Get generation from init_sections_and_labels
5866 and pass generation to output_rnglists.
5868 2017-11-23 Mike Stump <mikestump@comcast.net>
5869 Eric Botcazou <ebotcazou@adacore.com>
5871 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5872 * cfgloop.h (struct loop): Add unroll field.
5873 * function.h (struct function): Add has_unroll bitfield.
5874 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5875 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5876 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5877 (pass_rtl_unroll_loops::gate): Likewise.
5878 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5879 for which loop->unroll==1.
5880 (decide_unroll_constant_iterations): Use note for consistency and
5881 take loop->unroll into account. Return early if loop->unroll is set.
5882 Fix thinko in existing test.
5883 (decide_unroll_runtime_iterations): Use note for consistency and
5884 take loop->unroll into account.
5885 (decide_unroll_stupid): Likewise.
5886 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5887 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5888 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5890 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5891 (print_loop): Print loop->unroll if set.
5892 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5893 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5894 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5896 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5897 loop->unroll is set and smaller than the trip count. Otherwise bypass
5898 entirely the heuristics if loop->unroll is set. Remove dead note.
5899 Fix off-by-one bug in other note.
5900 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5901 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5903 (tree_unroll_loops_completely): Make static.
5904 (pass_complete_unroll::execute): Use correct type for variable.
5905 (pass_complete_unrolli::execute): Fix formatting.
5906 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5908 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5910 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5911 enable when TARGET_PREFER_AVX128 is set.
5913 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5915 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5917 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5918 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5920 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5922 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5923 frequencies. Fix estimation of aggregate parameters.
5925 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5927 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5930 2017-11-23 Tom de Vries <tom@codesourcery.com>
5932 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5934 2017-11-23 Tom de Vries <tom@codesourcery.com>
5936 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5939 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5942 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5943 sibcall_value_pcrel_fdpic): Use local variable instead of
5945 (calli_tbr_rel): Add missing operand 2.
5946 (call_valuei_tbr_rel): Add missing operand 3.
5948 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5951 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5952 bitpos/bitsize covering the whole destination, use store_expr only if
5953 the complex mode is the same. Otherwise, use expand_normal and if
5954 it returns CONCAT, subreg each part separately instead of trying to
5955 subreg the whole result.
5957 2017-11-23 Richard Biener <rguenther@suse.de>
5959 PR tree-optimization/23094
5960 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5961 come here from walking over backedges in the first iteration.
5962 (vn_reference_lookup_3): Skip clobbers that store the same value.
5964 2017-11-23 Richard Biener <rguenther@suse.de>
5966 PR tree-optimization/81403
5967 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5968 a block we need a leader relative to.
5969 (phi_translate_1): For nary processing require a leader from
5970 get_representative_for given we run expression simplification
5971 using match-and-simplify. Remove previous fix.
5973 2017-11-22 Jeff Law <law@redhat.com>
5975 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5976 Use new method allocate_value_range rather than accessing the
5977 vrp_value_range_pool data member directly.
5978 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5979 to use extract_range_from_stmt method to avoid need for
5980 extract_range_from_assignment method.
5981 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5982 method rather than setting values_propgated data member directly.
5983 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5984 and values propagated data members and extract_range_from_assignment
5985 method. Reorder private data members to conform to standards.
5986 Add new methods set_lattice_propagation_complete and
5987 allocate_value_range.
5989 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5991 PR rtl-optimization/83030
5992 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5993 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5994 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5995 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5996 CROSSING_JUMP_P flag is already set before setting it.
5997 * cfgrtl.c (fixup_partition_crossing): Likewise.
5998 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6001 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6003 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6004 Handle the case where both arguments are using gen_const_vec_series.
6006 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6009 * pretty-print.c (pp_format): Move quoting implementation to
6010 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6011 to pass address of "quote" local.
6012 (pp_begin_quote): New function.
6013 (pp_end_quote): New function.
6014 * pretty-print.h (printer_fn): Convert penultimate param from bool
6016 (pp_begin_quote): New decl.
6017 (pp_end_quote): New decl.
6018 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6019 param from bool to bool *.
6020 * tree-diagnostic.h (default_tree_printer): Likewise.
6022 2017-11-22 Jeff Law <law@redhat.com>
6024 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6025 Initialize vr_values.
6026 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6027 vr_values via vr_values class instance rather than delegators.
6028 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6029 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6030 (evrp_range_analyzer::push_value_range): Likewise.
6031 (evrp_range_analyzer::pop_value_range): Likewise.
6032 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6033 most delegators. Those remaining are exposed as public interfaces.
6034 Make vr_values a pointer and private.
6035 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6037 (evrp_range_analyzer::get_vr_value): New method.
6038 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6039 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6040 (class evrp_dom_walker): Attach evrp_folder class, initialize
6041 it in the ctor. Remove temporary delegators.
6042 (evrp_dom_walker::before_dom_children): Call methods in attached
6043 evrp_range_analyzer class via class instance pointer. Use
6044 free value_range_constant_singleton to remove need for
6045 op_with_constant_singleton_value delegator method. Do not
6046 create a vrp_prop class instance for every call! Narrow
6047 scope of a couple variables.
6048 (evrp_dom_walker::cleanup): Call methods in attached
6049 evrp_range_analyzer class via class instance pointer.
6050 * vr-values.h (class vr_values): Privatize many methods and
6053 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6055 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6056 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6057 * expr.c (expand_expr_real_2): Likewise.
6058 * fold-const.c (const_unop): Likewise.
6059 * optabs-tree.c (optab_for_tree_code): Likewise.
6060 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6061 * tree-inline.c (estimate_operator_cost): Likewise.
6062 * tree-pretty-print.c (dump_generic_node): Likewise.
6063 (op_code_prio): Likewise.
6064 (op_symbol_code): Likewise.
6065 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6066 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6067 * internal-fn.c (direct_internal_fn_optab): New function.
6068 (direct_internal_fn_array, direct_internal_fn_supported_p
6069 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6070 * fold-const-call.c (fold_const_reduction): New function.
6071 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6073 * tree-vect-loop.c: Include internal-fn.h.
6074 (reduction_code_for_scalar_code): Rename to...
6075 (reduction_fn_for_scalar_code): ...this and return an internal
6077 (vect_model_reduction_cost): Take an internal_fn rather than
6079 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6081 (vectorizable_reduction): Use internal functions rather than tree
6082 codes for the reduction operation. Update calls to the functions
6084 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6085 Use calls to internal functions rather than REDUC tree codes.
6086 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6088 2017-11-22 Olivier Hainque <hainque@adacore.com>
6090 * config/vxworks.c (vxworks_override_options): Pick default
6091 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6092 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6093 DWARF_GNAT_ENCODINGS_DEFAULT.
6094 * config/vxworksae.h: Likewise.
6096 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6098 PR tree-optimization/83104
6099 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6102 2017-11-22 Marek Polacek <polacek@redhat.com>
6103 H.J. Lu <hongjiu.lu@intel.com>
6104 Jason Merrill <jason@redhat.com>
6109 * calls.c (initialize_argument_information): Call
6110 warn_parameter_passing_abi target hook.
6111 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6112 argument onto stack.
6113 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6114 * common.opt: Update -fabi-version description.
6115 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6116 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6118 (ix86_is_empty_record): New function.
6119 (ix86_warn_parameter_passing_abi): New function.
6120 (TARGET_EMPTY_RECORD_P): Redefine.
6121 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6122 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6123 * doc/tm.texi: Regenerated.
6124 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6125 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6126 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6127 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6128 instead of int_size_in_bytes.
6129 * expr.c (copy_blkmode_to_reg): Likewise.
6130 * function.c (aggregate_value_p): Return 0 for empty types.
6131 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6133 (locate_and_pad_parm): Call arg size_in_bytes instead
6135 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6136 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6137 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6139 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6140 (std_gimplify_va_arg_expr): Skip empty records. Call
6141 arg_size_in_bytes instead size_in_bytes.
6142 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6143 * tree-core.h (tree_type_common): Add empty_flag.
6144 (tree_decl_common): Update comments.
6145 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6147 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6148 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6150 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6151 * tree.c (default_is_empty_type): New function.
6152 (default_is_empty_record): New function.
6153 (arg_int_size_in_bytes): New function.
6154 (arg_size_in_bytes): New function.
6155 (get_ultimate_context): New function.
6156 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6157 TRANSLATION_UNIT_WARN_EMPTY_P.
6158 (default_is_empty_record, arg_int_size_in_bytes,
6159 arg_size_in_bytes, get_ultimate_context): Declare.
6161 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6163 * config/arm/arm.c (cmse_clear_registers): New function.
6164 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6165 code by call to cmse_clear_registers.
6166 (cmse_nonsecure_entry_clear_before_return): Likewise.
6168 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6170 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6171 (vdot_s32, vdotq_s32): New.
6172 (vdot_lane_u32, vdotq_lane_u32): New.
6173 (vdot_lane_s32, vdotq_lane_s32): New.
6176 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6179 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6180 for unsigned values with fewer HWIs than the precision.
6181 (test_overflow): New function.
6182 (wide_int_cc_tests): Call it.
6184 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6186 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6187 has the same mode class as Pmode.
6189 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6191 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6192 padding_bits_to_clear_ptr.
6193 (cmse_nonsecure_entry_clear_before_return): Likewise.
6195 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6197 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6198 auto_sbitap instead of integer bitfield to control register needing
6201 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6203 PR tree-optimization/83044
6204 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6205 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6206 that need the upper bound. Subtract offset from
6207 get_addr_base_and_unit_offset only if positive and subtract it
6208 before division by eltsize rather than after it.
6211 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6212 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6213 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6217 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6218 expand_binop with *mul_widen_optab, make sure at least one of the
6219 operands doesn't have VOIDmode.
6222 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6224 PR rtl-optimization/82044
6225 PR tree-optimization/82042
6226 * dse.c (record_store): Check for overflow.
6227 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6228 add_wild_read instead of clear_rhs_from_active_local_stores on
6229 overflow and log it into dump_file.
6231 2017-11-22 Richard Biener <rguenther@suse.de>
6233 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6234 fake edges to exit when looking for a place to insert.
6235 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6237 (insert_into_preds_of_block): Commit edge insertion immediately,
6238 assert that doesn't require new BBs.
6239 (fini_pre): Release expressions.
6240 (pass_pre::execute): Shuffle things around a bit, if the fn
6241 is too large do not compute AVAIL either as this is really the
6244 2017-11-22 Richard Biener <rguenther@suse.de>
6246 PR tree-optimization/83089
6247 * tree-if-conv.c (pass_if_conversion::execute): If anything
6248 changed reset SCEV and free the number of iteration estimates.
6250 2017-11-21 Martin Sebor <msebor@redhat.com>
6252 PR tree-optimization/82945
6253 * calls.h (warn_nonstring_bound): Remove unused function.
6255 2017-11-21 Martin Sebor <msebor@redhat.com>
6257 PR tree-optimization/82945
6258 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6259 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6260 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6262 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6263 * calls.h (get_attr_nonstring_decl): Declare new function.
6264 * doc/extend.texi (attribute nonstring): Update.
6265 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6266 get_attr_nonstring_decl and handle it.
6267 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6268 detection of nul-termination.
6269 (strlen_to_stridx): Change to a pointer.
6270 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6271 (pass_strlen::execute): Same.
6273 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6275 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6276 for the new option -mprefer-vector-width=[none|128|256|512].
6277 * config/i386/i386.c (ix86_target_string): remove old style options
6278 -mprefer-avx256 and make -mprefer-avx128 as alias.
6279 (ix86_option_override_internal): Apply defaults for the
6280 -mprefer-vector-width=[128|256] option.
6281 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6282 Implement macros to work with -mprefer-vector-width=.
6283 * config/i386/i386.opt: Implemented option
6284 -mprefer-vector-width=[none|128|256|512].
6285 * doc/invoke.texi: Documentation for
6286 -mprefer-vector-width=[none|128|256|512].
6288 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6290 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6292 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6294 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6296 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6298 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6300 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6301 * expr.c (expand_expr_real_2): Likewise.
6302 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6303 fold_binary_loc): Likewise.
6304 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6305 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6306 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6307 MINUS_EXPR transformations.
6308 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6309 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6310 * tree-inline.c (estimate_operator_cost): Likewise.
6311 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6312 op_symbol_code): Likewise.
6313 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6314 * vr-values.c (extract_range_from_binary_expr): Likewise.
6315 * varasm.c (initializer_constant_valid_p_1): Likewise.
6316 * tree.def: New tree code POINTER_DIFF_EXPR.
6318 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6320 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6321 integer suffix to movbe mnemonic.
6322 (*bswaphi2_movbe): Ditto.
6323 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6325 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6328 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6329 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6330 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6332 2017-11-21 Martin Liska <mliska@suse.cz>
6334 * tree-inline.c (expand_call_inline): Remove not needed
6337 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6338 Jakub Jelinek <jakub@redhat.com>
6341 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6342 Remove static keyword from f variable.
6344 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6346 PR tree-optimization/83086
6347 * gimple-ssa-store-merging.c
6348 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6349 rather than n.base_addr.
6351 2017-11-21 Martin Liska <mliska@suse.cz>
6353 PR rtl-optimization/82044
6354 PR tree-optimization/82042
6355 * dse.c (check_mem_read_rtx): Check for overflow.
6357 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6359 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6362 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6364 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6365 and its corresponding call arg location note.
6367 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6368 Andrew Burgess <andrew.burgess@embecosm.com>
6370 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6372 (arc_return_slot_offset): Likewise.
6373 (arc_eh_return_address_location): New declaration.
6374 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6375 (MUST_SAVE_REGISTER): Add exception handler case.
6376 (MUST_SAVE_RETURN_ADDR): Likewise.
6377 (arc_frame_pointer_required): Likewise.
6378 (arc_frame_pointer_needed): New function.
6379 (arc_compute_frame_size): Changed.
6380 (arc_expand_prologue): Likewise.
6381 (arc_expand_epilogue): Likewise.
6382 (arc_initial_elimination_offset): Likewise.
6383 (arc_return_slot_offset): Delete.
6384 (arc_eh_return_address_location): New function.
6385 (arc_builtin_setjmp_frame_value): Likewise.
6386 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6387 (EH_RETURN_STACKADJ_RTX): Define.
6388 (EH_RETURN_HANDLER_RTX): Likewise.
6389 * config/arc/arc.md (eh_return): Delete.
6391 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6393 * print-tree.h (debug_vec_tree): Remove prototype.
6394 * gdbinit.in (pvt): Remove macro.
6396 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6398 PR tree-optimization/83047
6399 * gimple-ssa-store-merging.c
6400 (imm_store_chain_info::output_merged_store): If the loads with the
6401 same vuse are in different basic blocks, for load_gsi pick a load
6402 location that is dominated by the other loads.
6405 * config/i386/i386.c (ix86_memmodel_check): Start
6406 -Winvalid-memory-model diagnostics with lowercase letter.
6409 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6410 set in_cold_section_p to the partition containing loc_list->first.
6411 When seeing loc_list->last_before_switch node, update secname and
6412 perform range_across_switch second partition handling only after that.
6415 * run-rtl-passes.c: Include debug.h.
6416 (run_rtl_passes): Call debug_hooks->assembly_start.
6417 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6421 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6422 OPTAB_DIRECT in calls to expand_simple_binop.
6424 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6427 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6429 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6432 * spellcheck.h (best_match::blithely_get_best_candidate): New
6435 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6437 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6438 with lower case letter.
6440 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6442 * config/i386/i386.md (bswaphi2): New expander.
6443 (*bswaphi2_movbe): New insn pattern.
6444 (bswaphi -> rorhi pepehole2): New peephole pattern.
6446 2017-11-20 Jeff Law <law@redhat.com>
6448 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6449 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6450 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6451 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6452 gimple-ssa-evrp-analyze.h.
6454 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6455 set BB_VISITED here.
6456 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6458 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6460 * tree-predcom.c: Add general comment on Store-Store chains.
6461 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6463 (get_chain_last_ref_at): Rename into...
6464 (get_chain_last_write_at): ...this.
6465 (get_chain_last_write_before_load): New function.
6466 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6467 CT_STORE_STORE when write reference is added.
6468 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6469 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6470 (initialize_root_vars_store_elim_1): Ditto.
6471 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6472 definition is created.
6473 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6474 chain by replacing it with dominant stored value.
6476 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6478 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6480 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6482 * vr-values.c (extract_range_from_binary_expr): Use a full range
6485 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6487 * config/arm/arm.md (R4_REGNUM): Define constant.
6488 (nonsecure_call_internal): Remove r4 clobber.
6489 (nonsecure_call_value_internal): Likewise.
6490 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6491 clobber and resequence match_operands.
6492 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6493 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6494 (nonsecure_call_value_reg_thumb2): Likewise.
6496 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6498 PR tree-optimization/78821
6499 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6500 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6501 to the address of the base rather than the base itself.
6502 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6503 (find_bswap_or_nop_finalize): New function.
6504 (find_bswap_or_nop): Use it.
6505 (bswap_replace): Return a tree rather than bool, change first
6506 argument from gimple * to gimple_stmt_iterator, allow inserting
6507 into an empty sequence, allow ins_stmt to be NULL - then emit
6508 all stmts into gsi. Fix up MEM_REF address gimplification.
6509 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6510 (struct store_immediate_info): Add N and INS_STMT non-static
6512 (store_immediate_info::store_immediate_info): Initialize them
6513 from newly added ctor args.
6514 (merged_store_group::apply_stores): Formatting fixes. Sort by
6516 (stmts_may_clobber_ref_p): For stores call also
6517 refs_anti_dependent_p.
6518 (gather_bswap_load_refs): New function.
6519 (imm_store_chain_info::try_coalesce_bswap): New method.
6520 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6521 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6522 (imm_store_chain_info::output_merged_store): Fail if number of
6523 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6524 and NOP_EXPR rhs_code.
6525 (pass_store_merging::process_store): Compute n and ins_stmt, if
6526 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6527 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6529 (pass_store_merging::execute): Calculate dominators.
6531 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6532 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6533 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6534 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6535 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6536 class pass_optimize_bswap, bswap_replace,
6537 pass_optimize_bswap::execute): Moved to ...
6538 * gimple-ssa-store-merging.c: ... this file.
6539 Include optabs-tree.h.
6540 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6541 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6542 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6543 anonymous namespace, remove static keywords.
6544 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6545 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6547 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6550 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6552 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6554 * vec.h (debug_helper): New function.
6555 (DEFINE_DEBUG_VEC): New macro.
6556 * hash-set.h (debug_helper): New function.
6557 (DEFINE_DEBUG_HASH_SET): New macro.
6558 * cfg.c (debug_slim (edge)): New function.
6559 Call DEFINE_DEBUG_VEC for edges.
6560 Call DEFINE_DEBUG_HASH_SET for edges.
6561 * cfghooks.c (debug_slim (basic_block)): New function.
6562 Call DEFINE_DEBUG_VEC for basic blocks.
6563 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6564 * print-tree.c (debug_slim): New function to handle trees.
6565 Call DEFINE_DEBUG_VEC for trees.
6566 Call DEFINE_DEBUG_HASH_SET for trees.
6567 (debug (vec<tree, va_gc>) &): Remove.
6568 (debug (<vec<tree, va_gc>) *): Remove.
6569 * print-rtl.c (debug_slim): New function to handle const_rtx.
6570 Call DEFINE_DEBUG_VEC for rtx_def.
6571 Call DEFINE_DEBUG_VEC for rtx_insn.
6572 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6573 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6574 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6575 (debug (vec<rtx_insn *> *ptr): Remove.
6576 (debug_insn_vector): Remove.
6577 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6579 2017-11-20 Tom de Vries <tom@codesourcery.com>
6581 PR rtl-optimization/82020
6582 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6583 IF_THEN_ELSE condition.
6585 2017-11-19 Jeff Law <law@redhat.com>
6587 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6588 of degenerates resulting from ignoring an edge.
6590 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6593 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6595 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6598 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6599 for uninitialied counts.
6601 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6604 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6606 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6609 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6611 2017-11-19 Tom de Vries <tom@codesourcery.com>
6613 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6616 2017-11-19 Tom de Vries <tom@codesourcery.com>
6618 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6621 2017-11-19 Tom de Vries <tom@codesourcery.com>
6623 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6626 2017-11-19 Tom de Vries <tom@codesourcery.com>
6628 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6630 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6632 2017-11-19 Tom de Vries <tom@codesourcery.com>
6634 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6636 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6637 semicolon after MCORE_EXPORT_NAME call.
6639 2017-11-19 Tom de Vries <tom@codesourcery.com>
6642 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6643 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6645 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6647 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6648 (cgraph_edge::clone): Cleanup updating of profile.
6649 * ipa-cp.c (update_profiling_info): Likewise.
6650 * ipa-inline-transform.c (inline_transform): Likewise.
6651 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6652 * ipa-split.c (execute_split_functions): Do not split when function
6654 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6655 * profile-count.c (profile_count::dump): Add global0.
6656 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6658 (profile_count::to_sreal_scale): Likewise.
6659 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6660 (profile_count::combine_with_ipa_count): New function.
6661 * profile-count.h (profile_guessed_global0adjusted): New.
6662 (profile_count::adjusted_zero): New.
6663 (profile_count::global0adjusted): New.
6664 (profile_count::combine_with_ipa_count): New.
6665 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6666 correct profile of return block of split functions.
6667 (copy_cfg_body): Remove unused profile_count.
6668 (copy_body): Likewise.
6669 (expand_call_inline): Update.
6670 (tree_function_versioning): Update.
6672 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6674 * hash-set.h (hash_set::empty): New.
6675 * tree-ssa-threadbackward.h: Delete.
6676 * tree-ssa-threadbackward.c (class thread_jumps): New.
6677 Move max_threaded_paths into class.
6678 (fsm_find_thread_path): Remove arguments that are now in class.
6679 (profitable_jump_thread_path): Rename to...
6680 (thread_jumps::profitable_jump_thread_path): ...this.
6681 (convert_and_register_jump_thread_path): Rename to...
6682 (thread_jumps::convert_and_register_current_path): ...this.
6683 (check_subpath_and_update_thread_path): Rename to...
6684 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6685 (register_jump_thread_path_if_profitable): Rename to...
6686 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6687 (handle_phi): Rename to...
6688 (thread_jumps::handle_phi): ...this.
6689 (handle_assignment): Rename to...
6690 (thread_jumps::handle_assignment): ...this.
6691 (fsm_find_control_statement_thread_paths): Rename to...
6692 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6693 (find_jump_threads_backwards): Rename to...
6694 (thread_jumps::find_jump_threads_backwards): ...this.
6695 Initialize path local data.
6696 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6697 from within thread_jumps class.
6698 (pass_early_thread_jumps::execute): Same.
6700 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6702 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6704 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6706 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6707 to canonical location.
6709 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6712 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6714 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6716 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6718 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6719 rcrt1.o%s/grcrt1.o%s for -static-pie.
6721 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6723 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6724 ix86_shift_rotate_cost): Break out from ...
6725 (ix86_rtx_costs): ... here.
6726 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6729 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6731 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6732 when entry block was promoted unlikely.
6733 (estimate_bb_frequencies): Increase frequency scale.
6734 * profile-count.h (profile_count): Export precision info.
6736 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6738 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6739 disturbin profile of entry block.
6741 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6743 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6746 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6748 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6749 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6751 2017-11-17 Jeff Law <law@redhat.com>
6753 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6754 from evrp_dom_walker class. Various methods moved into new class.
6755 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6756 (evrp_range_analyzer::enter): New method.
6757 (evrp_range_analyzer::leave): New method.
6758 (evrp_dom_walker): Remove delegators no longer needed by this class.
6759 Replace vr_values data member with evrp_range_analyzer
6761 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6762 method extracted from evrp_dom_walker::before_dom_children.
6763 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6764 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6766 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6767 Add private copy constructor and move assignment operators.
6768 Privatize methods and class data where trivially possible.
6769 (evrp_dom_walker::cleanup): New function, extracted from
6770 execute_early_vrp. Simplify access to class data.
6772 * vr-values.h (get_output_for_vrp): Prototype.
6773 * vr-values.c (get_output_for_vrp): New function extracted from
6774 vrp_visit_assignment_or_call and extract_range_from_stmt.
6775 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6777 2017-11-17 Luis Machado <luis.machado@linaro.org>
6779 * config/aarch64/aarch64.c
6780 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6781 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6782 tune_params::AUTOPREFETCHER_WEAK.
6784 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6787 * config/arm/arm.c (arm_valid_target_attribute_rec):
6788 Parse "arch=" and "+<ext>".
6789 (arm_valid_target_attribute_tree): Re-init global options.
6790 (arm_option_override): Make non-static.
6791 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6792 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6793 (__ARM_FEATURE_CRC32): Support undef.
6794 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6795 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6797 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6799 * gdbinit.in (break-on-diagnostic): New command.
6801 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6803 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6804 return to indirect jump for EH return if control-flow
6805 protection is enabled. Change explicit 'false' argument in
6806 pro_epilogue_adjust_stack with a value of
6808 * config/i386/i386.md (simple_return_indirect_internal):
6809 Remove SImode restriction to support 64-bit.
6811 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6813 * combine.c (added_notes_insn): New.
6814 (try_combine): Handle added_notes_insn like added_links_insn.
6815 Rewrite return value code.
6816 (distribute_notes): Set added_notes_insn to the earliest insn we added
6819 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6821 PR rtl-optimization/82621
6822 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6823 dest of one of those SETs is unused.
6825 2017-11-17 Richard Biener <rguenther@suse.de>
6828 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6829 * tree-pretty-print.c (dump_generic_node): Handle
6830 annot_expr_parallel_kind.
6831 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6832 * gimplify.c (gimple_boolify): Likewise.
6834 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6836 * config.gcc (extra_headers): Add cet.h for x86 targets.
6837 * config/i386/cet.h: New file.
6838 * doc/install.texi: Add --enable-cet/--disable-cet.
6840 2017-11-17 Richard Biener <rguenther@suse.de>
6842 PR tree-optimization/83017
6843 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6844 (gen_parallel_loop): Properly count iterations.
6845 (parallelize_loops): Handle loop->can_be_parallel independent
6846 of flag_loop_parallelize_all. Make static profitability test match
6848 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6849 * invoke.texi (parloops-min-per-thread): Document.
6851 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6853 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6854 upstreaming review comments.
6856 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6858 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6859 with fast unaligned access.
6860 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6862 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6864 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6865 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6867 2017-11-17 Richard Biener <rguenther@suse.de>
6869 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6870 folding of references.
6872 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6875 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6876 of replacing call to strncmp with corresponding call to strcmp when
6879 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6881 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6882 option prefer-avx256 for skylake-avx512 configuration.
6883 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6884 (get_builtin_code_for_version): Ditto.
6886 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6887 Monk Chiang <sh.chiang04@gmail.com>
6889 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6890 (FIXED_REGISTERS): Reserve more register numbers.
6891 (CALL_USED_REGISTERS): Likewise.
6892 (REG_ALLOC_ORDER): Likewise.
6893 (REG_CLASS_CONTENTS): Likewise.
6894 (REGISTER_NAMES): Likewise.
6896 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6897 Kito Cheng <kito.cheng@gmail.com>
6899 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6901 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6903 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6905 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6906 parameter list for vec_splats.
6908 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6910 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6912 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6914 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6917 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6918 with the required version of automake 1.11.6.
6920 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6922 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6924 (DU_C2_3_power9): Correct reservation combinations.
6925 (FP_DIV_power9, VEC_DIV_power9): New.
6926 (power9-alu): Split out rotate/shift...
6927 (power9-rot): ...to here, correct dispatch resource.
6928 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6930 (power9-fp): Correct latency.
6931 (power9-sdiv): Add div/sqrt resource.
6932 (power9-ddiv): Correct latency, add div/sqrt resource.
6933 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6934 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6936 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6938 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6940 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6941 switch statement mapping KF built-ins to TF built-ins if we don't
6942 have the proper ISA 3.0 assembler support.
6944 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6946 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6947 (lower_emutls_function_body): Do not compute it.
6949 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6951 * ipa-split.c (split_bb_info): Turn time to sreal.
6952 (split_point): Likewise.
6953 (dump_split_point): Likewise.
6954 (fine_split_points): Likewise.
6955 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6957 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6959 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6962 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6964 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6966 (expensive_function_p): Remove useless assert.
6967 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6969 2017-11-16 Martin Sebor <msebor@redhat.com>
6971 PR tree-optimization/82588
6972 PR tree-optimization/82583
6973 * tree-vrp.c (check_array_ref): Handle flexible array members,
6974 string literals, and inner indices.
6975 (search_for_addr_array): Add detail to diagnostics.
6977 2017-11-16 Nathan Sidwell <nathan@acm.org>
6981 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6982 (SET_DECL_ASSEMBLER_NAME): Forward to
6983 overwrite_decl_assembler_name.
6984 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6985 (overwrite_decl_assembler_name): Declare.
6986 * tree.c (overwrite_decl_assembler_name): New.
6987 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6988 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6989 (LANG_HOOKS_INITIALIZER): Add it.
6990 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6991 * langhooks.c (lhd_set_decl_assembler_name): Use
6992 SET_DECL_ASSEMBLER_NAME.
6993 (lhd_overwrite_decl_assembler_name): Default implementation.
6995 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6996 Jackson Woodruff <jackson.woodruff@arm.com>
6998 PR tree-optimization/71026
6999 * match.pd: Canonicalize constant multiplies in division.
7001 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7003 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7004 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7005 Remove OPT_fomit_frame_pointer entry.
7006 * common/config/arc/arc-common.c: Likewise.
7007 * common/config/arm/arm-common.c: Likewise.
7008 * common/config/avr/avr-common.c: Likewise.
7009 * common/config/c6x/c6x-common.c: Likewise.
7010 * common/config/cr16/cr16-common.c: Likewise.
7011 * common/config/cris/cris-common.c: Likewise.
7012 * common/config/epiphany/epiphany-common.c: Likewise.
7013 * common/config/fr30/fr30-common.c: Likewise.
7014 * common/config/frv/frv-common.c: Likewise.
7015 * common/config/ia64/ia64-common.c: Likewise.
7016 * common/config/iq2000/iq2000-common.c: Likewise.
7017 * common/config/lm32/lm32-common.c: Likewise.
7018 * common/config/m32r/m32r-common.c: Likewise.
7019 * common/config/mcore/mcore-common.c: Likewise.
7020 * common/config/microblaze/microblaze-common.c: Likewise.
7021 * common/config/mips/mips-common.c: Likewise.
7022 * common/config/mmix/mmix-common.c: Likewise.
7023 * common/config/mn10300/mn10300-common.c: Likewise.
7024 * common/config/nios2/nios2-common.c: Likewise.
7025 * common/config/pa/pa-common.c: Likewise.
7026 * common/config/pdp11/pdp11-common.c: Likewise.
7027 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7028 * common/config/riscv/riscv-common.c: Likewise.
7029 * common/config/rs6000/rs6000-common.c: Likewise.
7030 * common/config/rx/rx-common.c: Likewise.
7031 * common/config/s390/s390-common.c: Likewise.
7032 * common/config/sh/sh-common.c: Likewise.
7033 * common/config/sparc/sparc-common.c: Likewise.
7034 * common/config/tilegx/tilegx-common.c: Likewise.
7035 * common/config/tilepro/tilepro-common.c: Likewise.
7036 * common/config/v850/v850-common.c: Likewise.
7037 * common/config/visium/visium-common.c: Likewise.
7038 * common/config/xstormy16/xstormy16-common.c: Likewise.
7039 * common/config/xtensa/xtensa-common.c: Likewise.
7040 * invoke.texi (-fomit-frame-pointer): Update documentation.
7042 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7044 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7046 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7048 * cfg.c (scale_bbs_frequencies_int,
7049 cale_bbs_frequencies_gcov_type): Remove.
7050 * cfg.h (scale_bbs_frequencies_int,
7051 cale_bbs_frequencies_gcov_type): Remove.
7053 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7055 * tree-ssa-loop-manip.c
7056 (scale_dominated_blocks_in_loop): Update to profile counts.
7057 (tree_transform_and_unroll_loop): Likewise.
7059 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7061 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7062 scale_bbs_frequencies_int.
7064 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7066 * final.c (compute_alignments): Use counts rather than frequencies.
7068 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7070 * cfgloopanal.c: Include sreal.h
7071 (average_num_loop_insns): Use counts and sreal for accounting.
7073 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7075 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7078 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7079 Kito Cheng <kito.cheng@gmail.com>
7081 * config/nds32/constraints.md: Provide more constraints.
7082 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7083 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7084 support constraints usage.
7086 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7088 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7090 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7091 Kito Cheng <kito.cheng@gmail.com>
7093 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7094 * config/nds32/nds32.opt: Refine the layout.
7095 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7096 TARGET_EXT_STRING): Support new options.
7097 * config/nds32/nds32.h: Likewise.
7098 * config/nds32/nds32.md: Likewise.
7099 * config/nds32/nds32-predicates.c: Likewise.
7100 * config/nds32/constraints.md: Likewise.
7101 * common/config/nds32/nds32-common.c: Likewise.
7103 2017-11-16 Julia Koval <julia.koval@intel.com>
7106 * config/i386/gfniintrin.h: Add sse check.
7107 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7109 2017-11-16 Julia Koval <julia.koval@intel.com>
7111 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7112 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7113 (ix86_handle_option): Handle -mavx512vbmi2.
7114 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7115 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7116 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7117 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7118 (ix86_valid_target_attribute_inner_p): Ditto.
7119 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7120 * config/i386/i386.opt (mavx512vbmi2): New option.
7121 * doc/invoke.texi: Add new option.
7123 2017-11-16 Julia Koval <julia.koval@intel.com>
7125 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7126 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7127 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7128 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7129 _mm512_gf2p8mul_epi8): New intrinsics.
7130 * config/i386/i386-builtin-types.def
7131 (V64QI_FTYPE_V64QI_V64QI): New type.
7132 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7133 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7134 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7135 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7136 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7137 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7139 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7141 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7142 explicitly as a stream of bytes.
7144 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7146 * config/rs6000/altivec.h (vec_xst_be): New #define.
7147 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7148 and externalize from *altivec_vperm_<mode>_internal.
7149 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7151 (XL_BE_V8HI): Likewise.
7152 (XL_BE_V4SI): Likewise.
7153 (XL_BE_V4SI): Likewise.
7154 (XL_BE_V2DI): Likewise.
7155 (XL_BE_V4SF): Likewise.
7156 (XL_BE_V2DF): Likewise.
7157 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7158 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7159 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7160 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7161 VSX_BUILTIN_VEC_XST_BE.
7162 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7163 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7165 (altivec_init_builtins): Replace conditional calls to def_builtin
7166 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7167 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7168 with unconditional calls. Remove calls to def_builtin for
7169 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7170 __builtin_vec_xst_be.
7171 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7172 to define_expand, and add alternate RTL generation for P8.
7173 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7174 vsx_ld_elemrev_v8hi.
7175 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7176 add alternate RTL generation for P8.
7177 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7178 vsx_ld_elemrev_v16qi.
7179 (vsx_st_elemrev_v8hi): Convert define_insn
7180 to define_expand, and add alternate RTL generation for P8.
7181 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7182 vsx_st_elemrev_v8hi.
7183 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7184 add alternate RTL generation for P8.
7185 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7186 vsx_st_elemrev_v16qi.
7188 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7191 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7192 TARGET_AVX512ER check.
7193 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7194 neither -mzeroupper nor -mno-zeroupper is used and
7195 TARGET_EMIT_VZEROUPPER is set.
7196 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7197 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7199 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7201 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7202 folding of vector compares.
7203 (fold_build_vec_cmp): New helper function.
7204 (fold_compare_helper): New helper function.
7205 (builtin_function_type): Add compare builtins to the list of functions
7206 having unsigned arguments. Cosmetic updates to comment indentation.
7207 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7208 the not+eq combination.
7210 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7212 PR tree-optimization/82726
7213 PR tree-optimization/70754
7214 * tree-predcom.c (order_drefs_by_pos): New function.
7215 (combine_chains): Move code setting has_max_use_after to...
7216 (try_combine_chains): ...here. New parameter. Sort combined chains
7217 according to position information.
7218 (tree_predictive_commoning_loop): Update call to above function.
7219 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7221 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7223 PR tree-optimization/82726
7225 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7227 PR tree-optimization/70754
7228 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7229 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7230 combined stmt before it if not NULL.
7231 (combine_chains): Process refs reversely and compute dominance point
7235 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7237 PR tree-optimization/79663
7238 * tree-predcom.c (combine_chains): Process refs in reverse order
7239 only for ZERO length chains, and add explaining comment.
7241 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7243 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7244 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7245 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7247 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7249 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7251 2017-11-15 Martin Liska <mliska@suse.cz>
7253 * tree-cfg.c (pass_warn_function_return::execute):
7254 Compare warn_return_type for greater than zero.
7256 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7260 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7261 to return true on Xeon and not on Xeon Phi.
7262 (ix86_check_avx256_register): Changed to ...
7263 (ix86_check_avx_upper_register): ... this. Add extra check for
7264 VALID_AVX512F_REG_OR_XI_MODE.
7265 (ix86_avx_u128_mode_needed): Changed
7266 ix86_check_avx256_register to ix86_check_avx_upper_register.
7267 (ix86_check_avx256_stores): Changed to ...
7268 (ix86_check_avx_upper_stores): ... this. Changed
7269 ix86_check_avx256_register to ix86_check_avx_upper_register.
7270 (ix86_avx_u128_mode_after): Changed
7271 avx_reg256_found to avx_upper_reg_found. Changed
7272 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7273 (ix86_avx_u128_mode_entry): Changed
7274 ix86_check_avx256_register to ix86_check_avx_upper_register.
7275 (ix86_avx_u128_mode_exit): Ditto.
7276 * config/i386/i386.h: (host_detect_local_cpu): New define.
7278 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7280 * config/arm/xgene1.md (xgene1): Split into automatons
7281 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7282 (xgene1_f_load): Adjust reservations and/or types.
7283 (xgene1_f_store): Likewise.
7284 (xgene1_load_pair): Likewise.
7285 (xgene1_store_pair): Likewise.
7286 (xgene1_fp_load1): Likewise.
7287 (xgene1_load1): Likewise.
7288 (xgene1_store1): Likewise.
7289 (xgene1_move): Likewise.
7290 (xgene1_alu): Likewise.
7291 (xgene1_simd): Likewise.
7292 (xgene1_bfm): Likewise.
7293 (xgene1_neon_load1): Likewise.
7294 (xgene1_neon_store1): Likewise.
7295 (xgene1_neon_logic): Likewise.
7296 (xgene1_neon_st1): Likewise.
7297 (xgene1_neon_ld1r): Likewise.
7298 (xgene1_alu_cond): Added.
7299 (xgene1_shift_reg): Likwise.
7300 (xgene1_bfx): Likewise.
7301 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7303 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7306 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7308 (can_widen_mult_without_libcall): New function.
7309 (expand_mul_overflow): If only checking unsigned mul overflow,
7310 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7311 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7312 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7313 (expand_DIVMOD): Formatting fix.
7314 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7315 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7316 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7318 PR tree-optimization/82977
7319 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7320 constructed temporary to strlen_to_stridx.put.
7322 2017-11-15 Martin Liska <mliska@suse.cz>
7324 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7325 * configure: Regenerate.
7327 2017-11-15 Martin Liska <mliska@suse.cz>
7330 * config/sh/sh-mem.cc: Use proper probability for
7333 2017-11-14 Jeff Law <law@redhat.com>
7335 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7336 the red zone for stack_clash_protection_final_dynamic_probe targets
7337 when the total dynamic stack size is zero bytes.
7339 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7340 blocks is post order.
7342 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7344 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7345 * final.c (rest_of_clean_state): Set it for the
7346 -fcompare-debug dump.
7347 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7348 class when TDF_COMPARE_DEBUG is set.
7350 * dwarf2out.c (gen_producer_string): Discard
7353 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7356 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7357 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7358 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7359 (__TGMATH_REAL_2_3): Remove macros.
7360 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7361 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7363 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7365 (remquo): Define using __TGMATH_REAL_3.
7367 2017-11-14 Jeff Law <law@redhat.com>
7369 * vr-values.c: New file with contents extracted from tree-vrp.c.
7370 * Makefile.in (OBJS): Add vr-values.o
7371 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7372 (set_value_range, set_and_canonicalize_value_range): Likewise.
7373 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7374 (value_range_constant_singleton, symbolic_range_p): Likewise.
7375 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7376 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7377 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7378 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7379 (find_case_label_range, find_case_label_index): Likewise.
7380 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7381 (range_int_cst_singleton_p, value_inside_range): Likewise.
7382 (get_single_symbol): Likewise.
7383 (switch_update): Move structure definition here.
7384 (to_remove_edges, to_update_switch_stmts): Provide externs.
7385 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7386 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7387 (vrp_val_is_min, set_value_range): Likewise.
7388 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7389 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7390 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7391 (range_is_nonnull, range_int_cst_p): Likewwise.
7392 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7393 (get_single_symbol, operand_less_p): Likewise
7394 (compare_values_warnv, compare_values): Likewise.
7395 (value_inside_range, value_range_constant_singleton): Likewise.
7396 (zero_nonzero_bitgs_from_vr): Likewise.
7397 (extract_range_from_binary_expr_1): Likewise.
7398 (overflow_comparison_p): Likewise.
7399 (to_remove_edges, to_update_switch_stmts): Likewise.
7400 (find_case_label-index, find_case_label_range): Likewise.
7401 (switch_update, set_value_range_to_nonnegative): Remove.
7402 (set_value_range_to_truthvalue): Likewise.
7403 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7404 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7405 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7406 (find_case_label_ranges, test_for_singularity): Likewise.
7407 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7408 (x_vr_values): Move to its remaining use site.
7410 2017-11-10 Jeff Law <law@redhat.com>
7412 * vr-values.h (VR_INITIALIZER): Move #define here.
7413 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7414 * Makefile.in (OBJS): Add tree-evrp.o
7415 * tree-vrp.h (assert_info): Move structure definition here.
7416 (set_value_range_to_varying): Prototype.
7417 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7418 (infer_value_range, register_edge_assert_for): Likewise.
7419 (stmt_interesting_for_vrp): Likewise.
7420 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7421 (set_value_range_to_varying): No longer static.
7422 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7423 (infer_value_range, register_edge_assert_for): Likewise.
7425 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7427 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7428 generate the XXBRD instruction.
7430 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7431 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7432 as being compatible if -mabi=ieeelongdouble.
7433 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7434 to setup float128 built-ins with hardware support.
7435 (BU_FLOAT128_HW_2): Likewise.
7436 (BU_FLOAT128_HW_3): Likewise.
7437 (BU_FLOAT128_HW_VSX_1): Likewise.
7438 (BU_FLOAT128_HW_VSX_2): Likewise.
7439 (scalar_extract_expq): Change float128 built-in functions to
7440 accommodate having both KFmode and TFmode functions. Use the
7441 KFmode variant as the default.
7442 (scalar_extract_sigq): Likewise.
7443 (scalar_test_neg_qp): Likewise.
7444 (scalar_insert_exp_q): Likewise.
7445 (scalar_insert_exp_qp): Likewise.
7446 (scalar_test_data_class_qp): Likewise.
7447 (sqrtf128_round_to_odd): Delete processing the round to odd
7448 built-in functions as special built-in functions, and define them
7449 as float128 built-ins. Use the KFmode variant as the default.
7450 (truncf128_round_to_odd): Likewise.
7451 (addf128_round_to_odd): Likewise.
7452 (subf128_round_to_odd): Likewise.
7453 (mulf128_round_to_odd): Likewise.
7454 (divf128_round_to_odd): Likewise.
7455 (fmaf128_round_to_odd): Likewise.
7456 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7457 support for KFmode and TFmode xststdcqp calls.
7458 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7459 point, switch the built-in handlers for the get/set float128
7460 exponent, get float128 mantissa, float128 test built-ins, and the
7461 float128 round to odd built-in functions. Eliminate creating the
7462 float128 round to odd built-in functions as special built-ins.
7463 (rs6000_init_builtins): Eliminate special creation of the float128
7464 round to odd built-in functions.
7465 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7466 function insns to support both TFmode and KFmode variants.
7467 (xsxsigqp_<mode>): Likewise.
7468 (xsiexpqpf_<mode>): Likewise.
7469 (xsiexpqp_<mode>): Likewise.
7470 (xststdcqp_<mode>): Likewise.
7471 (xststdcnegqp_<mode>): Likewise.
7472 (xststdcqp_<mode>): Likewise.
7474 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7476 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7477 unused path_in_freq_ptr parameter.
7478 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7480 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7482 * ipa-inline.c (edge_badness): Dump sreal frequency.
7483 (compute_inlined_call_time): Match natural implementaiton ...
7484 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7485 forgotten division by CGRAPH_FREQ_BASE.
7487 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7489 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7490 Solaris 11. Update comment.
7491 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7493 * config/sol2.h (STARTFILE_SPEC): Likewise.
7494 * configure: Regenerate.
7496 2017-11-14 Carl Love <cel@us.ibm.com>
7498 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7499 le_ and be_ prefixes to swap* variables. Remove
7500 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7503 2017-11-14 Jason Merrill <jason@redhat.com>
7505 Support GTY((cache)) on hash_map.
7506 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7507 (ggc_cache_remove): Override it instead of ggc_mx.
7508 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7509 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7510 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7511 (simple_cache_map_traits): Override maybe_mx.
7512 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7513 (hash_map): Friend gt_cleare_cache.
7514 (gt_cleare_cache): New.
7515 * tree.h (tree_cache_traits): New hash_map traits class.
7516 (tree_cache_map): New typedef.
7518 2017-11-14 Richard Biener <rguenther@suse.de>
7520 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7521 paramter and handling.
7522 (cleanup_control_flow_bb): Likewise.
7523 (cleanup_control_flow_pre): New helper performing a DFS walk
7524 to call cleanup_control_flow_bb in PRE order.
7525 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7526 via cleanup_control_flow_pre.
7528 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7530 * config/aarch64/aarch64-simd.md
7531 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7532 (*aarch64_simd_bsl<mode>_alt): Likewise.
7533 (aarch64_simd_bsldi_internal): New.
7534 (aarch64_simd_bsldi_alt): Likewise.
7536 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7538 * tracer.c (better_p): Do not compare frequencies.
7539 * reg-stack.c (better_edge): Likewise.
7540 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7543 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7545 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7546 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7547 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7548 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7549 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7550 * predict.c (maybe_hot_frequency_p): Inline to ...
7551 (maybe_hot_count_p): ... here; rewrite to counts.
7552 (counts_to_freqs): Rename to ...
7553 (update_max_bb_count): ... this one.
7554 (expensive_function_p): Use counts.
7555 (estimate_bb_frequencies): Update.
7556 (rebuild_frequencies): Update.
7557 * predict.h (counts_to_freqs): Rename to ...
7558 (update_max_bb_count): ... this one.
7559 * profile.c (compute_branch_probabilities): Add debug info
7560 * tree-inline.c (expand_call_inline): Update debug info.
7561 (optimize_inline_calls): Use update_max_bb_count..
7562 (tree_function_versioning): Use update_max_bb_count..
7563 * value-prof.c (gimple_value_profile_transformations):
7564 Do not use update_max_bb_count.
7566 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7568 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7569 always use frequencies.
7571 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7573 * bb-reorder.c: Remove frequencies from comments.
7574 (better_edge_p): Use profile counts.
7575 (find_traces): Dump profile counts.
7576 (rotate_loop): Use profile counts.
7577 (find_traces_1_round): Likewise.
7578 (connect_better_edge_p): Use counts instead of probabilities for
7580 (copy_bb_p): Drop early check for non-0 frequency.
7581 (sanitize_hot_paths): Update comments.
7583 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7585 * ipa-split.c (struct split_point): Add count.
7586 (consider_split): Do not compute incoming frequency; compute incoming
7587 count and store it to split_point.
7588 (split_function): Set count of the call to split part correctly.
7590 2017-11-13 Carl Love <cel@us.ibm.com>
7592 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7594 2017-11-13 Tom Tromey <tom@tromey.com>
7596 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7598 2017-11-13 Carl Love <cel@us.ibm.com>
7600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7601 Add support for builtins:
7602 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7603 vector {un,}signed {char,int,short},
7604 vector {un,}signed {char,int,short}) arguments.
7605 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7606 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7607 Add BU_P9V_AV_2 expansions for the builtins.
7608 * config/rs6000/altivec.h (vec_first_match_index,
7609 vec_first_mismatch_index, vec_first_match_or_eos_index,
7610 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7611 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7612 new extern declaration.
7613 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7614 * config/rs6000/vsx.md (first_match_index_<mode>,
7615 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7616 first_mismatch_or_eos_index_<mode>): Add define expand.
7617 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7618 * doc/extend.texi: Update the built-in documenation file for the new
7621 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7623 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7624 into the min/max operations for _Float<N> and _Float<N>X types.
7626 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7629 * dwarf2out.c (do_eh_frame): New static variable.
7630 (dwarf2out_begin_prologue): Set it.
7631 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7633 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7635 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7637 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7639 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7642 * debug.h (dwarf2out_do_eh_frame): Declare.
7643 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7644 (dwarf2out_do_frame): Use it.
7645 (dwarf2out_do_cfi_asm): Likewise.
7646 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7647 (dwarf2out_assembly_start): Likewise.
7648 (dwarf2out_begin_prologue): Fix comment.
7649 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7650 if the target needs either debug or unwind DWARF2 info.
7651 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7652 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7653 -ftrapping-math, -ftrapv and -fwrapv.
7655 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7657 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7658 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7659 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7660 (estimate_edge_size_and_time): Likewise.
7661 (ipa_merge_fn_summary_after_inlining): Likewise.
7662 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7663 (compute_uninlined_call_time): Use sreal_frequency.
7664 (compute_inlined_call_time): Likewise.
7665 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7666 * profile-count.c: Include sreal.h.
7667 (profile_count::to_sreal_scale): New.
7668 * profile-count.h: Forward declare sreal.
7669 (profile_count::to_sreal_scale): Declare.
7671 2017-11-13 Nathan Sidwell <nathan@acm.org>
7673 * diagnostic.c (maybe_line_and_column): New.
7674 (diagnostic_get_location_text): Use it.
7675 (diagnostic_report_current_module): Likewise.
7676 (test_diagnostic_get_location_text): Add tests.
7678 2017-11-13 Luis Machado <luis.machado@linaro.org>
7680 * doc/md.texi (Specifying processor pipeline description): Fix
7681 incorrect latency for the div instruction example.
7683 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7685 PR tree-optimization/78821
7686 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7687 that bit_not_p is the same.
7688 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7689 (split_group): Count precisely bit_not_p bits in each statement.
7690 (invert_op): New function.
7691 (imm_store_chain_info::output_merged_store): Use invert_op to
7692 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7693 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7695 2017-11-13 Martin Liska <mliska@suse.cz>
7697 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7698 (struct source_info): Likewise.
7699 (add_branch_counts): Likewise.
7700 (add_line_counts): Likewise.
7701 (function_summary): Likewise.
7702 (output_intermediate_line): Likewise.
7703 (generate_results): Likewise.
7705 2017-11-13 Martin Liska <mliska@suse.cz>
7707 * gcov.c (struct block_info): Remove typedef for block_t.
7708 (struct line_info): Likewise.
7709 (line_info::has_block): Likewise.
7710 (EXIT_BLOCK): Likewise.
7711 (unblock): Likewise.
7712 (circuit): Likewise.
7713 (get_cycles_count): Likewise.
7714 (process_file): Likewise.
7715 (read_graph_file): Likewise.
7716 (solve_flow_graph): Likewise.
7717 (find_exception_blocks): Likewise.
7718 (add_line_counts): Likewise.
7719 (accumulate_line_info): Likewise.
7720 (output_line_details): Likewise.
7722 2017-11-13 Martin Liska <mliska@suse.cz>
7724 * gcov.c (struct arc_info): Remove typedef for arc_t.
7725 (struct line_info): Likewise.
7726 (add_branch_counts): Likewise.
7727 (output_branch_count): Likewise.
7728 (function_info::~function_info): Likewise.
7729 (circuit): Likewise.
7730 (output_intermediate_line): Likewise.
7731 (read_graph_file): Likewise.
7732 (solve_flow_graph): Likewise.
7733 (find_exception_blocks): Likewise.
7734 (add_line_counts): Likewise.
7735 (accumulate_line_info): Likewise.
7736 (output_line_details): Likewise.
7737 (output_function_details): Likewise.
7739 2017-11-13 Martin Liska <mliska@suse.cz>
7741 * gcov.c (struct function_info): Remove typedef for function_t.
7742 (struct source_info): Likewise.
7743 (source_info::get_functions_at_location): Likewise.
7744 (solve_flow_graph): Likewise.
7745 (find_exception_blocks): Likewise.
7746 (add_line_counts): Likewise.
7747 (output_intermediate_file): Likewise.
7748 (process_file): Likewise.
7749 (generate_results): Likewise.
7750 (release_structures): Likewise.
7751 (read_graph_file): Likewise.
7752 (read_count_file): Likewise.
7753 (accumulate_line_counts): Likewise.
7754 (output_lines): Likewise.
7756 2017-11-13 Martin Liska <mliska@suse.cz>
7758 * gcov.c (function_info::function_info): Remove num_counts
7759 and add vector<gcov_type>.
7760 (function_info::~function_info): Use the vector.
7761 (process_file): Likewise.
7762 (read_graph_file): Likewise.
7763 (read_count_file): Likewise.
7764 (solve_flow_graph): Likewise.
7766 2017-11-13 Martin Liska <mliska@suse.cz>
7768 * gcov.c (function_info::is_artificial): New function.
7769 (process_file): Erase all artificial early.
7770 (generate_results): Skip as all artificial are already
7773 2017-11-13 Martin Liska <mliska@suse.cz>
7775 * gcov.c (read_graph_file): Store to global vector of functions.
7776 (read_count_file): Iterate the vector.
7777 (process_file): Likewise.
7778 (generate_results): Likewise.
7779 (release_structures): Likewise.
7781 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7783 PR tree-optimization/82954
7784 * gimple-ssa-store-merging.c
7785 (imm_store_chain_info::coalesce_immediate_stores): If
7786 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7788 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7790 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7791 Upddate call to ENDIAN_LANE_N.
7792 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7793 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7794 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7795 and use aarch64_endian_lane_rtx.
7796 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7798 2017-11-12 Tom de Vries <tom@codesourcery.com>
7800 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7802 2017-11-12 Tom de Vries <tom@codesourcery.com>
7804 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7806 2017-11-12 Tom de Vries <tom@codesourcery.com>
7808 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7810 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7811 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7812 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7814 2017-11-11 Martin Sebor <msebor@redhat.com>
7817 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7820 * prefic.c (translate_name): Replace strncpy with memcpy to
7821 avoid -Wstringop-truncation.
7823 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7825 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7827 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7829 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7831 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7833 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7836 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7838 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7839 enable generating XXBRH if the value is in a vector register.
7840 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7841 value is in a vector register.
7842 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7843 register to register bswap64's instead of doing the GPR sequence
7844 used on previous machines.
7845 (bswapdi2_xxbrd): New insn.
7846 (bswapdi2_reg): Disallow on ISA 3.0.
7847 (register to register bswap64 splitter): Do not split the insn on
7848 ISA 3.0 systems that use XXBRD.
7850 2017-11-10 Martin Sebor <msebor@redhat.com>
7853 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7855 (find_subframework_file): Same.
7857 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7859 * auto-profile.c (afdo_indirect_call): Drop frequency.
7860 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7861 (cgraph_node::create_edge): Drop frequency argument.
7862 (cgraph_node::create_indirect_edge): Drop frequency argument.
7863 (cgraph_edge::make_speculative): Drop frequency arguments.
7864 (cgraph_edge::resolve_speculation): Do not update frequencies
7865 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7866 (cgraph_node::dump): Check consistency in IPA mode.
7867 (cgraph_edge::maybe_hot_p): Use IPA counter.
7868 (cgraph_edge::verify_count_and_frequency): Rename to ...
7869 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7870 (cgraph_node::verify_node): Update.
7871 * cgraph.h (struct cgraph_edge): Drop frequency.
7872 (cgraph_edge::frequency): New function.
7873 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7875 (cgraph_edge::rebuild_edges): Likewise.
7876 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7877 (duplicate_thunk_for_node): Do not pass frequency.
7878 (cgraph_node::create_clone): Scale only counts.
7879 (cgraph_node::create_virtual_clone): Do not pass frequency.
7880 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7881 (cgraph_node::create_version_clone): Do not pass frequency.
7882 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7883 (cgraph_node::expand_thunk): Do not pass frequency.
7884 (cgraph_node::create_wrapper): Do not pass frequency.
7885 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7887 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7888 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7889 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7890 (gather_caller_stats): Use frequency function.
7891 (ipcp_cloning_candidate_p): Use frequency function.
7892 (ipcp_propagate_stage): Use frequency function.
7893 (get_info_about_necessary_edges): Use frequency function.
7894 (update_profiling_info): Update only IPA profile.
7895 (update_specialized_profile): Use frequency functoin.
7896 (perhaps_add_new_callers): Update only IPA profile.
7897 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7898 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7899 (dump_ipa_call_summary): Use frequency function.
7900 (estimate_edge_size_and_time): Use frequency function.
7901 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7902 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7903 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7904 (update_noncloned_counts): ... ths one; scale counts only.
7905 (clone_inlined_nodes): Do not scale frequency.
7906 (inline_call): Do not pass frequency.
7907 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7908 (compute_inlined_call_time): Use IPA profile.
7909 (want_inline_small_function_p): Use IPA profile.
7910 (want_inline_self_recursive_call_p): Use IPA profile.
7911 (edge_badness): Use IPA profile.
7912 (lookup_recursive_calls): Use IPA profile.
7913 (recursive_inlining): Do not pass frequency.
7914 (resolve_noninline_speculation): Do not update frequency.
7915 (inline_small_functions): Collect max of IPA profile.
7916 (dump_overall_stats): Dump IPA porfile.
7917 (dump_inline_stats): Dump IPA porfile.
7918 (ipa_inline): Collect IPA stats.
7919 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7920 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7921 (ipa_propagate_frequency): Use frequency function.
7922 (ipa_profile): Cleanup.
7923 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7924 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7925 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7926 (input_node): Do not stream frequency.
7927 (input_edge): Do not stream frequency.
7928 (merge_profile_summaries): Scale only IPA profiles.
7929 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7930 * predict.c (drop_profile): Do not recompute frequency.
7931 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7932 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7933 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7934 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7935 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7936 * tree-inline.c (copy_bb): Do not scale frequency.
7937 (expand_call_inline): Do not scale frequency.
7938 (tree_function_versioning): Do not scale frequency.
7939 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7941 2017-11-10 Julia Koval <julia.koval@intel.com>
7943 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7944 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7945 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7946 (_mm256_maskz_gf2p8affine_epi64_epi8)
7947 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7948 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7949 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7950 (__builtin_ia32_vgf2p8affineqb_v32qi)
7951 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7952 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7954 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7958 (arm_option_override): Refactor.
7959 (arm_option_reconfigure_globals): New.
7960 (arm_options_perform_arch_sanity_checks): New.
7961 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7963 (arm_options_perform_arch_sanity_checks): Likewise
7965 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7967 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7968 (power9-qpmul): New.
7969 * rs6000/rs6000.md ("type" attr): Add qmul.
7970 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7971 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7972 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7974 2017-11-10 Martin Sebor <msebor@redhat.com>
7977 * builtins.c (compute_objsize): Handle arrays that
7978 compute_builtin_object_size likes to fail for. Make extern.
7979 * builtins.h (compute_objsize): Declare.
7980 (check_strncpy_sizes): New function.
7981 (expand_builtin_strncpy): Call check_strncpy_sizes.
7982 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7983 -Wstringop-truncation.
7984 (gimple_fold_builtin_strncat): Same.
7985 * gimple.c (gimple_build_call_from_tree): Set call location.
7986 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7987 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7988 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7989 (handle_builtin_strlen): Use strlen_to_stridx.
7990 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7992 Use strlen_to_stridx.
7993 (pass_strlen::execute): Release strlen_to_stridx.
7994 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7995 (-Wstringop-truncation): Document new option.
7997 2017-11-10 Martin Liska <mliska@suse.cz>
7999 PR gcov-profile/82702
8000 * gcov.c (main): Handle intermediate files in a different
8002 (get_gcov_intermediate_filename): New function.
8003 (output_gcov_file): Remove support of intermediate files.
8004 (generate_results): Allocate intermediate file.
8005 (release_structures): Clean-up properly fn_end.
8006 (output_intermediate_file): Start iterating with line 1.
8008 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8010 PR tree-optimization/82929
8011 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8012 ops_swapped_p non-static data member.
8013 (store_immediate_info::store_immediate_info): Clear it.
8014 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8015 ops set ops_swapped_p.
8016 (count_multiple_uses): Handle ops_swapped_p.
8018 2017-11-10 Martin Liska <mliska@suse.cz>
8020 * coverage.c (coverage_init): Stream information about
8021 support of has_unexecuted_blocks.
8022 * doc/gcov.texi: Document that.
8023 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8024 * gcov.c (read_graph_file): Likewise.
8025 (output_line_beginning): Fix a small issue with
8028 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8030 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8031 reference of trivial component.
8033 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8036 * gimple-ssa-store-merging.c
8037 (pass_store_merging::terminate_all_aliasing_chains): For
8038 gimple_store_p stmts also call refs_output_dependent_p.
8040 PR rtl-optimization/82913
8041 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8044 2017-11-09 Jeff Law <law@redhat.com>
8046 * vr-values.h: New file with vr_values class.
8047 * tree-vrp.c: Include vr-values.h
8048 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8049 data objects into the vr_values class.
8050 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8051 (get_value_range): Make it a member function within vr_values class.
8052 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8053 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8054 (op_with_constant_singleton_value_range): Likewise.
8055 (extract_range_for_var_from_comparison_expr): Likewise.
8056 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8057 (extract_range_from_binary_expr): Likewise.
8058 (extract_range_from_unary_expr): Likewise.
8059 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8060 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8061 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8062 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8063 (compare_name_with_value, compare_names): Likewise.
8064 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8065 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8066 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8067 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8068 (extract_range_from_phi_node): Likewise.
8069 (simplify_truth_ops_using_ranges): Likewise.
8070 (simplify_div_or_mod_using_ranges): Likewise.
8071 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8072 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8073 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8074 (simplify_float_conversion_using_ranges): Likewise.
8075 (simplify_internal_call_using_ranges): Likewise.
8076 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8077 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8078 poitner via x_vr_values for calls into gimple folder.
8079 (vrp_initialize_lattice): Make this the vr_values ctor.
8080 (vrp_free_lattice): Make this the vr_values dtor.
8081 (set_vr_value): New function.
8082 (class vrp_prop): Add vr_values data member. Add various member
8083 functions as well as member functions that delegate to vr_values.
8084 (check_array_ref): Make a member function within vrp_prop class.
8085 (search_for_addr_array, vrp_initialize): Likewise.
8086 (vrp_finalize): Likewise. Revamp to avoid direct access to
8087 vr_value, values_propagated, etc.
8088 (check_array_bounds): Extract vrp_prop class instance pointer from
8089 walk info structure. Use it to call member functions.
8090 (check_all_array_refs): Make a member function within vrp_prop class.
8091 Smuggle class instance pointer via walk info structure.
8092 (x_vr_values): New local static.
8093 (vrp_valueize): Use x_vr_values to get class instance.
8094 (vr_valueize_1): Likewise.
8095 (class vrp_folder): Add vr_values data member. Add various member
8096 functions as well as member functions that delegate to vr_values.
8097 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8098 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8099 class instance from vr_values. Use it to call member functions.
8100 (vrp_dom_walker): Add vr_values data member.
8101 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8102 instance via x_vr_values.
8103 (identify_jump_threads): Accept vr_values as argument. Store
8104 it into the walker structure.
8105 (evrp_dom_walker): Add vr_values class data member. Add various
8107 (evrp_dom_walker::try_find_new_range): Use vr_values data
8108 member to access the memory allocator.
8109 (evrp_dom_walker::before_dom_children): Store vr_values class
8110 instance into the vrp_folder class.
8111 (evrp_dom_walker::push_value_range): Rework to avoid direct
8112 access to num_vr_values and vr_value.
8113 (evrp_dom_walker::pop_value_range): Likewise.
8114 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8115 Use vr_values to get to dump_all_value_ranges member function.
8116 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8117 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8118 Pass vr_values class instance down to identify_jump_threads.
8119 Remove call to vrp_free_lattice.
8120 (debug_all_value_ranges): Remove.
8122 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8123 (vrp_folder): Likewise.
8125 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8126 Get it from the existing bitmap instead.
8127 (vrp_intersect_ranges_1): Likewise.
8129 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8131 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8133 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8134 set bit_not_p to it.
8135 (imm_store_chain_info::coalesce_immediate_stores): Break group
8136 if bit_not_p is different.
8137 (count_multiple_uses, split_group,
8138 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8139 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8140 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8141 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8142 to store_immediate_info ctor.
8144 2017-11-09 Jim Wilson <jimw@sifive.com>
8146 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8147 (scan_prog_file): Likewise.
8149 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8151 * bb-reorder.c (max_entry_frequency): Remove.
8152 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8153 connect_traces, push_to_next_round_p): Remove prototypes.
8154 (find_traces_1_round): Use counts only.
8155 (push_to_next_round_p): Likewise.
8156 (find_traces): Likewise.
8157 (rotate_loop): Likewise.
8158 (find_traces_1_round): Likewise.
8159 (connect_traces): Likewise.
8160 (edge_order): Likewise.
8162 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8164 * config/arm/arm.c (output_return_instruction): Add comments to
8165 indicate requirement for cmse_nonsecure_entry return to account
8166 for the size of clearing instruction output here.
8167 (thumb_exit): Likewise.
8168 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8169 return in hardfloat mode.
8171 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8173 * config/rs6000/rs6000.c (machine_function): Add a bool,
8174 "toc_is_wrapped_separately".
8175 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8176 if it wasn't explicitly set or unset, we are optimizing for speed, and
8177 doing separate shrink-wrapping.
8178 (rs6000_get_separate_components): Enable the TOC component if
8179 saving the TOC register in the prologue.
8180 (rs6000_components_for_bb): Handle the TOC component.
8181 (rs6000_emit_prologue_components): Store the TOC register where needed.
8182 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8183 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8185 2017-11-09 Martin Jambor <mjambor@suse.cz>
8187 * ipa-param-manipulation.c: New file.
8188 * ipa-param-manipulation.h: Likewise.
8189 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8190 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8191 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8192 (ipa_parm_adjustment): Likewise.
8193 (ipa_parm_adjustment_vec): Likewise.
8194 (ipa_get_vector_of_formal_parms): Moved declaration to
8195 ipa-param-manipulation.h.
8196 (ipa_get_vector_of_formal_parm_types): Likewise.
8197 (ipa_modify_formal_parameters): Likewise.
8198 (ipa_modify_call_arguments): Likewise.
8199 (ipa_combine_adjustments): Likewise.
8200 (ipa_dump_param_adjustments): Likewise.
8201 (ipa_modify_expr): Likewise.
8202 (ipa_get_adjustment_candidate): Likewise.
8203 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8204 ipa-param-manipulation.c.
8205 (ipa_get_vector_of_formal_parm_types): Likewise.
8206 (ipa_modify_formal_parameters): Likewise.
8207 (ipa_modify_call_arguments): Likewise.
8208 (ipa_modify_expr): Likewise.
8209 (get_ssa_base_param): Likewise.
8210 (ipa_get_adjustment_candidate): Likewise.
8211 (index_in_adjustments_multiple_times_p): Likewise.
8212 (ipa_combine_adjustments): Likewise.
8213 (ipa_dump_param_adjustments): Likewise.
8214 * tree-sra.c: Also include ipa-param-manipulation.h
8215 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8218 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8219 Alan Hayward <alan.hayward@arm.com>
8220 David Sherwood <david.sherwood@arm.com>
8222 * doc/sourcebuild.texi (vect_masked_store): Document.
8224 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8225 Alan Hayward <alan.hayward@arm.com>
8226 David Sherwood <david.sherwood@arm.com>
8228 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8230 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8231 Alan Hayward <alan.hayward@arm.com>
8232 David Sherwood <david.sherwood@arm.com>
8234 * doc/sourcebuild.texi (vect_variable_length): Document.
8236 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8237 Alan Hayward <alan.hayward@arm.com>
8238 David Sherwood <david.sherwood@arm.com>
8240 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8242 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8243 Alan Hayward <alan.hayward@arm.com>
8244 David Sherwood <david.sherwood@arm.com>
8246 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8248 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8249 Alan Hayward <alan.hayward@arm.com>
8250 David Sherwood <david.sherwood@arm.com>
8252 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8253 previously undocumented selectors.
8254 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8256 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8258 * doc/rtl.texi (const_vector): Say that elements can be
8259 const_wide_ints too.
8260 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8261 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8262 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8263 * optabs.c (expand_vector_broadcast): Likewise.
8265 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8266 Alan Hayward <alan.hayward@arm.com>
8267 David Sherwood <david.sherwood@arm.com>
8269 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8270 scaled index even if the unscaled address was invalid.
8271 Don't increase the complexity of using a scale in that case.
8273 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8274 Alan Hayward <alan.hayward@arm.com>
8275 David Sherwood <david.sherwood@arm.com>
8277 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8278 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8279 UNITS_PER_WORD bytes.
8280 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8281 based on REGMODE_NATURAL_SIZE of the inner mode.
8282 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8283 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8284 * expmed.c (lowpart_bit_field_p): Divide the value up into
8285 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8286 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8287 whether something is likely to occupy more than one register.
8289 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8292 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8293 profile_count::adjust_for_ipa_scaling.
8294 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8295 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8297 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8299 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8301 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8302 (split_group): Add total_orig and total_new arguments, estimate the
8303 number of statements related to the store group without store merging
8304 and with store merging.
8305 (imm_store_chain_info::output_merged_store): Adjust split_group
8306 callers, punt if estimated number of statements with store merging
8307 is not smaller than estimated number of statements without it.
8309 (handled_load): Remove has_single_use checks.
8310 (pass_store_merging::process_store): Likewise.
8312 2017-11-09 Richard Biener <rguenther@suse.de>
8314 PR tree-optimization/82902
8315 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8317 2017-11-09 Martin Liska <mliska@suse.cz>
8320 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8323 2017-11-09 Martin Liska <mliska@suse.cz>
8325 PR tree-optimization/82669
8326 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8328 2017-11-09 Martin Liska <mliska@suse.cz>
8330 PR gcov-profile/48463
8331 * coverage.c (coverage_begin_function): Output also end locus
8332 of a function and information whether the function is
8334 * gcov-dump.c (tag_function): Parse and print the information.
8335 * gcov.c (INCLUDE_MAP): Add include.
8336 (INCLUDE_SET): Likewise.
8337 (struct line_info): Move earlier in the source file because
8338 of vector<line_info> in function_info structure.
8339 (line_info::line_info): Likewise.
8340 (line_info::has_block): Likewise.
8341 (struct source_info): Add new member index.
8342 (source_info::get_functions_at_location): New function.
8343 (function_info::group_line_p): New function.
8344 (output_intermediate_line): New function.
8345 (output_intermediate_file): Use the mentioned function.
8346 (struct function_start): New.
8347 (struct function_start_pair_hash): Likewise.
8348 (process_file): Add code that identifies group functions.
8349 Assign lines either to global or function scope.
8350 (generate_results): Skip artificial functions.
8351 (find_source): Assign index for each source file.
8352 (read_graph_file): Read new flag artificial and end_line.
8353 (add_line_counts): Assign it either to global of function scope.
8354 (accumulate_line_counts): Isolate core of the function to
8355 accumulate_line_info and call it for both function and global
8357 (accumulate_line_info): New function.
8358 (output_line_beginning): Fix GNU coding style.
8359 (print_source_line): New function.
8360 (output_line_details): Likewise.
8361 (output_function_details): Likewise.
8362 (output_lines): Iterate both source (global) scope and function
8364 (struct function_line_start_cmp): New class.
8365 * doc/gcov.texi: Reflect changes in documentation.
8367 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8370 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8371 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8372 and similarly for not instead of neg.
8374 2017-11-08 Andi Kleen <ak@linux.intel.com>
8376 * config/i386/i386.opt: Add -mforce-indirect-call.
8377 * config/i386/predicates.md: Check for flag_force_indirect_call.
8378 * doc/invoke.texi: Document -mforce-indirect-call
8380 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8382 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8384 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8385 config/riscv/riscv.c (predict.h): New include.
8386 (riscv_slow_unaligned_access_p): No longer static.
8387 (riscv_block_move_straight): Add require.
8388 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8390 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8393 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8394 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8395 nonimmediate_operand predicate for operand 1 instead of
8398 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8400 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8402 * config/aarch64/constraints.md (Uml): New constraint.
8403 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8406 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8408 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8409 of two vec_duplicates into a vec_concat.
8411 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8413 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8414 Simplify vec_merge of vec_duplicate and vec_concat.
8415 * config/aarch64/constraints.md (Utq): New constraint.
8416 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8419 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8421 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8422 Simplify vec_merge of vec_duplicate and const_vector.
8423 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8425 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8426 mode iterator. Update predicate on operand 1 to
8427 handle non-const_vec constants. Delete constraints.
8428 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8430 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8432 PR tree-optimization/78821
8433 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8435 (store_operand_info::store_operand_info): Initialize it to false.
8436 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8437 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8438 store in the group, and if chain_info is non-NULL, to ignore altogether
8440 (compatible_load_p): Fail if bit_not_p does not match.
8441 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8442 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8443 (pass_store_merging::process_store): Adjust
8444 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8445 call terminate_all_aliasing_chains newly when adding a store into
8446 a chain with non-NULL chain_info.
8448 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8450 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8452 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8454 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8456 (aarch64_layout_frame): Initialise emit_frame_chain.
8457 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8458 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8460 2017-11-08 Martin Liska <mliska@suse.cz>
8462 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8463 of gimple_call_internal_p.
8465 2017-11-07 Tom de Vries <tom@codesourcery.com>
8467 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8469 2017-11-07 Tom de Vries <tom@codesourcery.com>
8471 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8474 2017-11-08 Martin Liska <mliska@suse.cz>
8477 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8479 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8481 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8483 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8486 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8487 (*cmp<mode>_ccz_1): New insn with $k alternative.
8490 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8491 RTX_COMM_COMPARE as commutative as well.
8492 (ix86_binary_operator_ok): Formatting fix.
8493 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8494 *<code><mode>3<mask_name><round_saeonly_name>,
8495 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8496 *<s>mul<mode>3_highpart<mask_name>,
8497 *vec_widen_umult_even_v16si<mask_name>,
8498 *vec_widen_umult_even_v8si<mask_name>,
8499 *vec_widen_umult_even_v4si<mask_name>,
8500 *vec_widen_smult_even_v16si<mask_name>,
8501 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8502 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8503 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8504 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8505 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8506 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8507 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8508 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8509 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8510 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8511 ix86_binary_operator_ok. Formatting fixes.
8512 (*<plusminus_insn><mode>3<mask_name><round_name>,
8513 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8516 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8518 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8519 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8520 eq and ne if TARGET_ISEL.
8521 (cmp): New code_iterator.
8522 (UNS, UNSU_, UNSIK): New code_attrs.
8523 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8524 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8526 ("eq<mode>3"): ... this.
8527 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8529 ("ne<mode>3"): ... this.
8531 2017-11-07 Julia Koval <julia.koval@intel.com>
8534 * common/config/i386/i386-common.c
8535 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8536 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8537 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8538 * config/i386/i386.opt: Ditto.
8539 * config/i386/i386.c (ix86_target_string): Ditto.
8540 (ix86_option_override_internal): Ditto.
8541 (ix86_init_mpx_builtins): Move MPX to args2.
8542 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8543 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8544 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8545 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8546 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8547 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8548 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8550 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8553 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8554 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8555 (zero-extendsidi peephole2): Remove peephole.
8557 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8560 * stor-layout.c: Include attribs.h.
8561 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8562 explicit lookup of "aligned" attribute.
8564 2017-11-07 Andrew Waterman <andrew@sifive.com>
8566 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8567 (riscv_expand_block_move): Likewise.
8568 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8570 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8571 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8572 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8573 (riscv_adjust_block_mem): Likewise.
8574 (riscv_block_move_loop): Likewise.
8575 (riscv_expand_block_move): Likewise.
8576 * config/riscv/riscv.md (movmemsi): New pattern.
8578 2017-11-07 Michael Clark <michaeljclark@mac.com>
8580 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8581 (MUSL_DYNAMIC_LINKER): Likewise.
8583 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8585 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8586 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8589 2017-11-07 Richard Biener <rguenther@suse.de>
8591 * match.pd: Fix build.
8593 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8594 Jackson Woodruff <jackson.woodruff@arm.com>
8596 PR tree-optimization/71026
8597 * match.pd: Canonicalize negate in division.
8599 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8602 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8604 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8606 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8607 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8609 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8611 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8612 non-scalar integral types.
8613 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8614 (-(A-B), -(~A)): New transformations.
8616 2017-11-07 Tom de Vries <tom@codesourcery.com>
8618 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8619 semicolon after "do {} while (0)".
8620 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8621 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8622 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8623 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8624 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8626 2017-11-07 Tom de Vries <tom@codesourcery.com>
8628 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8629 after "do {} while (0)".
8630 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8631 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8632 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8633 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8634 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8636 2017-11-07 Tom de Vries <tom@codesourcery.com>
8639 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8641 (arm_rtx_costs_internal): Add missing semicolon after
8642 HANDLE_NARROW_SHIFT_ARITH call.
8644 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8646 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8647 disable isel if it was not set explicitly.
8649 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8651 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8652 (add_type_attribute) likewise.
8654 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8656 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8657 of struct ix86_frame.
8658 (ix86_initial_elimination_offset): Likewise.
8659 (ix86_expand_split_stack_prologue): Likewise.
8661 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8663 * tree-vrp.h (enum value_range_type): Update stale comment.
8665 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8666 Alan Hayward <alan.hayward@arm.com>
8667 David Sherwood <david.sherwood@arm.com>
8669 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8670 (aarch64_expand_vec_perm_const): Take the number of units too.
8671 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8672 (aarch64_expand_vec_perm_const): Likewise.
8673 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8674 (vec_perm<mode>): Update accordingly.
8676 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8677 Alan Hayward <alan.hayward@arm.com>
8678 David Sherwood <david.sherwood@arm.com>
8680 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8681 Take the number of units too.
8682 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8683 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8684 but check for a vector mode before rather than after the call.
8685 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8686 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8687 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8688 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8689 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8690 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8691 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8692 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8693 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8694 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8695 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8696 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8697 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8698 (aarch64_sqdmull2_n<mode>): Update accordingly.
8700 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8701 Alan Hayward <alan.hayward@arm.com>
8702 David Sherwood <david.sherwood@arm.com>
8704 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8705 the number of units too.
8706 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8707 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8708 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8709 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8710 (vec_store_lanesxi<mode>): Update accordingly.
8712 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8713 Alan Hayward <alan.hayward@arm.com>
8714 David Sherwood <david.sherwood@arm.com>
8716 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8717 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8718 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8719 of units rather than the mode.
8720 * config/aarch64/iterators.md (nunits): New mode attribute.
8721 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8722 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8723 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8724 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8725 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8726 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8727 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8728 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8729 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8730 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8731 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8732 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8733 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8734 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8735 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8736 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8737 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8738 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8739 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8740 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8741 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8742 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8743 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8744 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8745 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8746 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8747 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8748 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8749 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8750 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8751 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8752 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8753 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8754 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8755 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8756 (aarch64_simd_vec_setv2di): Likewise.
8758 2017-11-06 Carl Love <cel@us.ibm.com>
8760 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8762 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8763 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8765 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8767 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8768 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8769 Add power 8 macro expansions.
8770 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8771 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8772 power 8 instructions. (VSX_XXBR): Add iterator.
8774 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8776 * config/arm/arm.md (predicable_short_it): Change default to "no",
8777 improve documentation, remove uses that are identical to the default.
8778 (enabled_for_depr_it): Rename to enabled_for_short_it.
8779 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8780 Remove default uses.
8781 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8782 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8783 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8784 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8786 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8789 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8790 float128 helper macros, which are no longer used after deleting
8791 the old 'q' built-in functions, and moving the round to odd
8792 built-in functions to being special built-in functions.
8793 (BU_FLOAT128_2): Likewise.
8794 (BU_FLOAT128_1_HW): Likewise.
8795 (BU_FLOAT128_2_HW): Likewise.
8796 (BU_FLOAT128_3_HW): Likewise.
8797 (FABSQ): Delete old 'q' built-in functions.
8798 (COPYSIGNQ): Likewise.
8799 (SQRTF128_ODD): Move round to odd built-in functions to be
8800 special built-in functions, so that we can handle
8801 -mabi=ieeelongdouble.
8802 (TRUNCF128_ODD): Likewise.
8803 (ADDF128_ODD): Likewise.
8804 (SUBF128_ODD): Likewise.
8805 (MULF128_ODD): Likewise.
8806 (DIVF128_ODD): Likewise.
8807 (FMAF128_ODD): Likewise.
8808 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8809 built-in names to 'f128'.
8810 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8811 old 'q' built-in functions, as the machine independent code for
8812 'f128' built-in functions handles this.
8813 (rs6000_expand_builtin): Add expansion for float128 round to odd
8814 functions, keying off on -mabi=ieeelongdouble of whether to use
8815 the KFmode or TFmode variant.
8816 (rs6000_init_builtins): Initialize the _Float128 round to odd
8818 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8819 _Float128 'q' built-in functions are now mapped into the new
8820 'f128' built-in functions.
8822 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8824 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8825 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8827 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8830 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8831 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8832 * toplev.c: Include "ipa-fnsummary.h".
8833 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8835 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8837 PR tree-optimization/82838
8838 * gimple-ssa-store-merging.c
8839 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8840 on a separate gimple_seq which is then appended to seq.
8842 2017-11-06 Jeff Law <law@redhat.com>
8845 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8846 (get_probe_interval): New functions.
8847 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8848 (ix86_adjust_stack_and_probe): Likewise.
8849 (output_adjust_stack_and_probe): Likewise.
8850 (ix86_emit_probe_stack_range): Likewise.
8851 (ix86_expand_prologue): Likewise.
8853 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8855 PR tree-optimization/82816
8856 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8857 if the modes of the two types are the same.
8858 (convert_plusminus_to_widen): Likewise.
8860 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8862 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8864 (usadv16qi): New define_expand.
8865 (usadv8hi): New define_expand.
8867 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8870 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8872 (inline_transform): Likewise.
8873 * predict.c (counts_to_freqs): Remove useless conditional.
8874 * profile-count.h (profile_count::apply_scale): Move sanity check.
8875 * tree-inline.c (copy_bb): Always scale.
8876 (copy_cfg_body): Likewise.
8878 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8881 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8883 (*cmp_ite0): Add enabled_for_depr_it attribute.
8884 (*cmp_ite1): Likewise.
8886 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8888 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8891 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8893 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8894 return the value by pointer.
8895 (extract_range_from_multiplicative_op_1): Update accordingly.
8896 Return as soon as an operation fails.
8898 2017-11-05 Tom de Vries <tom@codesourcery.com>
8901 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8902 (DEF_SANITIZER_BUILTIN): ... here.
8903 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8904 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8906 2017-11-05 Tom de Vries <tom@codesourcery.com>
8909 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8911 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8912 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8913 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8915 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8916 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8917 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8919 2017-11-05 Tom de Vries <tom@codesourcery.com>
8922 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8925 2017-11-04 Michael Clark <michaeljclark@mac.com>
8927 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8928 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8930 (*addsi3_extended): Likewise.
8931 (*addsi3_extended2): Likewise.
8932 (<optab>si3): Likewise.
8933 (<optab>di3): Likewise.
8934 (<optab><mode>3): Likewise.
8935 (<*optabe>si3_internal): Likewise.
8936 (zero_extendqi<SUPERQI:mode>2): Likewise.
8937 (*add<mode>hi3): Likewise.
8938 (*xor<mode>hi3): Likewise.
8939 (<optab>di3): Likewise.
8940 (*<optab>si3_extend): Likewise.
8941 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8942 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8943 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8945 2017-11-04 Andrew Waterman <andrew@sifive.com>
8947 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8948 TARGET_STRICT_ALIGN based upon -mtune argument.
8950 2017-11-04 Andrew Waterman <andrew@sifive.com>
8952 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8954 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8956 * config/i386/i386.c (choose_basereg): Use optional scratch
8957 register and add assertion.
8958 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8959 needed, and don't allocate stack.
8960 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8961 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8962 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8964 2017-11-03 Jeff Law <law@redhat.com>
8966 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8967 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8968 to probe at the start of a noreturn function.
8970 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8972 PR tree-optimization/78821
8973 * gimple-ssa-store-merging.c: Update the file comment.
8974 (MAX_STORE_ALIAS_CHECKS): Define.
8975 (struct store_operand_info): New type.
8976 (store_operand_info::store_operand_info): New constructor.
8977 (struct store_immediate_info): Add rhs_code and ops data members.
8978 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8979 and op1r arguments to the ctor, initialize corresponding data members.
8980 (struct merged_store_group): Add load_align_base and load_align
8982 (merged_store_group::merged_store_group): Initialize them.
8983 (merged_store_group::do_merge): Update them.
8984 (merged_store_group::apply_stores): Pick the constant for
8985 encode_tree_to_bitpos from one of the two operands, or skip
8986 encode_tree_to_bitpos if neither operand is a constant.
8987 (class pass_store_merging): Add process_store method decl. Remove
8988 bool argument from terminate_all_aliasing_chains method decl.
8989 (pass_store_merging::terminate_all_aliasing_chains): Remove
8990 var_offset_p argument and corresponding handling.
8991 (stmts_may_clobber_ref_p): New function.
8992 (compatible_load_p): New function.
8993 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8994 if there is overlap and rhs_code is not INTEGER_CST. For
8995 non-overlapping stores terminate group if rhs is not mergeable.
8996 (get_alias_type_for_stmts): Change first argument from
8997 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8998 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8999 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9001 (get_location_for_stmts): Change first argument from
9002 auto_vec<gimple *> & to vec<gimple *> &.
9003 (struct split_store): Remove orig_stmts data member, add orig_stores.
9004 (split_store::split_store): Create orig_stores rather than orig_stmts.
9005 (find_constituent_stmts): Renamed to ...
9006 (find_constituent_stores): ... this. Change second argument from
9007 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9008 to info structures rather than the statements.
9009 (split_group): Rename ALLOW_UNALIGNED argument to
9010 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9011 it. Adjust find_constituent_stores caller.
9012 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9013 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9014 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9015 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9016 (mem_valid_for_store_merging): New function.
9017 (handled_load): New function.
9018 (pass_store_merging::process_store): New method.
9019 (pass_store_merging::execute): Use process_store method. Adjust
9020 terminate_all_aliasing_chains caller.
9022 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9024 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9025 Return true for more constants, symbols and label references.
9026 (aarch64_valid_floating_const): Remove unused function.
9028 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9031 * config/aarch64/aarch64.c (aarch64_layout_frame):
9032 Undo forcing of LR at bottom of frame.
9034 2017-11-03 Jeff Law <law@redhat.com>
9037 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9038 for int_registers_saved.
9040 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9041 extracted from tree-ssa-dom.c.
9042 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9043 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9044 (record_equivalences_from_incoming_edge): Add additional argument
9045 to single_pred_edge_ignoring_loop_edges call.
9046 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9047 (uncprop_dom_walker::before_dom_children): Add additional argument
9048 to single_pred_edge_ignoring_loop_edges call.
9049 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9050 single_pred_edge_ignoring_loop_edges rather than open coding.
9051 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9053 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9055 * match.pd (-(-A)): Rewrite.
9057 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9059 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9060 (rs6000_emit_int_cmove): New declaration.
9061 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9062 (rs6000_emit_sISEL): Delete.
9063 (rs6000_emit_int_cmove): Make non-static.
9064 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9065 instead of rs6000_emit_sISEL.
9067 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9069 * asan.c (create_cond_insert_point): Maintain profile.
9070 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9072 * basic-block.h (struct basic_block_def): Remove frequency.
9073 (EDGE_FREQUENCY): Use to_frequency
9074 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9076 (find_traces): Update to use to_frequency.
9077 (find_traces_1_round): Likewise; use only IPA counts.
9078 (bb_to_key): Likewise.
9079 (connect_traces): Use IPA counts only.
9080 (copy_bb_p): Update to use to_frequency.
9081 (fix_up_crossing_landing_pad): Likewise.
9082 (sanitize_hot_paths): Likewise.
9083 * bt-load.c (basic_block_freq): Likewise.
9084 * cfg.c (init_flow): Set count_max to uninitialized.
9085 (check_bb_profile): Remove frequencies; check counts.
9086 (dump_bb_info): Do not dump frequencies.
9087 (update_bb_profile_for_threading): Update counts only.
9088 (scale_bbs_frequencies_int): Likewise.
9089 (MAX_SAFE_MULTIPLIER): Remove.
9090 (scale_bbs_frequencies_gcov_type): Update counts only.
9091 (scale_bbs_frequencies_profile_count): Update counts only.
9092 (scale_bbs_frequencies): Update counts only.
9093 * cfg.h (struct control_flow_graph): Add count-max.
9094 (update_bb_profile_for_threading): Update prototype.
9095 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9096 (find_many_sub_basic_blocks): Likewise.
9097 * cfgcleanup.c (try_forward_edges): Likewise.
9098 (try_crossjump_to_edge): Likewise.
9099 * cfgexpand.c (expand_gimple_cond): Likewise.
9100 (expand_gimple_tailcall): Likewise.
9101 (construct_init_block): Likewise.
9102 (construct_exit_block): Likewise.
9103 * cfghooks.c (verify_flow_info): Check consistency of counts.
9104 (dump_bb_for_graph): Do not dump frequencies.
9105 (split_block_1): Do not update frequencies.
9106 (split_edge): Do not update frequencies.
9107 (make_forwarder_block): Do not update frequencies.
9108 (duplicate_block): Do not update frequencies.
9109 (account_profile_record): Do not update frequencies.
9110 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9111 for global heuristics.
9112 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9113 (expected_loop_iterations_unbounded): Use counts only.
9114 * cfgloopmanip.c (scale_loop_profile): Simplify.
9115 (create_empty_loop_on_edge): Simplify
9117 (duplicate_loop_to_header_edge): Simplify
9118 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9119 (update_br_prob_note): Take care of removing note when profile
9121 (relink_block_chain): Do not dump frequency.
9122 (rtl_account_profile_record): Use to_frequency.
9123 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9124 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9125 (cgraph_update_edges_for_call_stmt_node): Likewise.
9126 (cgraph_edge::verify_count_and_frequency): Update.
9127 (cgraph_node::verify_node): Temporarily disable frequency verification.
9128 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9129 to_cgraph_frequency.
9130 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9131 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9133 (cgraph_node::expand_thunk): Update profile.
9134 * except.c (dw2_build_landing_pads): Do not update frequency.
9135 * final.c (compute_alignments): Use to_frequency.
9136 (dump_basic_block_info): Do not dump frequency.
9137 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9138 (dump_gimple_bb_header): Do not dump frequency.
9139 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9141 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9142 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9143 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9144 (init_before_recovery): Do not update frequency.
9145 (sched_create_recovery_edges): Do not update frequency.
9146 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9147 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9148 (ipa_cp_c_finalize): Set max_count to uninitialized.
9149 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9150 (param_change_prob): Use counts.
9151 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9153 * ipa-split.c (consider_split): Use to_frequency.
9154 (split_function): Use to_frequency.
9155 * ira-build.c (loop_compare_func): Likewise.
9156 (mark_loops_for_removal): Likewise.
9157 (mark_all_loops_for_removal): Likewise.
9158 * loop-doloop.c (doloop_modify): Do not update frequency.
9159 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9161 * lto-streamer-in.c (input_function): Update count_max.
9162 * omp-expand.c (expand_omp_taskreg): Update count_max.
9163 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9164 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9165 (maybe_hot_count_p): Use ipa counts only.
9166 (maybe_hot_bb_p): Simplify.
9167 (maybe_hot_edge_p): Simplify.
9168 (probably_never_executed): Do not take frequency argument.
9169 (probably_never_executed_bb_p): Do not pass frequency.
9170 (probably_never_executed_edge_p): Likewise.
9171 (combine_predictions_for_bb): Check that profile is nonzero.
9172 (propagate_freq): Do not set frequency.
9173 (drop_profile): Simplify.
9174 (counts_to_freqs): Simplify.
9175 (expensive_function_p): Use to_frequency.
9176 (propagate_unlikely_bbs_forward): Simplify.
9177 (determine_unlikely_bbs): Simplify.
9178 (estimate_bb_frequencies): Add hack to silence graphite issues.
9179 (compute_function_frequency): Use ipa counts.
9180 (pass_profile::execute): Update.
9181 (rebuild_frequencies): Use counts only.
9182 (force_edge_cold): Use counts only.
9183 * profile-count.c (profile_count::dump): Dump new count types.
9184 (profile_count::differs_from_p): Check compatiblity.
9185 (profile_count::to_frequency): New function.
9186 (profile_count::to_cgraph_frequency): New function.
9187 * profile-count.h (struct function): Declare.
9188 (enum profile_quality): Add profile_guessed_local and
9189 profile_guessed_global0.
9190 (class profile_proability): Decrease number of bits to 29;
9191 update from_reg_br_prob_note and to_reg_br_prob_note.
9192 (class profile_count: Update comment; decrease number of bits
9193 to 61. Check compatibility.
9194 (profile_count::compatible_p): New private member function.
9195 (profile_count::ipa_p): New member function.
9196 (profile_count::operator<): Handle global zero correctly.
9197 (profile_count::operator>): Handle global zero correctly.
9198 (profile_count::operator<=): Handle global zero correctly.
9199 (profile_count::operator>=): Handle global zero correctly.
9200 (profile_count::nonzero_p): New member function.
9201 (profile_count::force_nonzero): New member function.
9202 (profile_count::max): New member function.
9203 (profile_count::apply_scale): Handle IPA scalling.
9204 (profile_count::guessed_local): New member function.
9205 (profile_count::global0): New member function.
9206 (profile_count::ipa): New member function.
9207 (profile_count::to_frequency): Declare.
9208 (profile_count::to_cgraph_frequency): Declare.
9209 * profile.c (OVERLAP_BASE): Delete.
9210 (compute_frequency_overlap): Delete.
9211 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9212 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9213 * sched-ebb.c (rank): Use counts only.
9214 * shrink-wrap.c (handle_simple_exit): Use counts only.
9215 (try_shrink_wrapping): Use counts only.
9216 (place_prologue_for_one_component): Use counts only.
9217 * tracer.c (find_best_predecessor): Use to_frequency.
9218 (find_trace): Use to_frequency.
9219 (tail_duplicate): Use to_frequency.
9220 * trans-mem.c (expand_transaction): Do not update frequency.
9221 * tree-call-cdce.c: Do not update frequency.
9222 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9223 (gimple_merge_blocks): Likewise.
9224 (gimple_split_edge): Likewise.
9225 (gimple_duplicate_sese_region): Likewise.
9226 (gimple_duplicate_sese_tail): Likewise.
9227 (move_sese_region_to_fn): Likewise.
9228 (gimple_account_profile_record): Likewise.
9229 (insert_cond_bb): Likewise.
9230 * tree-complex.c (expand_complex_div_wide): Likewise.
9231 * tree-eh.c (lower_resx): Update profile.
9232 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9234 (initialize_cfun): Do not initialize frequencies
9235 (freqs_to_counts): Delete.
9236 (copy_cfg_body): Ignore count parameter.
9237 (copy_body): Update.
9238 (expand_call_inline): Update count_max.
9239 (optimize_inline_calls): Update count_max.
9240 (tree_function_versioning): Update count_max.
9241 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9242 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9244 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9245 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9246 (try_peel_loop): Likewise.
9247 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9249 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9250 (tree_transform_and_unroll_loop): Do not use frequencies
9251 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9252 Use reliable prediction only.
9253 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9254 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9255 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9256 probability scaling.
9257 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9258 not update frequency
9259 (any_remaining_duplicated_blocks): Likewise.
9260 (update_profile): Likewise.
9261 (estimated_freqs_path): Delete.
9262 (freqs_to_counts_path): Delete.
9263 (clear_counts_path): Delete.
9264 (ssa_fix_duplicate_block_edges): Likewise.
9265 (duplicate_thread_path): Likewise.
9266 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9267 * tree-tailcall.c (decrease_profile): Do not update frequency.
9268 (eliminate_tail_call): Likewise.
9269 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9270 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9271 (optimize_mask_stores): Likewise.
9272 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9273 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9274 (ubsan_expand_ptr_ifn): Update profile.
9275 * value-prof.c (gimple_ic): Simplify.
9276 * value-prof.h (gimple_ic): Update prototype.
9277 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9278 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9280 (want_inline_self_recursive_call_p): Likewise.
9281 (resolve_noninline_speculation): Only cummulate defined counts.
9282 (inline_small_functions): Use nonzero_p.
9283 (ipa_inline): Do not access freed node.
9285 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9287 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9288 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9290 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9292 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9293 non-legitimate address.
9295 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9297 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9298 (*lt0_<mode>di, *lt0_<mode>si): New.
9300 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9302 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9303 TARGET_PAIRED_FLOAT.
9305 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9306 Jim Wilson <jim.wilson@linaro.org>
9308 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9309 * config/aarch64/aarch64-tune.md: Regenerated.
9310 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9311 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9313 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9315 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9316 (arc_expand_prologue): Restore blink for millicode.
9317 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9319 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9322 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9323 gen_vec_duplicate after forcing the scalar into a register.
9325 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9327 * combine (try_combine): Print the insns input to try_combine to the
9330 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9333 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9334 Remove second argument from aarch64_process_target_attr call.
9335 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9337 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9339 (aarch64_handle_attr_arch): Remove second argument.
9340 (aarch64_handle_attr_cpu): Ditto.
9341 (aarch64_handle_attr_tune): Ditto.
9342 (aarch64_handle_attr_isa_flags): Ditto.
9343 (aarch64_process_one_target_attr): Ditto.
9344 (aarch64_process_target_attr): Ditto.
9345 (aarch64_option_valid_attribute_p): Remove second argument.
9346 on aarch64_process_target_attr call.
9348 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9350 * diagnostic.c: Include "selftest-diagnostic.h".
9351 (selftest::assert_location_text): New function.
9352 (selftest::test_diagnostic_get_location_text): New function.
9353 (selftest::diagnostic_c_tests): Call it.
9355 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9357 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9358 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9359 (class selftest::test_diagnostic_context): Move to...
9360 * selftest-diagnostic.c: New file.
9361 * selftest-diagnostic.h: New file.
9363 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9365 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9366 offset range for FT32B.
9367 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9368 * config/ft32/ft32.md: Add TARGET_NOPM.
9369 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9370 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9372 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9374 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9376 2017-11-02 Jeff Law <law@redhat.com>
9378 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9379 virtual keyword on FINAL OVERRIDE members.
9381 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9382 virtuals together. Add virtual destructor.
9383 (substitute_and_fold_engine): Similarly.
9385 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9387 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9389 2017-11-02 Richard Biener <rguenther@suse.de>
9391 PR tree-optimization/82795
9392 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9394 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9396 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9398 (gcc_GAS_CHECK_FEATURE): Remove.
9399 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9400 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9401 * configure: Regenerate.
9403 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9405 * config/arc/arc.c (hwloop_optimize): Account for empty
9408 2017-11-02 Richard Biener <rguenther@suse.de>
9411 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9412 Truncate ARRAY_REF index and element size.
9414 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9416 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9418 2017-11-01 Jeff Law <law@redhat.com>
9420 * tree-ssa-ccp.c (ccp_folder): New class derived from
9421 substitute_and_fold_engine.
9422 (ccp_folder::get_value): New member function.
9423 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9424 (ccp_fold_stmt): Remove prototype.
9425 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9426 * tree-ssa-copy.c (copy_folder): New class derived from
9427 substitute_and_fold_engine.
9428 (copy_folder::get_value): Renamed from get_value.
9429 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9430 * tree-vrp.c (vrp_folder): New class derived from
9431 substitute_and_fold_engine.
9432 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9433 (vrp_folder::get_value): New member function.
9434 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9435 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9436 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9437 provide a class interface to folder/substitute routines.
9438 (ssa_prop_fold_stmt_fn): Remove typedef.
9439 (ssa_prop_get_value_fn): Likewise.
9440 (subsitute_and_fold): Remove prototype.
9441 (replace_uses_in): Likewise.
9442 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9443 Renamed from replace_uses_in. Call the virtual member function
9444 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9445 (substitute_and_fold_dom_walker): Remove initialization of
9446 data member entries for calbacks. Add substitute_and_fold_engine
9447 member and initialize it.
9448 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9449 member functions for get_value, replace_phi_args_in c
9450 replace_uses_in, and fold_stmt calls.
9451 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9452 substitute_and_fold. Remove assert. Update ctor call.
9454 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9455 (ssa_prop_visit_phi_fn): Likewise.
9456 (class ssa_propagation_engine): New class to provide an interface
9458 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9460 (ssa_prop_visit_phi): Likewise.
9461 (ssa_propagation_engine::simulate_stmt): Moved into class.
9462 Call visit_phi/visit_stmt from the class rather than via
9463 file scoped static variables.
9464 (ssa_propagation_engine::simulate_block): Moved into class.
9465 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9466 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9467 set file scoped statics for the visit_stmt/visit_phi callbacks.
9468 * tree-complex.c (complex_propagate): New class derived from
9469 ssa_propagation_engine.
9470 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9471 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9472 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9474 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9475 ssa_propagation_engine.
9476 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9477 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9478 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9479 * tree-ssa-copy.c (copy_prop): New class derived from
9480 ssa_propagation_engine.
9481 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9482 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9483 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9484 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9485 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9486 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9487 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9489 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9491 PR rtl-optimization/82778
9492 PR rtl-optimization/82597
9493 * compare-elim.c (struct comparison): Add in_a_setter field.
9494 (find_comparison_dom_walker::before_dom_children): Remove killed
9495 bitmap and df_simulate_find_defs call, instead walk the defs.
9496 Compute last_setter and initialize in_a_setter. Merge definitions
9497 with first initialization for a few variables.
9498 (try_validate_parallel): Use insn_invalid_p instead of
9499 recog_memoized. Return insn rather than just the pattern.
9500 (try_merge_compare): Fix up comment. Don't uselessly test if
9501 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9503 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9506 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9507 Alan Hayward <alan.hayward@arm.com>
9508 David Sherwood <david.sherwood@arm.com>
9510 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9511 aarch64_hard_regno_nregs to get the number of registers
9514 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9515 Alan Hayward <alan.hayward@arm.com>
9516 David Sherwood <david.sherwood@arm.com>
9518 * config/aarch64/constraints.md (Upl): Rename to...
9520 * config/aarch64/aarch64.md
9521 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9524 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9525 Alan Hayward <alan.hayward@arm.com>
9526 David Sherwood <david.sherwood@arm.com>
9528 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9529 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9532 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9533 Alan Hayward <alan.hayward@arm.com>
9534 David Sherwood <david.sherwood@arm.com>
9536 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9537 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9538 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9540 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9541 of a single element.
9542 * config/aarch64/iterators.md: Add a comment above the permute
9543 unspecs to say that they are generated directly by
9544 aarch64_expand_vec_perm_const.
9545 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9547 2017-11-01 Nathan Sidwell <nathan@acm.org>
9549 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9551 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9553 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9554 and medany code models, and describe what they do.
9556 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9558 Revert accidental duplicate:
9560 * combine.c (can_change_dest_mode): Reject changes in
9561 REGMODE_NATURAL_SIZE.
9563 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9565 PR rtl-optimization/64682
9566 PR rtl-optimization/69567
9567 PR rtl-optimization/69737
9568 PR rtl-optimization/82683
9569 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9570 register mentioned in the note, drop the note, unless it came from I3,
9571 in which case it should go to I3 again.
9573 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9575 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9576 and return false if not.
9577 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9579 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9581 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9582 range is known to be empty.
9584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9585 Alan Hayward <alan.hayward@arm.com>
9586 David Sherwood <david.sherwood@arm.com>
9588 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9589 and CONST_VECTOR_NUNITS instead of computing the number of units from
9590 the byte sizes of the vector and element.
9591 (simplify_binary_operation_1): Likewise.
9592 (simplify_const_binary_operation): Likewise.
9593 (simplify_ternary_operation): Likewise.
9595 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9596 Alan Hayward <alan.hayward@arm.com>
9597 David Sherwood <david.sherwood@arm.com>
9599 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9600 (int_mem_offset): ...this new function.
9601 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9602 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9603 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
9611 (compute_costs, find_decomposable_subregs, decompose_register)
9612 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9613 (resolve_clobber, dump_choices): Use it.
9615 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9616 Alan Hayward <alan.hayward@arm.com>
9617 David Sherwood <david.sherwood@arm.com>
9619 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9621 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9622 Alan Hayward <alan.hayward@arm.com>
9623 David Sherwood <david.sherwood@arm.com>
9625 * alias.c (find_base_value, find_base_term): Only process integer
9626 truncations. Check the precision rather than the size.
9628 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9629 Alan Hayward <alan.hayward@arm.com>
9630 David Sherwood <david.sherwood@arm.com>
9632 * machmode.h (is_narrower_int_mode): New function
9633 * optabs.c (expand_float, expand_fix): Use it.
9634 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9636 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9637 Alan Hayward <alan.hayward@arm.com>
9638 David Sherwood <david.sherwood@arm.com>
9640 * rtl.h (narrower_subreg_mode): New function.
9641 * ira-color.c (update_costs_from_allocno): Use it.
9643 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9644 Alan Hayward <alan.hayward@arm.com>
9645 David Sherwood <david.sherwood@arm.com>
9647 * optabs-query.h (convert_optab_p): New function, split out from...
9648 (convert_optab_handler): ...here.
9649 (widening_optab_handler): Delete.
9650 (find_widening_optab_handler): Remove permit_non_widening parameter.
9651 (find_widening_optab_handler_and_mode): Likewise. Provide an
9652 override that operates on mode class wrappers.
9653 * optabs-query.c (widening_optab_handler): Delete.
9654 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9655 parameter. Assert that the two modes are the same class and that
9656 the "from" mode is narrower than the "to" mode. Use
9657 convert_optab_handler instead of widening_optab_handler.
9658 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9659 instead of widening_optab_handler.
9660 * expr.c (expand_expr_real_2): Update calls to
9661 find_widening_optab_handler.
9662 * optabs.c (expand_widen_pattern_expr): Likewise.
9663 (expand_binop_directly): Take the insn_code as a parameter.
9664 (expand_binop): Only call find_widening_optab_handler for
9665 conversion optabs; use optab_handler otherwise. Update calls
9666 to find_widening_optab_handler and expand_binop_directly.
9667 Use convert_optab_handler instead of widening_optab_handler.
9668 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9669 find_widening_optab_handler and use scalar_mode rather than
9671 (convert_plusminus_to_widen): Likewise.
9673 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9674 Alan Hayward <alan.hayward@arm.com>
9675 David Sherwood <david.sherwood@arm.com>
9677 * machmode.h (fixed_size_mode): New class.
9678 * rtl.h (get_pool_mode): Return fixed_size_mode.
9679 * gengtype.c (main): Add fixed_size_mode.
9680 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9681 (get_raw_arg_mode): Likewise.
9682 * doc/tm.texi: Regenerate.
9683 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9684 * targhooks.c (default_get_reg_raw_mode): Likewise.
9685 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9686 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9687 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9688 (msp430_get_raw_result_mode): Likewise.
9689 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9690 * dbxout.c (dbxout_parms): Require fixed-size modes.
9691 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9692 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9693 * omp-low.c (lower_oacc_reductions): Likewise.
9694 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9695 (simplify_subreg): Update accordingly.
9696 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9697 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9698 that aren't fixed-size.
9699 (get_pool_mode): Return a fixed_size_mode.
9700 (output_constant_pool_2): Take a fixed_size_mode.
9702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9703 Alan Hayward <alan.hayward@arm.com>
9704 David Sherwood <david.sherwood@arm.com>
9706 * doc/rtl.texi (vec_series): Document.
9707 (const): Say that the operand can be a vec_series.
9708 * rtl.def (VEC_SERIES): New rtx code.
9709 * rtl.h (const_vec_series_p_1): Declare.
9710 (const_vec_series_p): New function.
9711 * emit-rtl.h (gen_const_vec_series): Declare.
9712 (gen_vec_series): Likewise.
9713 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9714 (gen_vec_series): Likewise.
9715 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9716 * simplify-rtx.c (simplify_unary_operation): Handle negations
9718 (simplify_binary_operation_series): New function.
9719 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9720 (test_vector_ops_series): New function.
9721 (test_vector_ops): Call it.
9722 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9723 gen_const_vec_series.
9724 (altivec_lvsr): Likewise.
9725 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9727 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9728 Alan Hayward <alan.hayward@arm.com>
9729 David Sherwood <david.sherwood@arm.com>
9731 * doc/rtl.texi (const): Update description of address constants.
9732 Say that vector constants are allowed too.
9733 * common.md (E, F): Use CONSTANT_P instead of checking for
9735 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9736 checking for CONST_VECTOR.
9737 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9739 * expr.c (expand_expr_real_2): Check for vector modes instead
9740 of checking for CONST_VECTOR.
9741 * rtl.h (const_vec_p): New function.
9742 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9743 (unwrap_const_vec_duplicate): Handle them here too.
9745 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9746 David Malcolm <dmalcolm@redhat.com>
9747 Alan Hayward <alan.hayward@arm.com>
9748 David Sherwood <david.sherwood@arm.com>
9750 * rtl.h (vec_duplicate_p): New function.
9751 * selftest-rtl.c (assert_rtx_eq_at): New function.
9752 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9753 (assert_rtx_eq_at): Declare.
9754 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9755 * selftest-run-tests.c (selftest::run_tests): Call it.
9756 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9757 (simplify_unary_operation_1): Recursively handle vector duplicates.
9758 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9760 (simplify_subreg): Handle subregs of vector duplicates.
9761 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9762 (selftest::simplify_rtx_c_tests): New functions.
9764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9765 Alan Hayward <alan.hayward@arm.com>
9766 David Sherwood <david.sherwood@arm.com>
9768 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9769 (gen_vec_duplicate): Likewise.
9770 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9772 (gen_const_vector): ...here.
9773 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9774 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9775 whose elements are all equal.
9776 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9777 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9778 (simplify_relational_operation): Likewise.
9779 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9781 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9782 (aarch64_expand_vector_init): Likewise.
9783 * config/arm/arm.c (neon_vdup_constant): Likewise.
9784 (neon_expand_vector_init): Likewise.
9785 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9786 (arm_block_set_unaligned_vect): Likewise.
9787 (arm_block_set_aligned_vect): Likewise.
9788 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9789 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9790 (expand_vec_perm_even_odd_pack): Likewise.
9791 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9792 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9793 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9794 gen_const_vec_duplicate.
9795 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9796 * config/mips/mips.c (mips_gen_const_int_vector): Use
9797 gen_const_vec_duplicate.
9798 (mips_expand_vector_init): Use CONST0_RTX.
9799 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9800 (define_split): Use gen_const_vec_duplicate.
9801 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9802 (define_split): Use gen_const_vec_duplicate.
9803 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9804 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9805 * config/spu/spu.c (spu_const): Likewise.
9807 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9808 Alan Hayward <alan.hayward@arm.com>
9809 David Sherwood <david.sherwood@arm.com>
9811 * combine.c (can_change_dest_mode): Reject changes in
9812 REGMODE_NATURAL_SIZE.
9814 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9816 * configure.ac (--enable-libssp): New.
9817 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9818 trying to determine target-specific default. Adjust indentation.
9819 * configure: Regenerated.
9820 * doc/install.texi (Configuration): Expand --disable-libssp
9823 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9825 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9828 2017-10-31 Martin Jambor <mjambor@suse.cz>
9831 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9833 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9835 * auto-profile.c (autofdo_source_profile::read): Use
9836 UNKNOWN_LOCATION rather than 0.
9837 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9838 (warning_at): ...this overload.
9839 (warning_at_rich_loc_n): Rename to...
9840 (warning_n): ...this overload.
9841 (error_at_rich_loc): Rename to...
9842 (error_at): ...this overload.
9843 (pedwarn_at_rich_loc): Rename to...
9844 (pedwarn): ...this overload.
9845 (permerror_at_rich_loc): Rename to...
9846 (permerror): ...this overload.
9847 (inform_at_rich_loc): Rename to...
9848 (inform): ...this overload.
9849 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9850 (diagnostic_n_impl_richloc): Rename to...
9851 (diagnostic_n_impl): ...this rich_location *-based decl.
9852 (inform_at_rich_loc): Rename to...
9853 (inform): ...this, and add an assertion.
9854 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9855 (warning_at_rich_loc): Rename to...
9856 (warning_at): ...this, and add an assertion.
9857 (warning_at_rich_loc_n): Rename to...
9858 (warning_n): ...this, and add an assertion.
9859 (warning_n): Update location_t-based implementation for removal of
9860 location_t-based diagnostic_n_impl.
9861 (pedwarn_at_rich_loc): Rename to...
9862 (pedwarn): ...this, and add an assertion.
9863 (permerror_at_rich_loc): Rename to...
9864 (permerror): ...this, and add an assertion.
9865 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9866 (error_at_rich_loc): Rename to...
9867 (error_at): ...this, and add an assertion.
9868 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9869 (driver::do_spec_on_infiles): Likewise.
9870 * substring-locations.c (format_warning_va): Update for renaming
9871 of inform_at_rich_loc.
9873 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9875 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9876 _Float<N>X built-in functions so that the variant without the
9877 "__builtin_" prefix is only enabled for the GNU C and Objective C
9878 languages when they are in non-strict ANSI/ISO mode.
9879 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9880 * target.def (floatn_builtin_p): Add a target hook to control
9881 whether _Float<N> and _Float<N>X built-in functions without the
9882 "__builtin_" prefix are enabled, and return true for C and
9883 Objective C in the default hook. Include langhooks.h in
9885 * targhooks.h (default_floatn_builtin_p): Likewise.
9886 * targhooks.c (default_floatn_builtin_p): Likewise.
9887 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9888 floatn_builtin_p target hook.
9889 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9891 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9892 Eric Botcazou <ebotcazou@adacore.com>
9894 PR rtl-optimization/81803
9895 * lra-constraints.c (curr_insn_transform): Also reload the whole
9896 register for a strict subreg no wider than a word if this is for
9897 a WORD_REGISTER_OPERATIONS target.
9899 2017-10-31 Jason Merrill <jason@redhat.com>
9901 * gdbinit.in: Skip over inlines from timevar.h.
9903 2017-10-31 Martin Liska <mliska@suse.cz>
9905 * doc/gcov.texi: Document new option.
9906 * gcov.c (print_usage): Likewise print it.
9907 (process_args): Support the argument.
9908 (format_count): New function.
9909 (format_gcov): Use the function.
9911 2017-10-31 Martin Liska <mliska@suse.cz>
9913 * gcov.c (struct name_map): do not use typedef.
9914 Define operator== and operator<.
9915 (name_search): Remove.
9916 (name_sort): Remove.
9917 (main): Do not allocate names.
9918 (process_file): Add vertical space.
9919 (generate_results): Use std::find.
9920 (release_structures): Do not release memory.
9921 (find_source): Use std::find.
9923 2017-10-31 Martin Liska <mliska@suse.cz>
9925 * gcov.c (struct line_info): Remove it's typedef.
9926 (line_info::line_info): Add proper ctor.
9927 (line_info::has_block): Do not use a typedef.
9928 (struct source_info): Do not use typedef.
9929 (circuit): Likewise.
9930 (get_cycles_count): Likewise.
9931 (output_intermediate_file): Iterate via vector iterator.
9932 (add_line_counts): Use std::vector methods.
9933 (accumulate_line_counts): Likewise.
9934 (output_lines): Likewise.
9936 2017-10-31 Martin Liska <mliska@suse.cz>
9938 * gcov.c (struct source_info): Remove typedef.
9939 (source_info::source_info): Add proper ctor.
9940 (accumulate_line_counts): Use struct, not it's typedef.
9941 (output_gcov_file): Likewise.
9942 (output_lines): Likewise.
9943 (main): Do not allocate an array.
9944 (output_intermediate_file): Use size of vector container.
9945 (process_file): Resize the vector.
9946 (generate_results): Do not preallocate, use newly added vector
9948 (release_structures): Do not release sources.
9949 (find_source): Use vector methods.
9950 (add_line_counts): Do not use typedef.
9952 2017-10-31 Martin Liska <mliska@suse.cz>
9954 * doc/gcov.texi: Document that.
9955 * gcov.c (add_line_counts): Mark lines with a non-executed
9957 (output_line_beginning): Handle such lines.
9958 (output_lines): Pass new argument.
9959 (output_intermediate_file): Print it in intermediate format.
9961 2017-10-31 Martin Liska <mliska@suse.cz>
9963 * color-macros.h: New file.
9964 * diagnostic-color.c: Factor out color related to macros to
9966 * doc/gcov.texi: Document -k option.
9967 * gcov.c (INCLUDE_STRING): Include string.h.
9968 (print_usage): Add -k option.
9969 (process_args): Parse it.
9970 (pad_count_string): New function.
9971 (output_line_beginning): Likewise.
9972 (DEFAULT_LINE_START): New macro.
9973 (output_lines): Support color output.
9975 2017-10-31 Martin Liska <mliska@suse.cz>
9977 PR gcov-profile/82633
9978 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9979 their interaction with GCOV infrastructure.
9980 * configure.ac: Add -fkeep-{inline,static}-functions to
9982 * configure: Regenerate.
9984 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9987 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9989 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9992 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9993 into a register if it does not fit into an immediate offset field.
9995 2017-10-31 Olivier Hainque <hainque@adacore.com>
9997 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9999 2017-10-31 Julia Koval <julia.koval@intel.com>
10001 * config.gcc: Add gfniintrin.h.
10002 * config/i386/gfniintrin.h: New.
10003 * config/i386/i386-builtin-types.def
10004 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10005 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10006 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10007 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10008 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10009 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10010 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10011 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10012 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10013 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10014 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10015 * config/i386/immintrin.h: Include gfniintrin.h.
10016 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10018 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10020 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10022 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10024 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10025 (arm_ashldi3_1bit): Remove pattern.
10026 (ashrdi3): Remove shift by 1 expansion.
10027 (arm_ashrdi3_1bit): Remove pattern.
10028 (lshrdi3): Remove shift by 1 expansion.
10029 (arm_lshrdi3_1bit): Remove pattern.
10030 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10031 cost of ashldi3 by 1.
10032 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10033 (<shift>di3_neon): Likewise.
10035 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10037 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10038 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10039 and (*aarch64_simd_mov<VQ:mode>).
10040 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10041 pattern alternative.
10042 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10043 attributes to match pattern alternative.
10045 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10047 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10048 * config/rs6000/emmintrin.h: New file.
10049 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10051 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10053 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10054 * (movdi_vfp_cortexa8): Remove pattern.
10056 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10058 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10060 (Specific, alpha*-dec-osf5.1): Remove.
10061 (Specific, mips-sgi-irix5): Remove.
10062 (Specific, mips-sgi-irix6): Remove.
10064 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10066 PR middle-end/22141
10067 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10068 arguments to clear_bit_region_be.
10070 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10072 * gimplify.c: Include memmodel.h.
10074 2017-10-30 Martin Jambor <mjambor@suse.cz>
10076 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10077 condition whether loop should be executed at all.
10079 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10082 gimple folding of vec_madd() intrinsics.
10083 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10084 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10085 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10087 2017-10-30 Richard Biener <rguenther@suse.de>
10089 PR tree-optimization/82762
10091 2017-10-23 Richard Biener <rguenther@suse.de>
10093 PR tree-optimization/82129
10095 2017-08-01 Richard Biener <rguenther@suse.de>
10097 PR tree-optimization/81181
10098 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10099 (compute_antic): ... end of iteration here.
10101 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10103 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10104 -std=iso9899:2017 and -std=gnu17.
10105 * doc/standards.texi (C Language): Document C17 support.
10106 * doc/cpp.texi (Overview): Mention -std=c17.
10107 (Standard Predefined Macros): Document C11 and C17 values of
10108 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10109 * doc/extend.texi (Inline): Do not list individual options for
10110 standards newer than C99.
10111 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10113 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10116 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10118 * asan.c (asan_finish_file): Align asan globals array by shadow
10121 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10123 PR middle-end/22141
10124 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10125 (struct store_immediate_info): Add bitregion_start and bitregion_end
10127 (store_immediate_info::store_immediate_info): Add brs and bre
10128 arguments and initialize bitregion_{start,end} from those.
10129 (struct merged_store_group): Add bitregion_start, bitregion_end,
10130 align_base and mask fields. Drop unnecessary struct keyword from
10131 struct store_immediate_info. Add do_merge method.
10132 (clear_bit_region_be): Use memset instead of loop storing zeros.
10133 (merged_store_group::do_merge): New method.
10134 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10135 stores as long as the surrounding bitregions have no gaps.
10136 (merged_store_group::merge_overlapping): Use do_merge.
10137 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10138 is byte aligned, rather than requiring that start and width are
10139 byte aligned. Drop unnecessary struct keyword from
10140 struct store_immediate_info. Allocate and populate also mask array.
10141 Make start of the arrays relative to bitregion_start rather than
10142 start and size them according to bitregion_{end,start} difference.
10143 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10144 struct store_immediate_info.
10145 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10146 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10147 struct keyword from struct store_immediate_info.
10148 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10149 between stores as long as the surrounding bitregions have no gaps.
10151 (struct split_store): Add orig non-static data member.
10152 (split_store::split_store): Initialize orig to false.
10153 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10154 if there is exactly a single original stmt. Change stmts argument
10155 to pointer from reference, if NULL, don't push anything to it. Add
10156 first argument, use it to optimize skipping over orig stmts that
10157 are known to be before bitpos already. Simplify.
10158 (split_group): Return unsigned int count how many stores are or
10159 would be needed rather than a bool. Add allow_unaligned argument.
10160 Change split_stores argument from reference to pointer, if NULL,
10161 only do a dry run computing how many stores would be produced.
10162 Rewritten algorithm to use both alignment and misalign if
10163 !allow_unaligned and handle bitfield stores with gaps.
10164 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10165 from bitregion_start instead of start. Compute allow_unaligned
10166 here, if true, do 2 split_group dry runs to compute which one
10167 produces fewer stores and prefer aligned if equal. Punt if
10168 new count is bigger or equal than original before emitting any
10169 statements, rather than during that. Remove no longer needed
10170 new_ssa_names tracking. Replace num_stmts with
10171 split_stores.length (). Use 32-bit stack allocated entries
10172 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10173 if possible. Handle bitfields with gaps.
10174 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10175 Compute bitregion_{start,end} for the stores and construct
10176 store_immediate_info with that. Formatting fixes.
10178 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10181 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10182 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10184 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10186 * gimplify.c: Include tm_p.h.
10188 * common.opt (gcoff): Re-add as ignored option.
10189 (gcoff1, gcoff2, gcoff3): Likewise.
10191 * Makefile.in (OBJS): Delete sdbout.o.
10192 (GTFILES): Delete $(srcdir)/sdbout.c.
10193 * debug.h: Delete sdb_debug_hooks.
10194 * final.c: Delete sdbout.h include.
10195 (final_scan_insn): Delete SDB_DEBUG check.
10196 (rest_of_clean_state): Likewise.
10197 * output.h: Delete sdb_begin_function_line.
10198 * sdbout.c: Delete.
10199 * sdbout.h: Delete.
10200 * toplev.c: Delete sdbout.h include.
10201 (process_options): Delete SDB_DEBUG check.
10202 * tree-core.h (tree_type_common): Delete pointer field of
10204 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10205 TYPE_SYMTAB_POINTER.
10206 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10207 (TYPE_SYMTAB_IS_POINTER): Delete.
10208 (TYPE_SYMTAB_IS_DIE): Renumber.
10209 * xcoffout.c: Refer to former sdbout.c file.
10210 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10212 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10213 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10214 (Debugging Options): Delete -gcoff.
10215 (-gstabs): Delete SDB reference.
10217 (-gcoff@var{level}): Delete.
10218 * doc/passes.texi (Debugging information output): Delete SDB and
10219 sdbout.c references.
10220 * doc/tm.texi: Regenerate.
10221 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10222 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10224 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10225 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10226 (SDB_DEBUGGING_INFO): Delete.
10227 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10228 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10229 * target.def (output_source_filename): Delete COFF reference.
10231 * common.opt (gcoff): Delete.
10232 (gxcoff+): Update Negative chain.
10233 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10235 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10236 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10237 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10238 and SDB references.
10239 (expand_function_start): Change sdb reference to past tense.
10240 (expand_function_end): Change sdb reference to past tense.
10241 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10242 * opts.c (debug_type_names): Delete coff entry.
10243 (common_handle_option): Delete OPT_gcoff case.
10244 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10246 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10247 * config/cris/cris.h: Delete SDB reference in comment.
10248 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10249 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10250 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10251 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10253 (ix86_expand_prologue): Likewise.
10254 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10255 * config/ia64/ia64.h: Likewise.
10256 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10257 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10259 * config/mmix/mmix.h: Likewise.
10260 * config/nds32/nds32.c: Likewise.
10261 * config/stormy/storym16.h: Likewise.
10262 * config/visium/visium.h: Likewise.
10263 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10265 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10267 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10268 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10269 FRAME_POINTER_REGNUM point at high end of local var area.
10271 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10273 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10274 Move comment around. Do not reset best_edge for a copiable
10275 destination if the copy would cause a partition change.
10276 (better_edge_p): Remove redundant check.
10278 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10280 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10282 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10284 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10285 for math functions that have _Float<N> and _Float<N>X variants.
10286 (mathfn_built_in_2): Add support for math functions that have
10287 _Float<N> and _Float<N>X variants.
10288 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10289 (expand_builtin_mathfn_ternary): Add support for fma with
10290 _Float<N> and _Float<N>X variants.
10291 (expand_builtin): Likewise.
10292 (fold_builtin_3): Likewise.
10293 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10294 create math function _Float<N> and _Float<N>X variants as external
10296 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10297 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10298 the __builtin_ prefix and if not strict ansi, without the prefix.
10299 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10300 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10301 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10302 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10303 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10304 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10305 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10306 function signatures for fma _Float<N> and _Float<N>X variants.
10307 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10308 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10309 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10310 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10311 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10312 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10313 * gencfn-macros.c (print_case_cfn): Add support for math functions
10314 that have _Float<N> and _Float<N>X variants.
10315 (print_define_operator_list): Likewise.
10316 (fltfn_suffixes): Likewise.
10318 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10319 for math functions that have _Float<N> and _Float<N>X variants.
10320 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10321 and _Float<N>X variants.
10322 (COPYSIGN): Likewise.
10325 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10326 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10328 (integer_valued_read_call_p): Likewise.
10329 * fold-const-call.c (fold_const_call_ss): Likewise.
10330 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10331 _Float<N> and _Float<N>X variants.
10332 (fold_const_call_ssss): Add support for fma _Float<N> and
10333 _Float<N>X variants.
10334 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10335 support for copysign and fma _Float<N> and _Float<N>X variants.
10336 (backprop::process_builtin_call_use): Likewise.
10337 * tree-call-cdce.c (can_test_argument_range); Add support for
10338 sqrt _Float<N> and _Float<N>X variants.
10339 (edom_only_function): Likewise.
10340 (get_no_error_domain): Likewise.
10341 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10342 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10343 copysign _Float<N> and _Float<N>X variants.
10344 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10345 handled by machine independent code.
10346 (FMAF128): Likewise.
10347 * doc/cpp.texi (Common Predefined Macros): Document defining
10348 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10349 fma _Float<N> and _Float<N>X variants.
10351 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10354 * config/i386/i386-modes.def (CCFPU): Remove definition.
10355 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10356 (ix86_cc_modes_compatible): Ditto.
10357 (ix86_expand_carry_flag_compare): Ditto.
10358 (ix86_expand_int_movcc): Ditto.
10359 (ix86_expand_int_addcc): Ditto.
10360 (ix86_reverse_condition): Ditto.
10361 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10362 Return true/false for unordered/ordered fp comparisons.
10363 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10364 (ix86_prepare_fp_compare_args): Update for rename.
10365 (ix86_expand_fp_compare): Update for rename. Generate unordered
10366 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10367 (ix86_expand_sse_compare_and_jump): Ditto.
10368 * config/i386/predicates.md (fcmov_comparison_operator):
10369 Remove CCFPU mode handling.
10370 (ix86_comparison_operator): Ditto.
10371 (ix86_carry_flag_operator): Ditto.
10372 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10373 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10374 (*cmpu<mode>_cc_i387): Ditto.
10375 (FPCMP): Remove mode iterator.
10376 (unord): Remove mode attribute.
10377 (unord_subst): New define_subst transformation
10378 (unord): New define_subst attribute.
10379 (unordered): Ditto.
10380 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10381 (*cmpi<unord>xf_i387): Ditto.
10382 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10383 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10384 using unord_subst transformation.
10385 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10386 (round_saeonly): Also handle CCFP mode.
10387 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10388 Remove UNSPEC_SAHF unspec handling.
10390 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10392 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10394 2017-10-27 Jeff Law <law@redhat.com>
10396 * gimple-ssa-sprintf.c: Include domwalk.h.
10397 (class sprintf_dom_walker): New class, derived from dom_walker.
10398 (sprintf_dom_walker::before_dom_children): New function.
10399 (struct call_info): Moved into sprintf_dom_walker class
10400 (compute_formath_length, handle_gimple_call): Likewise.
10401 (sprintf_length::execute): Call the dominator walker rather
10402 than walking the statements.
10404 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10405 gimple statement locations.
10406 (check_array_bounds): Corresponding changes. Get the statement's
10407 location directly from wi->stmt.
10409 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10412 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10414 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10416 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10417 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10419 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10422 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10423 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10424 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10425 maybe_get_pool_constant.
10426 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10429 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10431 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10432 2.26 caveat. Update gas and gld versions.
10433 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10436 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10438 * cgraph.h (set_malloc_flag): Declare.
10439 * cgraph.c (set_malloc_flag_1): New function.
10440 (set_malloc_flag): Likewise.
10441 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10442 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10444 (read_ipa_call_summary): Add support for reading is_return_callee.
10445 (write_ipa_call_summary): Stream is_return_callee.
10446 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10447 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10448 ipa-prop.h, ipa-fnsummary.h.
10449 (pure_const_names): Change to static.
10450 (malloc_state_e): Define.
10451 (malloc_state_names): Define.
10452 (funct_state_d): Add field malloc_state.
10453 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10454 (check_retval_uses): New function.
10455 (malloc_candidate_p): Likewise.
10456 (analyze_function): Add support for malloc attribute.
10457 (pure_const_write_summary): Stream malloc_state.
10458 (pure_const_read_summary): Add support for reading malloc_state.
10459 (dump_malloc_lattice): New function.
10460 (propagate_malloc): New function.
10461 (warn_function_malloc): New function.
10462 (ipa_pure_const::execute): Call propagate_malloc and
10463 ipa_free_fn_summary.
10464 (pass_local_pure_const::execute): Add support for malloc attribute.
10465 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10466 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10468 2017-10-27 Martin Liska <mliska@suse.cz>
10470 PR gcov-profile/82457
10471 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10472 for fork-like functions to be properly instrumented.
10474 2017-10-27 Richard Biener <rguenther@suse.de>
10476 PR middle-end/81659
10477 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10478 info when we redirected EH.
10480 2017-10-26 Michael Collison <michael.collison@arm.com>
10482 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10484 (<optab>_trunchf<GPI:mode>2: New pattern.
10485 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10486 * config/aarch64/iterators.md (wv): New mode attribute.
10487 (vf, VF): New mode attributes.
10488 (vgp, VGP): New mode attributes.
10489 (s): Update attribute with SImode and DImode prefixes.
10491 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10493 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10494 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10495 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10496 (nios2_option_overide): Initialize it. Don't allow R0-relative
10497 addressing with PIC.
10498 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10499 (nios2_symbolic_constant_p): Likewise.
10500 (nios2_legitimate_address_p): Likewise.
10501 (nios2_r0rel_section_name_p): New.
10502 (nios2_symbol_ref_in_r0rel_data_p): New.
10503 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10504 (r0rel_constant_p): New.
10505 (nios2_print_operand_address): Handle r0rel_constant_p.
10506 (nios2_cdx_narrow_form_p): Likewise.
10507 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10508 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10509 (Nios II Options): Document -mr0rel-sec.
10511 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10513 * config/nios2/nios2.c: Include xregex.h.
10514 (nios2_gprel_sec_regex): New.
10515 (nios2_option_overide): Initialize it. Don't allow GP-relative
10516 addressing with PIC.
10517 (nios2_small_section_name_p): Check for regex match.
10518 * config/nios2/nios2.opt (mgprel-sec=): New option.
10519 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10520 (Nios II Options): Document -mgprel-sec.
10522 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10524 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10526 2017-10-26 Tom de Vries <tom@codesourcery.com>
10528 PR tree-optimization/82707
10529 * gimple.c (gimple_copy): Fix unsharing of
10530 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10532 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10534 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10535 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10536 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10537 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10538 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10539 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10540 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10541 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10542 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10543 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10544 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10545 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10546 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10547 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10548 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10549 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10550 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10552 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10554 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10556 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10557 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10558 warning to rs6000.c. Remove the Undocumented flag, since it has
10560 (-mabi=ibmlongdouble): Likewise.
10561 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10562 already set, set the default format for long double.
10563 (rs6000_debug_reg_global): Print whether long double is IBM or
10565 (rs6000_option_override_internal): Rework setting long double
10566 format. Only warn if the user is changing the long double default
10567 and they did not use -Wno-psabi.
10568 * doc/invoke.texi (PowerPC options): Update the documentation for
10569 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10571 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10572 Alan Hayward <alan.hayward@arm.com>
10573 David Sherwood <david.sherwood@arm.com>
10575 * rtl.h (wider_subreg_mode): New function.
10576 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10577 rather than an unsigned int *.
10578 * ira-color.c (regno_max_ref_width): Replace with...
10579 (regno_max_ref_mode): ...this new variable.
10580 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10581 Use wider_subreg_mode.
10582 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10583 rather than an unsigned int *.
10584 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10585 (process_alt_operands): Likewise.
10586 (invariant_p): Likewise.
10587 * lra-spills.c (assign_mem_slot): Likewise.
10588 (add_pseudo_to_slot): Likewise.
10589 * lra.c (collect_non_operand_hard_regs): Likewise.
10590 (add_regs_to_insn_regno_info): Likewise.
10591 * reload1.c (regno_max_ref_width): Replace with...
10592 (regno_max_ref_mode): ...this new variable.
10593 (reload): Update accordingly. Update call to
10594 ira_sort_regnos_for_alter_reg.
10595 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10596 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10597 (scan_paradoxical_subregs): Likewise.
10599 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10601 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10602 (aarch64_frame): Add emit_frame_chain boolean.
10603 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10604 Move eh_return case to aarch64_layout_frame.
10605 (aarch64_layout_frame): Initialize emit_frame_chain.
10606 (aarch64_expand_prologue): Use emit_frame_chain.
10608 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10610 * config/aarch64/aarch64.c (aarch64_layout_frame):
10611 Ensure LR is always stored at the bottom of the callee-saves.
10612 Remove rarely used frame layout which saves callee-saves at top of
10613 frame, so the store of LR can be used as a valid probe in all cases.
10615 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10617 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10618 Improve unaligned TImode/TFmode base/offset split.
10620 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10625 * combine.c (find_single_use_1): Likewise.
10626 (expand_field_assignment): Likewise.
10627 (move_deaths): Likewise.
10628 * lra-constraints.c (simplify_operand_subreg): Likewise.
10629 (curr_insn_transform): Likewise.
10630 * lra.c (collect_non_operand_hard_regs): Likewise.
10631 (add_regs_to_insn_regno_info): Likewise.
10632 * rtlanal.c (reg_referenced_p): Likewise.
10633 (covers_regno_no_parallel_p): Likewise.
10635 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10637 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10638 Don't print any bits outside the precision of the value.
10639 * wide-int.cc (test_printing): Add some new tests.
10641 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10643 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10644 supports -xbrace_comment option.
10645 * configure: Regenerate.
10646 * config.in: Regenerate.
10647 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10648 (ASM_CPU_SPEC): Use it.
10650 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10652 * target.def (static_rtx_alignment): New hook.
10653 * targhooks.h (default_static_rtx_alignment): Declare.
10654 * targhooks.c (default_static_rtx_alignment): New function.
10655 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10656 * doc/tm.texi: Regenerate.
10657 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10658 instead of targetm.constant_alignment. Remove call to
10659 set_mem_attributes.
10660 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10661 (cris_preferred_mininum_alignment): New function, split out from...
10662 (cris_constant_alignment): ...here.
10663 (cris_static_rtx_alignment): New function.
10664 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10666 (ix86_constant_alignment): ...here.
10667 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10668 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10669 (mmix_static_rtx_alignment): New function.
10670 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10671 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10673 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10676 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10677 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10679 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10681 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10682 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10683 mark as a sign-extending load.
10684 (local_pic_load_u): Define.
10686 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10688 PR middle-end/82062
10689 * fold-const.c (operand_equal_for_comparison_p): Also return true
10690 if ARG0 is a simple variant of ARG1 with narrower precision.
10691 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10693 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10695 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10697 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10698 scatter_static, scatter_per_elt.
10699 * x86-tune-costs.h: Add new cost entries.
10701 2017-10-25 Richard Biener <rguenther@suse.de>
10703 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10704 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10705 class pass_fre): Move to ...
10706 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10707 class pass_fre): ... here and adjust for statistics.
10709 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10712 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10713 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10714 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10715 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10718 2017-10-25 Richard Biener <rguenther@suse.de>
10720 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10721 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10722 eliminate_push_avail, eliminate_insert): Move inside...
10723 (class eliminate_dom_walker): ... this class in preparation
10725 (fini_eliminate): Remove by merging with ...
10726 (eliminate): ... this function. Adjust for class changes.
10727 (pass_pre::execute): Remove fini_eliminate call.
10728 (pass_fre::execute): Likewise.
10730 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10733 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10734 (VPERMI2, VPERMI2I): New mode iterators.
10735 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10736 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10738 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10739 mode iterator. Remove 3 old define_insn patterns.
10740 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10741 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10742 VPERMI2 mode iterator, remove the other two expanders.
10743 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10744 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10745 instructions, remove the other two patterns.
10746 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10747 mode iterator, remove the other two patterns.
10748 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10749 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10750 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10751 and adjust argument order accordingly.
10752 (ix86_expand_vec_perm): Adjust caller.
10753 (expand_vec_perm_1): Likewise.
10754 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10755 (expand_vec_perm_vpermt2_vpshub2): ... this.
10756 (ix86_expand_vec_perm_const_1): Adjust caller.
10757 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10760 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10761 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10762 (vec_shl_<mode>): Remove unused expander.
10763 (avx512bw_<shift_insn><mode>3): New define_insn.
10764 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10765 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10767 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10770 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10773 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10775 PR rtl-optimization/82396
10776 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10777 (autopref_multipass_init): Simplify initialization.
10778 (autopref_rank_data): Simplify sort order.
10779 * gcc/sched-int.h (autopref_multipass_data_): Remove
10780 multi_mem_insn_p, min_offset and max_offset.
10782 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10784 PR middle-end/60580
10785 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10786 Check special value of flag_omit_frame_pointer.
10787 (aarch64_can_eliminate): Likewise.
10788 (aarch64_override_options_after_change_1): Simplify handling of
10789 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10791 2017-10-24 Richard Biener <rguenther@suse.de>
10793 PR tree-optimization/82697
10794 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10795 zero for conditional load and unconditional store.
10797 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10799 * doc/install.texi: Document bootstrap-cet.
10801 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10804 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10805 ENDBR instruction at function entrance if function is only
10808 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10811 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10812 patterns to better describe from which operation the CF is computed.
10813 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10814 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10815 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10816 is 0, use _0 suffixed expanders instead of emitting a comparison
10819 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10821 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10822 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10824 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10826 PR middle-end/82569
10827 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10828 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10829 * loop-iv.c (iv_get_reaching_def): Likewise.
10830 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10831 variable is promoted and the partition contains undefined values.
10833 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10835 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10837 (nios2_address_cost): Define.
10838 (nios2_legitimize_address): Recognize (exp + constant) directly.
10839 (TARGET_ADDRESS_COST): Define.
10841 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10843 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10844 (nios2_symbolic_memory_operand_p): Declare.
10845 (nios2_split_large_constant): Declare.
10846 (nios2_split_symbolic_memory_operand): Declare.
10847 * config/nios2/nios2.c: Adjust includes.
10848 (nios2_symbolic_constant_allowed): New.
10849 (nios2_symbolic_constant_p): New.
10850 (nios2_plus_symbolic_constant_p): New.
10851 (nios2_valid_addr_expr_p): Recognize addresses involving
10852 symbolic constants.
10853 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10854 (nios2_symbolic_memory_operand_p): New.
10855 (nios2_large_constant_p): New.
10856 (nios2_split_large_constant): New.
10857 (nios2_split_plus_large_constant): New.
10858 (nios2_split_symbolic_memory_operand): New.
10859 (nios2_legitimize_address): Code refactoring. Handle addresses
10860 involving symbolic constants.
10861 (nios2_emit_move_sequence): Likewise.
10862 (nios2_print_operand): Improve error output.
10863 (nios2_print_operand_address): Handle LO_SUM.
10864 (nios2_cdx_narrow_form_p): Likewise.
10865 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10866 operands involving symbolic constants.
10867 (movhi_internal, movsi_internal): Likewise.
10868 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10869 (extendhisi2, extendqi<mode>2): Likewise.
10871 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10873 * tree-pass.h (PROP_rtl_split_insns): Define.
10874 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10876 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10878 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10880 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10883 * target.def (const_not_ok_for_debug_p): Default to
10884 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10885 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10886 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10887 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10888 which targetm.const_not_ok_for_debug_p returned true.
10889 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10891 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10893 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10894 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10895 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10897 (ix86_const_not_ok_for_debug_p): New function.
10898 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10899 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10901 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10904 * system.h: Conditionally include "unique-ptr.h" if
10905 INCLUDE_UNIQUE_PTR is defined.
10906 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10907 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10909 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10911 * config/rl78/rl78.md: New define_expand "subdi3".
10913 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10916 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10917 DF_REF_INSN if DF_REF_INSN_INFO is false.
10919 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10921 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10922 xmm_move instead of sse_move.
10923 (sse_store_index): New function.
10924 (ix86_register_move_cost): Be more sensible about mismatch stall;
10925 model AVX moves correctly; make difference between sse->integer and
10927 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10928 moves; make difference between SSE and AVX.
10929 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10930 and zmm_move. Increase size of sse load and store tables;
10931 add unaligned load and store tables; add ssemmx_to_integer.
10932 * x86-tune-costs.h: Update all entries according to real
10933 move latencies from Agner Fog's manual and chip documentation.
10935 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10938 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10939 * config/i386/constraints.md (Wf): New constraint.
10940 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10941 (cmp<dwi>_doubleword): Removed.
10942 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10943 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10944 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10945 expand with cmp<dwi>_doubleword. For LTU and GEU use
10946 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10948 * common.opt (gcolumn-info): Enable by default.
10949 * doc/invoke.texi (gcolumn-info): Document new default.
10951 2017-10-23 Richard Biener <rguenther@suse.de>
10953 PR tree-optimization/82672
10954 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10955 Fold the stmt if we propagated into it.
10957 2017-10-23 Richard Biener <rguenther@suse.de>
10959 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10960 (bitmap_remove_expr_from_set): ... this. All callers call this
10961 for non-constant values.
10962 (bitmap_set_subtract): Rename to...
10963 (bitmap_set_subtract_expressions): ... this. Adjust and
10965 (bitmap_set_contains_value): Remove superfluous check.
10966 (bitmap_set_replace_value): Inline into single caller ...
10967 (bitmap_value_replace_in_set): ... here and simplify.
10968 (dependent_clean): Merge into ...
10969 (clean): ... this using an overload. Adjust.
10970 (prune_clobbered_mems): Adjust.
10971 (compute_antic_aux): Likewise.
10972 (compute_partial_antic_aux): Likewise.
10974 2017-10-23 Richard Biener <rguenther@suse.de>
10976 PR tree-optimization/82129
10978 2017-08-01 Richard Biener <rguenther@suse.de>
10980 PR tree-optimization/81181
10981 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10982 (compute_antic): ... end of iteration here.
10984 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10986 * target.def (starting_frame_offset): New hook.
10987 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10988 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10989 * doc/tm.texi.in: Regenerate.
10990 * hooks.h (hook_hwi_void_0): Declare.
10991 * hooks.c (hook_hwi_void_0): New function.
10992 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10993 STARTING_FRAME_OFFSET.
10994 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10995 * reload1.c (reload): Likewise.
10996 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10997 instead of STARTING_FRAME_OFFSET.
10998 * function.c (try_fit_stack_local): Likewise.
10999 (assign_stack_local_1): Likewise
11000 (instantiate_virtual_regs): Likewise.
11001 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11002 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11003 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11004 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11005 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11006 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11007 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11008 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11009 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11010 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11011 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11012 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11013 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11014 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11015 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11016 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11017 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11018 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11019 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11020 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11021 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11022 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11023 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11024 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11025 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11026 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11027 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11028 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11029 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11030 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11031 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11032 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11033 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11034 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11035 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11036 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11037 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11038 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11039 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11040 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11041 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11042 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11043 return a HOST_WIDE_INT.
11044 (avr_builtin_setjmp_frame_value): Use it instead of
11045 STARTING_FRAME_OFFSET.
11046 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11047 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11048 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11050 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11051 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11052 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11053 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11054 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11055 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11056 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11057 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11058 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11059 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11060 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11061 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11063 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11064 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11065 * config/mips/mips.c (mips_compute_frame_info): Refer to
11066 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11067 (mips_starting_frame_offset): New function.
11068 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11069 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11070 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11071 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11072 and return a HOST_WIDE_INT.
11073 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11074 (mmix_initial_elimination_offset): Refer to
11075 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11076 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11077 * config/pa/pa.c (pa_starting_frame_offset): New function.
11078 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11079 (pa_expand_prologue): Likewise.
11080 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11081 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11082 !FRAME_GROWS_DOWNWARD handling to...
11083 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11084 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11085 !FRAME_GROWS_DOWNWARD handling to...
11086 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11087 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11088 !FRAME_GROWS_DOWNWARD handling to...
11089 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11090 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11092 (rs6000_starting_frame_offset): New function.
11093 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11094 !FRAME_GROWS_DOWNWARD handling to...
11095 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11096 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11097 !FRAME_GROWS_DOWNWARD handling to...
11098 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11099 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11100 !FRAME_GROWS_DOWNWARD handling to...
11101 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11102 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11103 (rs6000_starting_frame_offset): New function.
11104 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11105 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11106 * config/vax/vax.c (vax_starting_frame_offset): New function.
11107 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11108 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11109 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11110 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11111 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11112 * system.h (STARTING_FRAME_OFFSET): Poison.
11114 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11116 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11117 SCALAR_TYPE_MODE instead of TYPE_MODE.
11119 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11120 Alan Hayward <alan.hayward@arm.com>
11121 David Sherwood <david.sherwood@arm.com>
11123 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11125 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11126 Alan Hayward <alan.hayward@arm.com>
11127 David Sherwood <david.sherwood@arm.com>
11129 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11131 2017-10-23 Richard Biener <rguenther@suse.de>
11133 PR tree-optimization/82129
11134 * tree-ssa-pre.c (bitmap_set_and): Remove.
11135 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11136 canonicalizing expressions in the set to those with lowest
11137 ID rather than taking that from the first edge.
11139 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11141 * combine.c (rtx_equal_for_field_assignment_p): Use
11142 byte_lowpart_offset.
11144 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11149 to a SUBREG_PROMOTED_VAR.
11151 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11152 Alan Hayward <alan.hayward@arm.com>
11153 David Sherwood <david.sherwood@arm.com>
11155 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11156 (expand_debug_source_expr): Likewise.
11157 * combine.c (combine_simplify_rtx): Likewise.
11158 * cse.c (fold_rtx): Likewise.
11159 * optabs.c (expand_float): Likewise.
11160 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11161 (simplify_binary_operation_1): Likewise.
11163 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11164 Alan Hayward <alan.hayward@arm.com>
11165 David Sherwood <david.sherwood@arm.com>
11167 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11168 (record_promoted_value): Likewise.
11169 * expr.c (expand_expr_real_2): Likewise.
11170 * ree.c (update_reg_equal_equiv_notes): Likewise.
11171 (combine_set_extension): Likewise.
11172 * rtlanal.c (low_bitmask_len): Likewise.
11173 * simplify-rtx.c (neg_const_int): Likewise.
11174 (simplify_binary_operation_1): Likewise.
11176 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11177 Alan Hayward <alan.hayward@arm.com>
11178 David Sherwood <david.sherwood@arm.com>
11180 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11181 * regcprop.c (maybe_mode_change): Likewise.
11182 * reload1.c (alter_reg): Likewise.
11184 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11186 * inchash.h (inchash::hash::add_wide_int): New function.
11187 * lto-streamer-out.c (hash_tree): Use it.
11189 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11191 * inchash.h (inchash::hash::add_wide_int): Rename to...
11192 (inchash::hash::add_hwi): ...this.
11193 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11194 (polymorphic_call_target_hasher::hash): Likewise.
11195 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11196 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11197 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11198 * lto-streamer-out.c (hash_tree): Likewise.
11199 * optc-save-gen.awk: Likewise.
11200 * tree.c (add_expr): Likewise.
11202 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11205 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11206 for ordered inequality comparisons even with TARGET_IEEE_FP.
11208 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11211 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11212 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11213 Expand with cmp<dwi>_doubleword.
11215 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11217 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11218 List CET intrinsics.
11219 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11220 specific to -fcf-protection option.
11222 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11224 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11225 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11226 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11227 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11228 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11229 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11230 (extra_objs): Add cet.o for Linux/x86 targets.
11231 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11232 * config/i386/cet.c: New file.
11233 * config/i386/cetintrin.h: Likewise.
11234 * config/i386/t-cet: Likewise.
11235 * config/i386/cpuid.h (bit_SHSTK): New.
11236 (bit_IBT): Likewise.
11237 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11238 pass IBT and SHSTK bits.
11239 * config/i386/i386-builtin-types.def
11240 (VOID_FTYPE_UNSIGNED_PVOID): New.
11241 (VOID_FTYPE_UINT64_PVOID): Likewise.
11242 * config/i386/i386-builtin.def: Add CET intrinsics.
11243 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11244 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11245 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11246 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11248 * config/i386/i386.c (rest_of_insert_endbranch): New.
11249 (pass_data_insert_endbranch): Likewise.
11250 (pass_insert_endbranch): Likewise.
11251 (make_pass_insert_endbranch): Likewise.
11252 (ix86_notrack_prefixed_insn_p): Likewise.
11253 (ix86_target_string): Add -mibt, -mshstk flags.
11254 (ix86_option_override_internal): Add flag_cf_protection
11256 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11257 (ix86_print_operand): Add 'notrack' prefix output.
11258 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11259 (ix86_expand_builtin): Expand CET intrinsics.
11260 (x86_output_mi_thunk): Add 'endbranch' instruction.
11261 * config/i386/i386.h (TARGET_IBT): New.
11262 (TARGET_IBT_P): Likewise.
11263 (TARGET_SHSTK): Likewise.
11264 (TARGET_SHSTK_P): Likewise.
11265 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11266 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11267 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11268 (builtin_setjmp_setup): New pattern.
11269 (builtin_longjmp): Likewise.
11270 (rdssp<mode>): Likewise.
11271 (incssp<mode>): Likewise.
11272 (saveprevssp): Likewise.
11273 (rstorssp): Likewise.
11274 (wrss<mode>): Likewise.
11275 (wruss<mode>): Likewise.
11276 (setssbsy): Likewise.
11277 (clrssbsy): Likewise.
11278 (nop_endbr): Likewise.
11279 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11281 * config/i386/immintrin.h: Include <cetintrin.h>.
11282 * config/i386/linux-common.h
11283 (file_end_indicate_exec_stack_and_cet): New prototype.
11284 (TARGET_ASM_FILE_END): New.
11286 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11288 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11289 latencies instead of having separate table; make difference between
11290 integer and float costs.
11291 * i386.h (processor_costs): Remove scalar_stmt_cost,
11292 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11293 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11295 * x86-tune-costs.h: Remove entries which has been removed in
11296 procesor_costs from all tables; make cond_taken_branch_cost
11297 and cond_not_taken_branch_cost COST_N_INSNS based.
11299 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11301 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11303 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11305 * config/i386/i386.md (isa): Remove fma_avx512f.
11306 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11307 <avx512>_fmadd_<mode>_mask3<round_name>,
11308 <avx512>_fmsub_<mode>_mask<round_name>,
11309 <avx512>_fmsub_<mode>_mask3<round_name>,
11310 <avx512>_fnmadd_<mode>_mask<round_name>,
11311 <avx512>_fnmadd_<mode>_mask3<round_name>,
11312 <avx512>_fnmsub_<mode>_mask<round_name>,
11313 <avx512>_fnmsub_<mode>_mask3<round_name>,
11314 <avx512>_fmaddsub_<mode>_mask<round_name>,
11315 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11316 <avx512>_fmsubadd_<mode>_mask<round_name>,
11317 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11318 (*vec_widen_umult_even_v16si<mask_name>,
11319 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11320 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11322 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11324 * extend.texi: Add 'nocf_check' documentation.
11325 * gimple.texi: Add second parameter to
11326 gimple_build_call_from_tree.
11327 * invoke.texi: Add -fcf-protection documentation.
11328 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11330 2017-10-20 Richard Biener <rguenther@suse.de>
11332 PR tree-optimization/82473
11333 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11334 the largest input type.
11336 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11338 * c-attribs.c (handle_nocf_check_attribute): New function.
11339 (c_common_attribute_table): Add 'nocf_check' handling.
11340 * gimple-parser.c: Add second argument NULL to
11341 gimple_build_call_from_tree.
11342 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11343 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11345 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11346 * common.opt: Add fcf-protection flag.
11347 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11348 * flag-types.h: Add enum cf_protection_level.
11349 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11350 Add 'nocf_check' attribute propagation to gimple call.
11351 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11352 (gimple_build_call_from_tree): Update prototype.
11353 (gimple_call_nocf_check_p): New function.
11354 (gimple_call_set_nocf_check): Likewise.
11355 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11356 * ipa-icf.c: Add nocf_check attribute in statement hash.
11357 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11358 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11359 * toplev.c (process_options): Add flag_cf_protection handling.
11361 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11363 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11365 2017-10-20 Richard Biener <rguenther@suse.de>
11367 PR tree-optimization/82603
11368 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11369 remove false predicated stores.
11371 2017-10-20 Richard Biener <rguenther@suse.de>
11373 * graphite-isl-ast-to-gimple.c
11374 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11375 Remove return value and simplify, dump copied stmt after lhs
11377 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11378 Reduce dump verbosity.
11379 (gsi_insert_earliest): Likewise.
11380 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11381 * graphite.c (print_global_statistics): Adjust dumping.
11382 (print_graphite_scop_statistics): Likewise.
11383 (print_graphite_statistics): Do not dump loops here.
11384 (graphite_transform_loops): But here.
11386 2017-10-20 Nicolas Roche <roche@adacore.com>
11388 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11389 * configure: Regenerate.
11391 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11394 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11395 functions when optimizing replace GIMPLE_RETURN stmts with
11396 calls to __builtin_unreachable ().
11399 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11400 for -fsanitize=thread link of executables.
11401 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11402 link of executables.
11405 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11406 New mode iterators.
11407 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11408 define_insns for logical vector shifts to use VI248_AVX512BW
11409 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11410 condition, useless isa and prefix attributes. Change the first
11411 2 of these define_insns to ...
11412 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11413 define_insn for avx512vl.
11414 (<shift_insn><mode>3): ... and this, new define_insn without
11415 masking for non-avx512vl.
11418 * config/i386/sse.md (*andnot<mode>3,
11419 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11420 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11421 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11422 not applied use empty suffix even for TARGET_AVX512VL.
11423 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11424 is applied, supply evex,evex or evex,evex,evex instead of just
11427 2017-10-20 Julia Koval <julia.koval@intel.com>
11429 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11430 (OPTION_MASK_ISA_GFNI_UNSET): New.
11431 (ix86_handle_option): Handle OPT_mgfni.
11432 * config/i386/cpuid.h (bit_GFNI): New.
11433 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11434 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11435 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11436 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11437 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11438 * config/i386/i386.opt: Add mgfni.
11440 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11442 * config/msp430/msp430.c (msp430_option_override): Disable
11443 -fdelete-null-pointer-checks.
11444 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11446 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11448 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11449 of x87 and SSE instructions.
11451 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11453 * asan.c (create_cond_insert_point): Do not update edge count.
11454 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11455 (afdo_propagate_circuit): Likewise.
11456 (afdo_calculate_branch_prob): Likewise.
11457 (afdo_annotate_cfg): Likewise.
11458 * basic-block.h (struct edge_def): Remove count.
11459 (edge_def::count): New accessor.
11460 * bb-reorder.c (rotate_loop): Update.
11461 (find_traces_1_round): Update.
11462 (connect_traces): Update.
11463 (sanitize_hot_paths): Update.
11464 * cfg.c (unchecked_make_edge): Update.
11465 (make_single_succ_edge): Update.
11466 (check_bb_profile): Update.
11467 (dump_edge_info): Update.
11468 (update_bb_profile_for_threading): Update.
11469 (scale_bbs_frequencies_int): Update.
11470 (scale_bbs_frequencies_gcov_type): Update.
11471 (scale_bbs_frequencies_profile_count): Update.
11472 (scale_bbs_frequencies): Update.
11473 * cfganal.c (connect_infinite_loops_to_exit): Update.
11474 * cfgbuild.c (compute_outgoing_frequencies): Update.
11475 (find_many_sub_basic_blocks): Update.
11476 * cfgcleanup.c (try_forward_edges): Update.
11477 (try_crossjump_to_edge): Update
11478 * cfgexpand.c (expand_gimple_cond): Update
11479 (expand_gimple_tailcall): Update
11480 (construct_exit_block): Update
11481 * cfghooks.c (verify_flow_info): Update
11482 (redirect_edge_succ_nodup): Update
11483 (split_edge): Update
11484 (make_forwarder_block): Update
11485 (duplicate_block): Update
11486 (account_profile_record): Update
11487 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11488 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11489 * cfgloopmanip.c (scale_loop_profile): Update.
11491 (lv_adjust_loop_entry_edge): Update.
11492 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11493 (force_nonfallthru_and_redirect): Update.
11494 (purge_dead_edges): Update.
11495 (rtl_flow_call_edges_add): Update.
11496 * cgraphunit.c (init_lowered_empty_function): Update.
11497 (cgraph_node::expand_thunk): Update.
11498 * gimple-pretty-print.c (dump_probability): Update.
11499 (dump_edge_probability): Update.
11500 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11501 * haifa-sched.c (sched_create_recovery_edges): Update.
11502 * hsa-gen.c (convert_switch_statements): Update.
11503 * ifcvt.c (dead_or_predicable): Update.
11504 * ipa-inline-transform.c (inline_transform): Update.
11505 * ipa-split.c (split_function): Update.
11506 * ipa-utils.c (ipa_merge_profiles): Update.
11507 * loop-doloop.c (add_test): Update.
11508 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11509 * lto-streamer-in.c (input_cfg): Update.
11510 (input_function): Update.
11511 * lto-streamer-out.c (output_cfg): Update.
11512 * modulo-sched.c (sms_schedule): Update.
11513 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11514 * predict.c (maybe_hot_edge_p): Update.
11515 (unlikely_executed_edge_p): Update.
11516 (probably_never_executed_edge_p): Update.
11517 (dump_prediction): Update.
11518 (drop_profile): Update.
11519 (propagate_unlikely_bbs_forward): Update.
11520 (determine_unlikely_bbs): Update.
11521 (force_edge_cold): Update.
11522 * profile.c (compute_branch_probabilities): Update.
11523 * reg-stack.c (better_edge): Update.
11524 * shrink-wrap.c (handle_simple_exit): Update.
11525 * tracer.c (better_p): Update.
11526 * trans-mem.c (expand_transaction): Update.
11527 (split_bb_make_tm_edge): Update.
11528 * tree-call-cdce.c: Update.
11529 * tree-cfg.c (gimple_find_sub_bbs): Update.
11530 (gimple_split_edge): Update.
11531 (gimple_duplicate_sese_region): Update.
11532 (gimple_duplicate_sese_tail): Update.
11533 (gimple_flow_call_edges_add): Update.
11534 (insert_cond_bb): Update.
11535 (execute_fixup_cfg): Update.
11536 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11537 * tree-complex.c (expand_complex_div_wide): Update.
11538 * tree-eh.c (lower_resx): Update.
11539 (unsplit_eh): Update.
11540 (cleanup_empty_eh_move_lp): Update.
11541 * tree-inline.c (copy_edges_for_bb): Update.
11542 (freqs_to_counts): Update.
11543 (copy_cfg_body): Update.
11544 * tree-ssa-dce.c (remove_dead_stmt): Update.
11545 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11546 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11547 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11548 (unloop_loops): Update.
11549 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11550 * tree-ssa-loop-split.c (connect_loops): Update.
11551 (split_loop): Update.
11552 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11553 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11554 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11555 * tree-ssa-reassoc.c (branch_fixup): Update.
11556 * tree-ssa-tail-merge.c (replace_block_by): Update.
11557 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11558 (compute_path_counts): Update.
11559 (update_profile): Update.
11560 (recompute_probabilities): Update.
11561 (update_joiner_offpath_counts): Update.
11562 (estimated_freqs_path): Update.
11563 (freqs_to_counts_path): Update.
11564 (clear_counts_path): Update.
11565 (ssa_fix_duplicate_block_edges): Update.
11566 (duplicate_thread_path): Update.
11567 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11568 (case_bit_test_cmp): Update.
11569 (collect_switch_conv_info): Update.
11570 (gen_inbound_check): Update.
11571 (do_jump_if_equal): Update.
11572 (emit_cmp_and_jump_insns): Update.
11573 * tree-tailcall.c (decrease_profile): Update.
11574 (eliminate_tail_call): Update.
11575 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11576 (vect_do_peeling): Update.
11577 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11578 * ubsan.c (ubsan_expand_null_ifn): Update.
11579 (ubsan_expand_ptr_ifn): Update.
11580 * value-prof.c (gimple_divmod_fixed_value): Update.
11581 (gimple_mod_pow2): Update.
11582 (gimple_mod_subtract): Update.
11583 (gimple_ic): Update.
11584 (gimple_stringop_fixed_value): Update.
11586 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11589 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11591 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11593 PR rtl-optimization/82395
11594 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11595 based on non_spilled_static_chain_regno_p.
11597 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11599 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11600 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11601 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11603 2017-10-19 Martin Sebor <msebor@redhat.com>
11605 PR tree-optimization/82596
11606 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11608 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11610 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11611 (get_mem_refs_of_builtin_call): Likewise.
11612 * builtins.c (expand_builtin_apply): Adjust call to
11613 allocate_dynamic_stack_space.
11614 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11615 the third argument to allocate_dynamic_stack_space, otherwise -1.
11616 (expand_builtin): Deal with all alloca variants.
11617 (is_inexpensive_builtin): Likewise.
11618 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11619 * calls.c (special_function_p): Deal with all alloca variants.
11620 (initialize_argument_information): Adjust call to
11621 allocate_dynamic_stack_space.
11622 (expand_call): Likewise.
11623 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11624 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11625 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11626 use it for the stack usage computation.
11627 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11628 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11629 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11630 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11631 (in_loop_p): Remove first argument and useless check.
11632 (pass_walloca::execute): Remove useless test and adjust call to above.
11633 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11634 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11635 (gimplify_call_expr): Deal with all alloca variants.
11636 * hsa-gen.c (gen_hsa_alloca): Likewise.
11637 (gen_hsa_insns_for_call): Likewise.
11638 * ipa-pure-const.c (special_builtin_state): Likewise.
11639 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11640 * tree-object-size.c (alloc_object_size): Likewise.
11641 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11642 (call_may_clobber_ref_p_1): Likewise.
11643 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11644 (ccp_fold_stmt): Likewise.
11645 (optimize_stack_restore): Likewise.
11646 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11647 (mark_all_reaching_defs_necessary_1): Likewise.
11648 (propagate_necessity): Likewise.
11649 (eliminate_unnecessary_stmts): Likewise.
11650 * tree.c (build_common_builtin_nodes): Build
11651 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11652 (build_alloca_call_expr): New function.
11653 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11654 (CASE_BUILT_IN_ALLOCA): Likewise.
11655 (build_alloca_call_expr): Declare.
11656 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11658 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11661 * dwarf2out.c (new_die_raw): New static inline function.
11662 (new_die): Use it to create the DIE.
11663 (add_AT_external_die_ref): Likewise.
11664 (clone_die): Likewise.
11665 (clone_as_declaration): Likewise.
11666 (dwarf2out_vms_debug_main_pointer): Likewise.
11667 (base_type_die): Likewise. Remove early return for corner cases.
11668 Do not call add_pubtype on the DIE here.
11669 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11670 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11671 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11672 native order exists for base types, attach the DIE manually and call
11673 add_pubtype on it. Do not equate a reverse order DIE to the type.
11675 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11677 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11678 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11680 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11682 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11684 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11685 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11686 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11687 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11688 * opts.c (sanitizer_opts): Add builtin.
11689 * ubsan.c (instrument_builtin): New function.
11690 (pass_ubsan::execute): Call it.
11691 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11692 * doc/invoke.texi: Document -fsanitize=builtin.
11694 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11695 builtins, store max (log2 (align), 0) into uchar field instead of
11696 align into uptr field.
11697 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11698 store uchar 0 field instead of uptr 0 field.
11699 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11700 instead of passing one address of struct with 2 locations pass
11701 two addresses of structs with 1 location each.
11702 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11703 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11704 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11705 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11706 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11707 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11708 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11709 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11711 2017-10-19 Martin Liska <mliska@suse.cz>
11714 * file-find.c (remove_prefix): Remove.
11715 * file-find.h (remove_prefix): Likewise.
11716 * gcc-ar.c: Remove smartness of lookup.
11718 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11720 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11721 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11722 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11724 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11727 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11728 (setcc + and to xor + setcc): New peephole2.
11730 2017-10-19 Tom de Vries <tom@codesourcery.com>
11732 * doc/sourcebuild.texi (Test Directives, Variants of
11733 dg-require-support): Add dg-require-stack-size.
11735 2017-10-19 Martin Liska <mliska@suse.cz>
11738 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11739 that have a large alignment.
11740 (gimplify_target_expr): Likewise.
11742 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11744 PR rtl-optimization/82602
11745 * ira.c (rtx_moveable_p): Return false for volatile asm.
11747 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11750 * config/i386/i386-modes.def (CCGZ): New CC mode.
11751 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11752 * config/i386/predicates.md (ix86_comparison_operator):
11754 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11755 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11756 with double-word subtraction.
11757 (put_condition_code): Handle CCGZmode.
11759 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11761 * wide-int.cc (debug (const wide_int &)): New.
11762 (debug (const wide_int *)): New.
11763 (debug (const widest_int &)): New.
11764 (debug (const widest_int *)): New.
11766 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11768 PR middle-end/82556
11769 * lra-constraints.c (curr_insn_transform): Use non-input operand
11770 instead of output one for matched reload.
11772 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11774 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11775 (tree-ssa-loop-ivopts.h): New header file.
11776 (struct builtin_info): New fields.
11777 (classify_builtin_1): Compute and record base and offset parts for
11778 memset builtin partition by calling strip_offset.
11779 (offset_cmp, fuse_memset_builtins): New functions.
11780 (finalize_partitions): Fuse adjacent memset partitions by calling
11782 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11783 Expose the interface.
11784 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11786 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11788 PR tree-optimization/82574
11789 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11790 that data reference must be executed exactly once per iteration
11791 against the outermost loop in nest.
11792 (classify_partition): Update call to above function.
11794 2017-10-18 Richard Biener <rguenther@suse.de>
11796 PR tree-optimization/82591
11797 * graphite.c (graphite_transform_loops): Move code gen message
11799 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11800 Here. Handle scop_to_isl_ast failing.
11801 (scop_to_isl_ast): Limit the number of ISL operations.
11803 2017-10-18 Richard Biener <rguenther@suse.de>
11805 * graphite-isl-ast-to-gimple.c
11806 (translate_isl_ast_to_gimple::set_rename): Simplify.
11807 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11808 (graphite_copy_stmts_from_block): ... here.
11809 (copy_bb_and_scalar_dependences): Simplify.
11810 (add_parameters_to_ivs_params): Canonicalize.
11811 (generate_entry_out_of_ssa_copies): Simplify.
11812 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11813 by passing in ISL dimension.
11814 (parameter_index_in_region_1): Rename to ...
11815 (parameter_index_in_region): ... this.
11816 (extract_affine): Adjust assert, pass down parameter index.
11817 (add_param_constraints): Use range-info when available.
11818 (build_scop_context): Adjust.
11819 * sese.c (new_sese_info): Adjust.
11820 (free_sese_info): Likewise.
11821 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11822 Remove unused typedefs.
11823 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11825 2017-10-18 Martin Liska <mliska@suse.cz>
11827 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11829 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11831 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11832 (s390_sched_init): Do not reset s390_sched_state if we entered the
11833 current basic block via a fallthru edge and all others are unlikely.
11835 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11837 * config/s390/s390.c (NUM_SIDES): New variable.
11838 (LONGRUNNING_THRESHOLD): New variable.
11839 (LATENCY_FACTOR): New variable.
11840 (s390_sched_score): Decrease score for long-running instructions on
11842 (s390_sched_variable_issue): Perform bookkeeping for long-running
11845 2017-10-18 Richard Biener <rguenther@suse.de>
11847 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11848 Simplify with removal of the parameter rename map.
11849 (set_rename): Likewise.
11850 (should_copy_to_new_region): Likewise.
11851 (graphite_copy_stmts_from_block): Likewise.
11852 (copy_bb_and_scalar_dependences): Remove initialization of
11853 unused copied_bb_map.
11854 (copy_def): Remove.
11855 (copy_internal_parameters): Likewise.
11856 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11857 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11858 Use INTEGRAL_TYPE_P.
11859 (parameter_index_in_region_1): Rename to ...
11860 (assign_parameter_index_in_region): ... this. Assert we have
11861 a parameter we handle.
11862 (scan_tree_for_params): Adjust.
11863 * sese.h (parameter_rename_map_t): Remove.
11864 (struct sese_info_t): Remove unused parameter_rename_map and
11865 copied_bb_map members.
11866 * sese.c (new_sese_info): Adjust.
11867 (free_sese_info): Likewise.
11869 2017-10-18 Martin Liska <mliska@suse.cz>
11872 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11873 on an abnormal edge.
11875 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11877 * doc/invoke.texi (ffunction-sections and fdata-sections):
11880 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11882 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11883 the use statement can throw internally.
11885 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11887 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11888 any RTX present on the RHS of a SET.
11889 * compare-elim.c (try_eliminate_compare): Restore comment.
11891 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11893 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11894 may be also called on tcc_type nodes.
11895 * langhooks.c (lhd_tree_size): Likewise.
11897 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11899 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11900 format_warning_at_substring.
11901 (maybe_warn): Convert source_range * param to a location_t. Pass
11902 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11903 (format_directive): Remove code to extract source_ranges and
11904 source_range * in favor of just a location_t.
11905 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11907 * substring-locations.c (format_warning_va): Convert
11908 source_range * param to a location_t.
11909 (format_warning_at_substring): Likewise.
11910 * substring-locations.h (format_warning_va): Likewise.
11911 (format_warning_at_substring): Likewise.
11913 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11915 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11917 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11918 and scatter/gather ops.
11920 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11921 vec_gather_load and vec_scatter_store.
11922 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11923 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11924 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11925 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11926 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11927 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11929 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11931 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11932 Detect FTST instruction and handle its register pops. Only pop
11933 second operand if can_pop_second_op is true.
11934 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11935 set can_pop_second_op to false in the compare_for_stack_reg call.
11937 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11938 output_fp_compare for stack register operands.
11939 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11940 instructions here. Do not emit stack register pops here. Assert
11941 that FCOMPP pops next to top stack register. Rewrite function.
11943 2017-10-17 Nathan Sidwell <nathan@acm.org>
11945 PR middle-end/82577
11946 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11947 use DECL_ASSEMBLER_NAME_RAW.
11949 PR middle-end/82546
11950 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11953 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11954 Wilco Dijkstra <wilco.dijkstra@arm.com>
11956 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11957 converstion to Pmode from the buf_addr.
11959 2017-10-17 Richard Biener <rguenther@suse.de>
11961 * graphite-dependences.c (scop_get_reads_and_writes): Change
11962 output parameters to references.
11964 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11966 PR 71026/tree-optimization
11967 * fold-const.c (distribute_real_division): Removed.
11968 (fold_binary_loc): Remove calls to distribute_real_divison.
11970 2017-10-17 Richard Biener <rguenther@suse.de>
11972 * graphite-scop-detection.c
11973 (scop_detection::stmt_has_simple_data_refs_p): Always use
11974 the full nest as region.
11975 (try_generate_gimple_bb): Likewise.
11976 * sese.c (scalar_evolution_in_region): Simplify now that
11977 SCEV can handle instantiation in regions.
11978 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11979 in the non-loop part of a function if requested.
11981 2017-10-17 Richard Biener <rguenther@suse.de>
11983 PR tree-optimization/82563
11984 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11986 (graphite_regenerate_ast_isl): Call it.
11987 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11989 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11991 PR tree-optimization/82549
11992 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11993 Formatting fixes. Instead of calling make_bit_field_ref with negative
11996 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11998 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11999 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12000 _mm_maskz_reduce_ss): New.
12001 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12002 __builtin_ia32_reducess_mask): Ditto..
12003 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12004 * config/i386/sse.md (reduces<mode>): Renamed to ...
12005 (reduces<mode><mask_scalar_name>): ... this.
12006 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12008 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12009 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12011 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12013 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12014 * selftest-run-tests.c (selftest::run_tests): Call
12015 selftest::unique_ptr_tests_cc_tests.
12016 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12017 * unique-ptr-tests.cc: New file.
12019 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12022 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12024 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12025 (make_hard_regno_born, make_hard_regno_dead): Update
12026 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12028 2017-10-16 Jeff Law <law@redhat.com>
12030 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12032 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12034 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12036 2017-10-16 Olivier Hainque <hainque@adacore.com>
12038 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12039 with_cpu if we were configured for an e500v2 target cpu name.
12041 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12043 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12044 * doc/invoke.texi: Document +nodsp as a valid extension for
12047 2017-10-16 Martin Liska <mliska@suse.cz>
12049 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12050 (test_set_range): Likewise.
12051 (test_range_functions): Rename to ...
12052 (test_bit_in_range): ... this.
12053 (sbitmap_c_tests): Add new test.
12055 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12057 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12059 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12060 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12062 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12064 * config/aarch64/aarch64-builtins.c
12065 (aarch64_types_quadopu_lane_qualifiers): New.
12066 (TYPES_QUADOPU_LANE): New.
12067 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12068 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12069 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12070 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12071 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12072 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12073 (Vdottype, DOTPROD): New.
12074 (sur): Add SDOT and UDOT.
12076 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12078 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12079 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12080 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12081 Add TARGET_DOTPROD.
12082 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12083 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12084 Enable TARGET_DOTPROD.
12085 (cortex-a75.cortex-a55): Likewise.
12086 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12088 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12090 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12091 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12093 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12095 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12096 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12097 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12098 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12099 * config/arm/types.md (neon_dot, neon_dot_q): New.
12100 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12102 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12104 * config/arm/arm.h (TARGET_DOTPROD): New.
12105 * config/arm/arm.c (arm_arch_dotprod): New.
12106 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12107 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12108 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12109 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12110 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12111 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12112 * doc/invoke.texi (armv8.2-a): Document dotprod
12114 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12116 * i386.c (ix86_vec_cost): New function.
12117 (ix86_rtx_costs): Handle vector operations better.
12118 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12119 * x86-tune-costs.h: Add new costs to all tables.
12121 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12123 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12125 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12126 divsd, sqrtss and sqrtsd
12127 * x86-tune-costs.h: Add new entries to all costs.
12128 (znver1_cost): Fix to match real instruction latencies.
12130 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12131 Michael Collison <michael.collison@arm.com>
12133 * compare-elim.c: Include emit-rtl.h.
12134 (can_merge_compare_into_arith): New function.
12135 (try_validate_parallel): Likewise.
12136 (try_merge_compare): Likewise.
12137 (try_eliminate_compare): Call the above when no previous clobber
12139 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12142 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12144 PR middle-end/62263
12145 PR middle-end/82498
12146 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12147 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12149 PR middle-end/62263
12150 PR middle-end/82498
12151 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12152 to be any operand_equal_p operands. For & (B - 1) require
12153 B to be power of 2. Recognize
12154 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12156 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12159 * optabs.c (expand_memory_blockage): Fix call of
12160 targetm.have_memory_blockage.
12162 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12165 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12166 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12167 objects to extra_objs instead of overwriting it.
12169 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12171 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12172 Use any_fp_register_operand as operand[3] predicate. Simplify
12173 equality test for operands[2] and operands[4] memory location.
12174 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12175 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12176 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12177 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12178 any_fp_register_operand as operand[1] predicate. Simplify
12179 equality test for operands[0] and operands[3] memory location.
12180 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12181 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12182 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12184 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12186 * target-insns.def: Add memory_blockage.
12187 * optabs.c (expand_memory_blockage): New function.
12188 (expand_asm_memory_barrier): Rename ...
12189 (expand_asm_memory_blockage): ... to this.
12190 (expand_mem_thread_fence): Call expand_memory_blockage
12191 instead of expand_asm_memory_barrier.
12192 (expand_mem_singnal_fence): Ditto.
12193 (expand_atomic_load): Ditto.
12194 (expand_atomic_store): Ditto.
12195 * doc/md.texi (Standard Pattern Names For Generation):
12196 Document memory_blockage instruction pattern.
12198 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12200 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12201 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12202 * config/rl78/rl78.md: New define_expand "adddi3".
12204 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12206 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12209 2017-10-13 Jeff Law <law@redhat.com>
12211 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12213 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12216 * internal-fn.c (expand_mul_overflow): If both operands have
12217 the same highpart of -1 or 0 and the topmost bit of lowpart
12218 is different, overflow is if res <= 0 rather than res < 0.
12220 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12222 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12223 TARGET_P9_VECTOR code for unaligned_load case.
12225 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12227 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12229 2017-10-13 Nathan Sidwell <nathan@acm.org>
12231 * tree-core.h (tree_contains_struct): Make bool.
12232 * tree.c (tree_contains_struct): Likewise.
12233 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12234 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12235 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12236 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12237 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12239 2017-10-13 Richard Biener <rguenther@suse.de>
12241 * graphite-isl-ast-to-gimple.c
12242 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12243 parameters and dominance check.
12244 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12245 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12246 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12247 Do not update SSA form here or do intermediate IL verification.
12248 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12249 (graphite_initialize): Remove check on the number of loops in
12250 the function and inline into graphite_transform_loops.
12251 (graphite_finalize): Inline into graphite_transform_loops.
12252 (graphite_transform_loops): Perform SSA update and IL verification
12254 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12256 2017-10-13 Richard Biener <rguenther@suse.de>
12258 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12259 graphite_expression_type_precision): Avoid global constructor
12261 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12262 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12263 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12264 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12265 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12266 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12268 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12271 * config/i386/i386.h (ix86_red_zone_size): New.
12272 * config/i386/i386.md (push peephole2s): Replace
12273 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12275 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12276 Alan Hayward <alan.hayward@arm.com>
12277 David Sherwood <david.sherwood@arm.com>
12279 * combine.c (can_change_dest_mode): Reject changes in
12280 REGMODE_NATURAL_SIZE.
12282 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12283 Alan Hayward <alan.hayward@arm.com>
12284 David Sherwood <david.sherwood@arm.com>
12286 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12287 (expand_debug_source_expr): Likewise.
12288 * combine.c (combine_simplify_rtx): Likewise.
12289 * cse.c (fold_rtx): Likewise.
12290 * fwprop.c (canonicalize_address): Likewise.
12291 * targhooks.c (default_shift_truncation_mask): Likewise.
12293 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12294 Alan Hayward <alan.hayward@arm.com>
12295 David Sherwood <david.sherwood@arm.com>
12297 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12298 (widened_mode): Likewise.
12299 (expand_unop): Likewise.
12300 * ree.c (transform_ifelse): Likewise.
12301 (merge_def_and_ext): Likewise.
12302 (combine_reaching_defs): Likewise.
12303 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12305 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12306 Alan Hayward <alan.hayward@arm.com>
12307 David Sherwood <david.sherwood@arm.com>
12309 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12310 * combine.c (gen_lowpart_for_combine): Likewise.
12311 * dwarf2out.c (rtl_for_decl_location): Likewise.
12312 * final.c (alter_subreg): Likewise.
12313 * rtlhooks.c (gen_lowpart_general): Likewise.
12314 (gen_lowpart_if_possible): Likewise.
12316 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12317 Alan Hayward <alan.hayward@arm.com>
12318 David Sherwood <david.sherwood@arm.com>
12320 * calls.c (expand_call): Use subreg_lowpart_offset.
12321 * cse.c (cse_insn): Likewise.
12322 * regcprop.c (copy_value): Likewise.
12323 (copyprop_hardreg_forward_1): Likewise.
12325 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12328 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12329 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12330 =Q constraints to +Q and into insn condition add check
12331 that operands[0] and operands[1] are equal.
12332 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12333 =Q constraints to +Q and into insn condition add check
12334 that operands[0] is equal to either operands[1] or operands[2].
12337 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12338 instead of handling MINUS_EXPR twice (once for each argument),
12339 canonicalize operand order and handle just once, use rtype where
12340 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12343 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12344 any values of __C while still being pattern recognizable as a simple
12345 rotate instruction.
12347 2017-10-13 Richard Biener <rguenther@suse.de>
12349 PR tree-optimization/82451
12351 2017-10-02 Richard Biener <rguenther@suse.de>
12353 PR tree-optimization/82355
12354 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12355 a mapping for the enclosing loop but avoid generating one for
12356 the loop tree root.
12357 (copy_bb_and_scalar_dependences): Remove premature codegen
12358 error on PHIs in blocks duplicated into multiple places.
12359 * graphite-scop-detection.c
12360 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12361 in the region use it as loop and nest to analyze the DR in.
12362 (try_generate_gimple_bb): Likewise.
12363 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12364 (add_loop_constraints): For blocks in a loop not in the region
12365 create a dimension with a single iteration.
12366 * sese.h (gbb_loop_at_index): Remove assert.
12368 * cfgloop.c (loop_preheader_edge): For the loop tree root
12369 return the single successor of the entry block.
12370 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12371 Reset the SCEV hashtable and niters.
12372 * graphite-scop-detection.c
12373 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12374 assert that we only have POLYNOMIAL_CHREC that vary in loops
12375 contained in the region.
12376 (scop_detection::graphite_can_represent_expr): Adjust.
12377 (scop_detection::stmt_has_simple_data_refs_p): For loops
12378 not in the region set loop to NULL. The nest is now the
12379 entry edge to the region.
12380 (try_generate_gimple_bb): Likewise.
12381 * sese.c (scalar_evolution_in_region): Adjust for
12382 instantiate_scev change.
12383 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12384 Make nest parameter the edge into the region.
12385 (create_data_ref): Likewise.
12386 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12387 entry edge into a region and adjust instantiate_scev calls.
12388 (create_data_ref): Likewise.
12389 (graphite_find_data_references_in_stmt): Likewise.
12390 (find_data_references_in_stmt): Pass the loop preheader edge
12391 from the nest argument.
12392 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12393 parameter the edge into the region.
12394 (instantiate_parameters): Use the loop preheader edge as entry.
12395 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12397 (get_instantiated_value_entry): Make instantiate_below parameter
12398 the edge into the region.
12399 (instantiate_scev_name): Likewise. Adjust dominance checks,
12400 when we cannot use loop-based instantiation instantiate by
12401 walking use-def chains.
12402 (instantiate_scev_poly): Adjust.
12403 (instantiate_scev_binary): Likewise.
12404 (instantiate_scev_convert): Likewise.
12405 (instantiate_scev_not): Likewise.
12406 (instantiate_array_ref): Remove.
12407 (instantiate_scev_3): Likewise.
12408 (instantiate_scev_2): Likewise.
12409 (instantiate_scev_1): Likewise.
12410 (instantiate_scev_r): Do not blindly handle N-operand trees.
12411 Do not instantiate array-refs. Handle all constants and invariants.
12412 (instantiate_scev): Make instantiate_below parameter
12413 the edge into the region.
12414 (resolve_mixers): Use the loop preheader edge for the region
12415 parameter to instantiate_scev_r.
12416 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12418 2017-10-13 Richard Biener <rguenther@suse.de>
12420 PR tree-optimization/82525
12421 * graphite-isl-ast-to-gimple.c
12422 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12424 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12425 Fail code generation when we cannot represent the isl integer.
12426 (binary_op_to_tree): Elide modulo operations that are no-ops
12427 in the type we code generate. Remove now superfluous code
12430 2017-10-13 Richard Biener <rguenther@suse.de>
12432 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12433 (scop_detection::harmful_loop_in_region): Remove premature
12434 IV type restriction.
12435 (scop_detection::graphite_can_represent_scev): We can handle
12436 pointer IVs just fine.
12438 2017-10-13 Alan Modra <amodra@gmail.com>
12440 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12441 "Clobbers and Scratch Registers". Add paragraph on
12442 alternative to clobbers for scratch registers and OpenBLAS
12445 2017-10-13 Alan Modra <amodra@gmail.com>
12447 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12448 example of a memory input for a string of known length. Move
12449 commentary out of table. Add a number of new examples
12450 covering array memory inputs.
12452 2017-10-12 Martin Liska <mliska@suse.cz>
12454 PR tree-optimization/82493
12455 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12456 (test_range_functions): New function.
12457 (sbitmap_c_tests): Likewise.
12458 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12459 * selftest.h (sbitmap_c_tests): New function.
12461 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12463 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12465 * config/rs6000/amo.h: Fix spacing issue.
12467 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12470 * config/i386/i386.md (*ashl<mode>3_mask_1,
12471 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12472 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12475 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12477 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12478 (profile_probability): Set max_probability
12479 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12481 * profile-count.c (profile_probability::differs_from_p): Do not
12482 rely on max_probaiblity == 10000
12484 2017-10-12 Jeff Law <law@redhat.com>
12486 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12489 2017-10-12 Martin Sebor <msebor@redhat.com>
12493 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12494 (handle_alias_pairs): Call it.
12495 * common.opt (-Wattribute-alias): New option.
12496 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12497 * doc/invoke.texi (-Wattribute-alias): Document.
12499 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12502 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12504 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12506 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12507 (make_hard_regno_born, make_hard_regno_dead): Update
12508 bb_killed_pseudos and bb_gen_pseudos.
12510 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12512 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12514 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12516 * config/alpha/alpha.c (alpha_split_conditional_move):
12517 Use std::swap instead of manually swapping.
12518 (alpha_stdarg_optimize_hook): Ditto.
12519 (alpha_canonicalize_comparison): Ditto.
12521 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12523 * tree-loop-distribution.c (struct builtin_info): New struct.
12524 (struct partition): Refactor fields into struct builtin_info.
12525 (partition_free): Free struct builtin_info.
12526 (build_size_arg_loc, build_addr_arg_loc): Delete.
12527 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12528 information from struct builtin_info.
12529 (find_single_drs): New function refactored from classify_partition.
12530 Also moved builtin validity checks to this function.
12531 (compute_access_range, alloc_builtin): New functions.
12532 (classify_builtin_st, classify_builtin_ldst): New functions.
12533 (classify_partition): Refactor code into functions find_single_drs,
12534 classify_builtin_st and classify_builtin_ldst.
12535 (distribute_loop): Don't do runtime alias check when distributing
12537 (find_seed_stmts_for_distribution): New function.
12538 (pass_loop_distribution::execute): Refactor code finding seed
12539 stmts into above function. Support distribution for the innermost
12540 two-level loop nest. Adjust dump information.
12542 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12544 * tree-loop-distribution.c: Adjust the general comment.
12545 (NUM_PARTITION_THRESHOLD): New macro.
12546 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12547 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12548 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12549 in call to build_partition_graph.
12550 (finalize_partitions): New parameter. Make loop distribution more
12551 conservative by fusing more partitions.
12552 (distribute_loop): Don't do runtime alias check in case of loop nest
12554 (find_seed_stmts_for_distribution): New function.
12555 (prepare_perfect_loop_nest): New function.
12556 (pass_loop_distribution::execute): Refactor code finding seed stmts
12557 and loop nest into above functions. Support loop nest distribution.
12558 Adjust dump information accordingly.
12560 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12562 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12563 and set PTYPE_SEQUENTIAL for merged partition.
12565 2017-10-12 Richard Biener <rguenther@suse.de>
12567 PR tree-optimization/69728
12569 2017-09-19 Richard Biener <rguenther@suse.de>
12571 PR tree-optimization/69728
12572 * graphite-sese-to-poly.c (schedule_error): New global.
12573 (add_loop_schedule): Handle empty domain by failing the
12575 (build_original_schedule): Handle schedule_error.
12577 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12578 domain by returning an unchanged schedule.
12580 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12582 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12583 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12585 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12587 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12590 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12593 * expr.c (store_field): Don't optimize away bitsize == 0 store
12594 from CALL_EXPR with addressable return type.
12596 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12598 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12599 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12600 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12601 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12602 TARGET_ISEL instead of TARGET_ISEL<sel>.
12604 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12606 * config/rs6000/rs6000.c
12607 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12609 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12611 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12613 (reg_or_cint_operand): Fix comment.
12614 (reg_or_zero_operand): New predicate.
12615 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12616 * config/rs6000/rs6000.c (output_isel): Delete.
12617 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12618 instead of reg_or_cint_operand. Output instruction directly (not via
12620 (isel_unsigned_<mode>): Ditto.
12621 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12622 gpc_reg_operand. Add an instruction alternative for this. Output
12623 instruction directly.
12624 (*isel_reversed_unsigned_<mode>): Ditto.
12626 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12628 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12629 (TARGET_CANONICALIZE_COMPARISON): Define.
12631 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12634 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12635 Check whether the dest is REG before adding REG_EQUIV note.
12637 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12640 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12642 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12643 (make_hard_regno_born, make_hard_regno_dead): Update
12644 bb_killed_pseudos and bb_gen_pseudos.
12646 2017-10-11 Nathan Sidwell <nathan@acm.org>
12648 * incpath.h (enum incpath_kind): Name enum, prefix values.
12649 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12650 * incpath.c (heads, tails): Use INC_MAX.
12651 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12652 (merge_include_chains, split_quote_chain,
12653 register_include_chains): Update incpath_kind names.
12654 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12655 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12657 (add_framework_path, darwin_register_objc_includes): Likewise.
12658 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12660 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12662 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12663 Do not use float_operator operator predicate.
12664 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12665 * config/i386/predicates.md (float_operator): Remove predicate.
12667 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12669 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12670 (*jccxf_i387): Ditto.
12671 (*jcc<mode>_i387): Ditto.
12672 (*jccu<mode>_i387): Ditto.
12673 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12674 (*jcc_*_i387 splitters): Remove.
12675 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12676 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12677 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12680 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12682 * profile-count.h (slow_safe_scale_64bit): New function.
12683 (safe_scale_64bit): New inline.
12684 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12685 * profile-count.c: Include wide-int.h
12686 (slow_safe_scale_64bit): New.
12688 2017-10-11 Nathan Sidwell <nathan@acm.org>
12690 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12691 HAS_DECL_ASSEMBLER_NAME_P.
12692 * gimple-expr.c (gimple_decl_printable_name: Check
12693 HAS_DECL_ASSEMBLER_NAME_P too.
12694 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12695 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12696 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12697 * passes.c (rest_of_decl_compilation): Check
12698 HAS_DECL_ASSEMBLER_NAME_P too.
12699 * recog.c (verify_changes): Likewise.
12700 * tree-pretty-print.c (dump_decl_name): Likewise.
12701 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12703 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12704 (SET_DECL_ASSEMBLER_NAME): Use it.
12705 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12706 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12707 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12709 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12711 * config.gcc (i386, x86_64): Add extra objects.
12712 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12713 (ix86_min_insn_size): Declare.
12714 (ix86_issue_rate): Declare.
12715 (ix86_adjust_cost): Declare.
12716 (ia32_multipass_dfa_lookahead): Declare.
12717 (ix86_macro_fusion_p): Declare.
12718 (ix86_macro_fusion_pair_p): Declare.
12719 (ix86_bd_has_dispatch): Declare.
12720 (ix86_bd_do_dispatch): Declare.
12721 (ix86_core2i7_init_hooks): Declare.
12722 (ix86_atom_sched_reorder): Declare.
12723 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12724 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12725 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12726 (rip_relative_addr_p): Rename to ...
12727 (ix86_rip_relative_addr_p): ... this one; export.
12728 (memory_address_length): Update.
12729 (ix86_issue_rate): Move to x86-tune-sched.c.
12730 (ix86_flags_dependent): Move to x86-tune-sched.c.
12731 (ix86_agi_dependent): Move to x86-tune-sched.c.
12732 (exact_dependency_1): Move to x86-tune-sched.c.
12733 (exact_store_load_dependency): Move to x86-tune-sched.c.
12734 (ix86_adjust_cost): Move to x86-tune-sched.c.
12735 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12736 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12737 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12738 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12739 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12740 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12741 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12742 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12743 (min_insn_size): Rename to ...
12744 (ix86_min_insn_size): ... this one; export.
12745 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12746 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12747 (core2i7_first_cycle_multipass_backtrack): Move to
12748 x86-tune-sched-core.c.
12749 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12750 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12751 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12752 (ix86_avoid_jump_mispredicts): Update.
12753 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12754 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12755 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12756 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12757 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12758 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12759 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12760 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12761 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12762 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12763 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12764 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12765 (BIG): Move to ix86-tune-sched-bd.c.
12766 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12767 (enum insn_path): Move to ix86-tune-sched-bd.c.
12768 (get_mem_group): Move to ix86-tune-sched-bd.c.
12769 (is_cmp): Move to ix86-tune-sched-bd.c.
12770 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12771 (is_branch): Move to ix86-tune-sched-bd.c.
12772 (is_prefetch): Move to ix86-tune-sched-bd.c.
12773 (init_window): Move to ix86-tune-sched-bd.c.
12774 (allocate_window): Move to ix86-tune-sched-bd.c.
12775 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12776 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12777 (process_end_window): Move to ix86-tune-sched-bd.c.
12778 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12779 (find_constant): Move to ix86-tune-sched-bd.c.
12780 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12781 (has_immediate): Move to ix86-tune-sched-bd.c.
12782 (get_insn_path): Move to ix86-tune-sched-bd.c.
12783 (get_insn_group): Move to ix86-tune-sched-bd.c.
12784 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12785 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12786 (add_insn_window): Move to ix86-tune-sched-bd.c.
12787 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12788 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12789 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12790 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12791 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12792 (do_dispatch): Move to ix86-tune-sched-bd.c.
12793 (has_dispatch): Move to ix86-tune-sched-bd.c.
12794 * config/i386/t-i386: Add new object files.
12795 * config/i386/x86-tune-costs.h: New file.
12796 * config/i386/x86-tune-sched-atom.c: New file.
12797 * config/i386/x86-tune-sched-bd.c: New file.
12798 * config/i386/x86-tune-sched-core.c: New file.
12799 * config/i386/x86-tune-sched.c: New file.
12801 2017-10-11 Liu Hao <lh_mouse@126.com>
12803 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12804 the generic version below instead.
12805 (should_colorize): Recognize Windows consoles as terminals
12807 * pretty-print.c [__MINGW32__] (write_all): New function.
12808 [__MINGW32__] (find_esc_head): Likewise.
12809 [__MINGW32__] (find_esc_terminator): Likewise.
12810 [__MINGW32__] (eat_esc_sequence): Likewise.
12811 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12813 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12816 2017-10-11 Richard Biener <rguenther@suse.de>
12818 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12819 Properly call analyze_scalar_evolution with the loop of the stmt.
12821 2017-10-11 Richard Biener <rguenther@suse.de>
12823 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12824 * tree-core.h (tree_base): Add chrec_var union member.
12825 * tree.h (CHREC_VAR): Remove.
12826 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12827 * tree-chrec.h (build_polynomial_chrec): Adjust.
12828 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12829 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12831 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12833 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12834 * match.pd: ... here.
12835 ((T) X == (T) Y): Relax condition.
12837 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12839 PR tree-optimization/82472
12840 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12842 (break_alias_scc_partitions): Update postorder number.
12844 2017-10-11 Martin Liska <mliska@suse.cz>
12847 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12849 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12851 2017-10-11 Martin Liska <mliska@suse.cz>
12853 * print-rtl.c (print_insn): Move declaration of idbuf
12854 to same scope as name.
12856 2017-10-11 Martin Liska <mliska@suse.cz>
12861 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12862 volatile arguments.
12864 2017-10-11 Martin Liska <mliska@suse.cz>
12867 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12868 volatile arguments.
12870 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12872 * config.gcc (default_gnu_indirect_function): Default to yes for
12873 arm*-*-linux* with glibc.
12875 2017-10-11 Richard Biener <rguenther@suse.de>
12877 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12878 default-defs and types we do not want to analyze.
12879 (interpret_loop_phi): Replace unreachable code with an assert.
12880 (compute_scalar_evolution_in_loop): Remove and inline ...
12881 (analyze_scalar_evolution_1): ... here, replacing condition with
12882 what makes the intent clearer. Remove handling of cases
12883 get_scalar_evolution now handles.
12885 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12887 PR rtl-optimization/81434
12888 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12889 comment for main loop. In sched_group_found if, also add checks for
12890 pass and min_cost_group.
12892 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12894 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12895 (rs6000_insn_cost): New function.
12896 * config/rs6000/rs6000.md (cost): New attribute.
12898 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12899 H.J. Lu <hongjiu.lu@intel.com>
12903 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12904 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12905 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12906 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12907 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12908 OPTION_MASK_ISA_MMX and some other bit set require both
12909 mmx and the ISAs without the mmx bit.
12910 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12911 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12912 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12913 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12914 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12915 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12916 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12917 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12918 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12919 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12920 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12921 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12922 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12923 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12924 Add OPTION_MASK_ISA_MMX.
12926 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12928 * config.gcc (armv7*-*-freebsd*): New target.
12929 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12931 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12933 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12934 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12937 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12939 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12940 a template parameter.
12941 (WIDE_INT_REF_FOR): Update accordingly.
12942 * tree.h (wi::int_traits <const_tree>): Delete.
12943 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12944 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12945 (wi::tree_to_wide_ref): New typedef.
12946 (wi::to_wide): New function.
12947 * calls.c (get_size_range): Use wi::to_wide when operating on
12948 trees as wide_ints.
12949 * cgraph.c (cgraph_node::create_thunk): Likewise.
12950 * config/i386/i386.c (ix86_data_alignment): Likewise.
12951 (ix86_local_alignment): Likewise.
12952 * dbxout.c (stabstr_O): Likewise.
12953 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12954 * expr.c (const_vector_from_tree): Likewise.
12955 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12956 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12957 (fold_negate_expr_1, int_const_binop_1, const_binop)
12958 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12959 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12960 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12961 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12962 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12963 (fold_not_const, round_up_loc): Likewise.
12964 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12965 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12966 (alloca_call_type): Likewise.
12967 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12968 * godump.c (go_output_typedef): Likewise.
12969 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12970 * internal-fn.c (get_min_precision): Likewise.
12971 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12972 * ipa-polymorphic-call.c
12973 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12974 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12975 (ipa_modify_call_arguments): Likewise.
12976 * match.pd: Likewise.
12977 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12978 * print-tree.c (print_node_brief, print_node): Likewise.
12979 * stmt.c (expand_case): Likewise.
12980 * stor-layout.c (layout_type): Likewise.
12981 * tree-affine.c (tree_to_aff_combination): Likewise.
12982 * tree-cfg.c (group_case_labels_stmt): Likewise.
12983 * tree-data-ref.c (dr_analyze_indices): Likewise.
12984 (prune_runtime_alias_test_list): Likewise.
12985 * tree-dump.c (dequeue_and_dump): Likewise.
12986 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12987 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12988 * tree-pretty-print.c (dump_generic_node): Likewise.
12989 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12990 (simple_iv_with_niters): Likewise.
12991 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12992 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12993 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12994 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12995 (refine_value_range_using_guard, number_of_iterations_ne_max)
12996 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12997 (get_cst_init_from_scev, record_nonwrapping_iv)
12998 (scev_var_range_cant_overflow): Likewise.
12999 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13000 * tree-ssa-pre.c (compute_avail): Likewise.
13001 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13002 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13003 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13004 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13005 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13006 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13007 (dump_case_nodes, try_switch_expansion): Likewise.
13008 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13009 (vect_do_peeling): Likewise.
13010 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13011 * tree-vect-stmts.c (vectorizable_load): Likewise.
13012 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13013 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13014 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13015 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13016 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13017 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13018 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13019 (evrp_dom_walker::before_dom_children): Likewise.
13020 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13021 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13022 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13023 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13024 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13025 (get_range_pos_neg): Likewise.
13026 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13027 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13028 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13029 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13030 * config/avr/avr.c (avr_fold_builtin): Likewise.
13031 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13032 * config/msp430/msp430.c (msp430_attr): Likewise.
13033 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13034 * config/powerpcspe/powerpcspe-c.c
13035 (altivec_resolve_overloaded_builtin): Likewise.
13036 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13037 (rs6000_expand_ternop_builtin): Likewise.
13038 * config/rs6000/rs6000-c.c
13039 (altivec_resolve_overloaded_builtin): Likewise.
13040 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13041 (rs6000_expand_ternop_builtin): Likewise.
13042 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13044 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13046 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13047 when copying loop nest with only one inner loop.
13049 2017-10-10 Richard Biener <rguenther@suse.de>
13051 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13052 blocks if SCEV is active.
13053 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13055 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13056 (scev_initialize): Assert we are not yet initialized.
13058 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13060 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13061 inner loop's exit stmt by making it always exit the loop, otherwise
13062 we would generate an infinite empty loop.
13064 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13066 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13067 renaming variables in new preheader if it's deleted.
13069 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13071 * tree-loop-distribution.c (struct partition): Remove unused field
13072 loops of the structure.
13073 (partition_alloc, partition_free): Ditto.
13074 (build_rdg_partition_for_vertex): Ditto.
13076 2017-10-09 Jeff Law <law@redhat.com>
13078 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13079 return type to match prototype and documentation.
13081 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13083 * config/rs6000/rs6000.c (processor_costs): Move to ...
13084 * config/rs6000/rs6000.h: ... here.
13085 (rs6000_cost): Declare.
13087 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13089 * except.c (setjmp_fn): New global variable.
13090 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13091 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13092 if DONT_USE_BUILTIN_SETJMP is defined.
13094 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13096 * target.def (insn_cost): New hook.
13097 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13098 * doc/tm.texi: Regenerate.
13099 * rtlanal.c (insn_cost): Use the new hook.
13101 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13103 * combine.c (combine_validate_cost): Compute the new insn_cost,
13104 not just pattern_cost.
13105 (try_combine): Adjust comment.
13107 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13109 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13111 * combine.c (uid_insn_cost): Adjust comment.
13112 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13114 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13115 * dse.c (find_shift_sequence): Ditto.
13116 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13117 (bb_valid_for_noce_process_p): Use pattern_cost.
13118 * rtl.h (insn_rtx_cost): Delete.
13119 (pattern_cost): New prototype.
13120 (insn_cost): New prototype.
13121 * rtlanal.c (insn_rtx_cost): Rename to...
13122 (pattern_cost): ... this.
13125 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13127 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13128 (*jcc<mode>_0_r_i387): Ditto.
13129 (*jccxf_r_i387): Ditto.
13130 (*jcc<mode>_r_i387): Ditto.
13131 (*jccu<mode>_r_i387): Ditto.
13132 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13133 (*jcc): Rename from *jcc_1.
13135 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13137 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13138 deferred rescans after the lvx/stvx recombination pre-pass.
13140 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13142 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13143 memory operation instruction support.
13144 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13145 (rs6000-ibm-aix[789]*): Likewise.
13146 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13147 Document new functions.
13149 2017-10-09 Richard Biener <rguenther@suse.de>
13151 PR tree-optimization/82397
13152 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13153 equality only for semantically equal trees.
13155 2017-10-09 Richard Biener <rguenther@suse.de>
13157 PR tree-optimization/82449
13158 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13159 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13160 allow constant addresses.
13161 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13164 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13166 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13169 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13172 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13175 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13178 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13180 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13183 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13184 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13186 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13188 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13189 (WI_BINARY_PREDICATE_RESULT): Likewise.
13190 (wi::binary_traits::operator_result): New type.
13191 (wi::binary_traits::predicate_result): Likewise.
13192 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13193 (generic_wide_int::operator==, generic_wide_int::operator!=)
13194 (generic_wide_int::operator&, generic_wide_int::and_not)
13195 (generic_wide_int::operator|, generic_wide_int::or_not)
13196 (generic_wide_int::operator^, generic_wide_int::operator+
13197 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13199 (operator~, unary operator-, operator==, operator!=, operator&)
13200 (operator|, operator^, operator+, binary operator-, operator*): New
13202 * expr.c (get_inner_reference): Use wi::bit_and_not.
13203 * fold-const.c (fold_binary_loc): Likewise.
13204 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13205 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13206 (bit_value_binop): Likewise.
13207 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13208 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13209 (extract_range_from_binary_expr_1): Likewise.
13210 (masked_increment): Likewise.
13211 (simplify_bit_ops_using_ranges): Likewise.
13213 2017-10-09 Martin Jambor <mjambor@suse.cz>
13216 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13217 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13218 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13220 (hsa_fixup_mov_insn_type): New function.
13221 (hsa_op_with_type::get_in_type): Use it.
13222 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13223 immediates in an assert.
13224 (hsa_op_with_type::extend_int_to_32bit): New method.
13225 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13226 types. Convert to dest type if necessary.
13227 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13228 (reg_for_gimple_ssa): Pass false as min32int to
13229 hsa_type_for_scalar_tree_type.
13230 (gen_hsa_addr): Fixup type when creating addresable temporary.
13231 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13232 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13233 necessary. Call hsa_fixup_mov_insn_type.
13234 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13235 extend operands and convert to dest type if necessary.
13236 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13237 to dest type if necessary.
13238 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13239 if conversion nt necessary and size matches.
13240 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13241 to dest type if necessary.
13242 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13243 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13245 (gen_hsa_clrsb): Likewise.
13246 (gen_hsa_ffs): Likewise.
13247 (gen_hsa_divmod): Extend operands and convert to dest type if
13249 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13251 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13253 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13254 Remove empty default arguments. Use a brace block as output
13256 (conditional return): Ditto.
13258 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13259 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13261 (group_ending_nop): Ditto.
13262 (doloop_end): Ditto.
13263 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13264 (splitters for those): Ditto.
13266 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13268 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13269 a conditional jump (and the compare for it) so that pc_rtx is the
13271 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13272 for the deleted and renamed ctr<mode>_internal[234] patterns.
13273 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13274 Delete second conditional return pattern.
13275 (ctr<mode>_internal2): Delete this second bdnz pattern.
13276 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13277 (ctr<mode>_internal4): Delete this second bdz pattern.
13279 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13281 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13282 (always_initialized_rtx_for_ssa_name_p): New predicate.
13283 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13284 (finish_out_of_ssa): Free new field of SA.
13285 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13286 * tree-ssa-coalesce.c: Include tree-ssa.h.
13287 (get_parm_default_def_partitions): Remove extern keyword.
13288 (get_undefined_value_partitions): New function.
13289 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13290 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13291 uninitialized bits.
13292 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13294 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13296 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13298 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13300 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13301 for targets that preffer 128bit.
13303 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13305 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13307 2017-10-08 Olivier Hainque <hainque@adacore.com>
13309 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13310 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13312 (thumb_set_return_address): Likewise.
13314 2017-10-08 Olivier Hainque <hainque@adacore.com>
13316 * common/config/arm/arm-common.c (arm_except_unwind_info):
13317 Handle DWARF2_UNWIND_INFO.
13319 2017-10-07 Michael Collison <michael.collison@arm.com>
13321 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13324 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13326 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13327 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13328 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13329 defined, force the creation of a new block for a dispatch label.
13331 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13333 * invoke.texi (Wsuggest-attribute=cold): Document.
13334 * common.opt (Wsuggest-attribute=cold): New
13335 * ipa-pure-const.c (warn_function_cold): New function.
13336 * predict.c (compute_function_frequency): Use it.
13337 * predict.h (warn_function_cold): Declare.
13339 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13341 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13344 2017-10-06 Martin Liska <mliska@suse.cz>
13346 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13347 keyword for member functions.
13348 (struct sanopt_tree_couple): New struct.
13349 (struct sanopt_tree_couple_hash): New function.
13350 (struct sanopt_ctx): Add new hash_map.
13351 (has_dominating_ubsan_ptr_check): New function.
13352 (record_ubsan_ptr_check_stmt): Likewise.
13353 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13354 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13355 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13357 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13360 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13361 aarch64_simd_valid_immediate on CONST_VECTORs.
13362 (aarch64_reg_or_bic_imm): Likewise.
13364 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13366 PR rtl-optimization/82396
13367 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13369 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13371 * graphite-dependences.c (scop_get_reads): Move code to...
13372 (scop_get_must_writes): Move code to...
13373 (scop_get_may_writes): Move code to...
13374 (scop_get_reads_and_writes): ... here.
13375 (scop_get_dependences): Call scop_get_reads_and_writes.
13377 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13379 PR tree-optimization/82434
13380 * fold-const.h (can_native_encode_type_p,
13381 can_native_encode_string_p): Remove.
13382 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13383 don't encode anything, just return what would be otherwise returned.
13384 (native_encode_fixed, native_encode_complex, native_encode_vector):
13386 (native_encode_string): Likewise. Inline by hand
13387 can_native_encode_string_p.
13388 (can_native_encode_type_p): Remove.
13389 (can_native_encode_string_p): Remove.
13390 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13391 STRING_CSTs using can_native_encode_string_p, test all
13392 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13393 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13394 argument from native_encode_expr.
13395 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13396 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13399 2017-10-06 Richard Biener <rguenther@suse.de>
13401 PR tree-optimization/82397
13402 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13403 operand_equal_p but rely on data_ref_compare_tree for detecting
13405 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13406 to match up with dr_group_sort_cmp.
13408 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13411 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13413 * config/s390/s390-builtin-types.def: Regenerate.
13415 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13418 * config/s390/s390-builtin-types.def: Regenerate.
13419 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13420 Change flag from B_VXE to B_VX.
13421 (s390_vec_min_dbl): Remove B_VXE flag.
13423 2017-10-06 Richard Biener <rguenther@suse.de>
13425 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13426 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13427 translate_isl_ast_to_gimple::is_valid_rename,
13428 translate_isl_ast_to_gimple::get_rename,
13429 translate_isl_ast_to_gimple::get_def_bb_for_const,
13430 translate_isl_ast_to_gimple::get_new_name,
13431 translate_isl_ast_to_gimple::collect_all_ssa_names,
13432 translate_isl_ast_to_gimple::copy_loop_phi_args,
13433 translate_isl_ast_to_gimple::collect_all_ssa_names,
13434 translate_isl_ast_to_gimple::copy_loop_phi_args,
13435 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13436 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13437 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13438 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13439 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13440 translate_isl_ast_to_gimple::copy_cond_phi_args,
13441 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13442 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13443 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13444 translate_isl_ast_to_gimple::rename_uses,
13445 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13446 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13447 (set_rename_for_each_def): Likewise.
13448 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13449 here. Handle rewriting SCEV analyzable uses here.
13450 (copy_bb_and_scalar_dependences): Generate code for PHI
13452 (graphite_regenerate_ast_isl): Adjust.
13453 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13454 (add_write, add_read): New functions.
13455 (build_cross_bb_scalars_def): Use it and simplify.
13456 (build_cross_bb_scalars_use): Likewise.
13457 (graphite_find_cross_bb_scalar_vars): Inline into...
13458 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13459 simulating out-of-SSA. Compute liveout and add dependencies.
13460 (build_scops): Force an empty entry block.
13461 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13463 (sese_build_liveouts): Declare.
13464 (sese_trivially_empty_bb_p): Likewise.
13465 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13466 compute liveout and debug_liveout.
13467 (sese_bad_liveouts_use): Remove.
13468 (sese_reset_debug_liveouts_bb): Likewise.
13469 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13470 (sese_build_liveouts): Build liveout and debug_liveout and store
13472 (new_sese_info): Adjust.
13473 (free_sese_info): Likewise.
13474 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13475 do not build liveout here.
13476 (move_sese_in_condition): Adjust region entry.
13477 (scev_analyzable_p): Match up with chrec_apply requirements.
13478 (sese_trivially_empty_bb_p): New.
13479 * tree-into-ssa.c (get_reaching_def): Properly support generating
13480 default-defs for incremental rewrite of anonymous names.
13482 2017-10-06 Richard Biener <rguenther@suse.de>
13484 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13485 precision do not perform modulo reduction.
13487 2017-10-06 Richard Biener <rguenther@suse.de>
13489 PR tree-optimization/82436
13490 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13491 conservatively choose the vectorization factor when checking
13492 whether we can perform the required load permutation.
13493 (vect_transform_slp_perm_load): Assert when we may not fail.
13495 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13497 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13498 message for incompatible -msdata=* and -mcall-* options.
13500 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13502 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13503 rate for post-reload scheduling.
13505 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13507 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13509 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13511 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13512 to improve monte carlo in scimark.
13514 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13516 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13517 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13518 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13519 pentium4_cost, nocona_cost): Set reassociation width to 1.
13520 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13521 width to 2 for fp operations and 1 otherwise.
13522 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13524 (atom_cost): Set reassociation width to 2.
13525 (slm_cost, generic_cost): Set fp reassociation width
13526 to 2 and 1 otherwise.
13527 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13528 (core_cost): Set fp reassociation width to 4 and vector to 2.
13529 (ix86_reassociation_width): Rewrite using cost table; special case
13530 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13531 and TARGET_AVX128_OPTIMAL.
13532 * config/i386/i386.h (processor_costs): Add
13533 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13534 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13535 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13536 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13537 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13538 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13540 2017-10-05 Nathan Sidwell <nathan@acm.org>
13542 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13544 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13546 * config/arm/arm.c (arm_test_fpu_data): New.
13547 (arm_run_selftests): Call arm_test_fpu_data.
13549 2017-10-04 Nathan Sidwell <nathan@acm.org>
13551 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13553 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13555 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13557 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13558 check type for aarch64_simd_valid_immediate.
13559 (aarch64_output_simd_mov_immediate): Update prototype.
13560 (aarch64_simd_valid_immediate): Update prototype.
13561 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13562 support for ORR-immediate.
13563 (and<mode>3): modified pattern to add support for BIC-immediate.
13564 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13565 now checks for valid immediate for BIC and ORR based on new enum
13567 (aarch64_output_simd_mov_immediate): Function now used to output
13568 BIC/ORR imm as well based on new enum argument.
13569 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13571 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13572 (aarch64_reg_or_bic_imm): Likewise.
13574 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13576 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13577 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13579 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13583 PR rtl-optimization/82396
13584 * haifa-sched.c (autopref_multipass_init): Simplify
13586 (autopref_rank_data): Simplify sort order.
13587 * sched-int.h (autopref_multipass_data_): Remove
13588 multi_mem_insn_p, min_offset and max_offset.
13590 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13592 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13594 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13596 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13597 vect_doubleint_cvt.
13599 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13601 * doc/sourcebuild.texi: Document vect_long_mult.
13603 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13605 PR tree-optimization/82413
13606 * fold-const.c (build_range_check): Use widest_int when comparing
13607 the maximum ETYPE value with HIGH.
13609 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13611 PR rtl-optimization/82396
13612 * haifa-sched.c (autopref_multipass_init): Simplify
13614 (autopref_rank_data): Simplify sort order.
13615 * sched-int.h (autopref_multipass_data_): Remove
13616 multi_mem_insn_p, min_offset and max_offset.
13618 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13620 PR tree-optimization/82381
13621 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13622 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13625 PR tree-optimization/82374
13626 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13627 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13628 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13629 current_function_decl to the new decl.
13631 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13633 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13634 helper macro for IEEE float128 hardware built-in functions.
13635 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13637 (TRUNCF128_ODD): Likewise.
13638 (ADDF128_ODD): Likewise.
13639 (SUBF128_ODD): Likewise.
13640 (MULF128_ODD): Likewise.
13641 (DIVF128_ODD): Likewise.
13642 (FMAF128_ODD): Likewise.
13643 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13644 UNSPEC_TRUNC_ROUND_TO_ODD.
13645 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13646 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13647 floating point round to odd instructions.
13648 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13649 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13650 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13651 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13652 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13653 (trunc<mode>sf2_hw): Change the truncate with round to odd
13654 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13655 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13656 to odd hardware instructions.
13657 (sub<mode>3_odd): Likewise.
13658 (mul<mode>3_odd): Likewise.
13659 (div<mode>3_odd): Likewise.
13660 (sqrt<mode>2_odd): Likewise.
13661 (fma<mode>4_odd): Likewise.
13662 (fms<mode>4_odd): Likewise.
13663 (nfma<mode>4_odd): Likewise.
13664 (nfms<mode>4_odd): Likewise.
13665 (trunc<mode>df2_odd): Change the truncate with round to odd
13666 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13668 * doc/extend.texi (PowerPC built-in functions): Update documentation
13669 for existing IEEE float128-bit built-in functions. Add built-in
13670 functions that generate the IEEE 128-bit floating point round to
13673 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13675 PR rtl-optimization/77729
13676 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13677 to (X&(C1&~C2))|C2 transformations.
13679 2017-10-03 Martin Jambor <mjambor@suse.cz>
13681 PR tree-optimization/82363
13682 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13683 mismatch, mark lacc written regardless of racc.
13685 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13687 PR tree-optimization/82381
13688 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13689 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13690 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13691 fallthrough into reassoc_stmt_dominates_stmt_p.
13694 * combine.c (combine_instructions): Don't combine in unreachable
13697 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13701 function to not use the have_cpu variable. Do not set cpu_index,
13702 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13703 or the default cpu.
13704 (rs6000_valid_attribute_p): Remove duplicate initializations of
13705 old_optimize and func_optimize.
13706 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13707 (rs6000_activate_target_options): Make global.
13708 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13711 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13713 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13714 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13715 Set *poffset to 0 and *psize and *pmax_size to -1 if
13716 *poffset + *psize overflows in HOST_WIDE_INT.
13718 PR tree-optimization/82387
13719 PR tree-optimization/82388
13720 PR tree-optimization/82389
13721 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13722 instead of live_bytes non-NULL.
13724 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13727 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13730 2017-10-02 Richard Biener <rguenther@suse.de>
13732 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13733 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13734 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13736 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13738 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13739 requested precision matches the type's.
13740 * calls.c (alloc_max_size): Calculate the new candidate size as
13741 a widest_int and use wi::to_widest when comparing it with the
13742 current candidate size.
13743 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13744 zero rather than integer_zero_node.
13745 * match.pd: Check for a no-op conversion before using wi::add
13746 rather than after. Use tree_to_uhwi when summing small shift
13747 counts into an unsigned int.
13749 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13750 Alan Hayward <alan.hayward@arm.com>
13751 David Sherwood <david.sherwood@arm.com>
13754 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13755 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13756 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13757 POINTER_AND_FP_REGS.
13759 2017-10-02 Richard Biener <rguenther@suse.de>
13761 PR tree-optimization/82355
13762 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13763 a mapping for the enclosing loop but avoid generating one for
13764 the loop tree root.
13765 (copy_bb_and_scalar_dependences): Remove premature codegen
13766 error on PHIs in blocks duplicated into multiple places.
13767 * graphite-scop-detection.c
13768 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13769 in the region use it as loop and nest to analyze the DR in.
13770 (try_generate_gimple_bb): Likewise.
13771 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13772 (add_loop_constraints): For blocks in a loop not in the region
13773 create a dimension with a single iteration.
13774 * sese.h (gbb_loop_at_index): Remove assert.
13776 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13778 * omp-expand.c (adjust_context_scope): New function.
13779 (expand_parallel_call): Call adjust_context_scope.
13781 2017-10-01 Jeff Law <law@redhat.com>
13783 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13784 dom_opt_dom_walker class with direct access to private members.
13785 Add comments. Call test_for_singularity.
13786 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13787 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13788 m_dummy_cond anymore.
13789 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13791 (pass_dominator:execute): Build the dummy_cond here and pass it
13792 to the dom_opt_dom_walker ctor.
13793 (test_for_singularity): New function.
13795 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13796 Maya Rashish <coypu@sdf.org>
13798 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13799 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13800 (alpha*-*-netbsd*) Use nbsd_tm_file.
13801 (arm*-*-netbsdelf*) Likewise.
13802 (i[34567]86-*-netbsdelf*) Likewise.
13803 (x86_64-*-netbsd*) Likewise.
13804 (mips*-*-netbsd*) Likewise.
13805 (powerpc-*-netbsd*) Likewise.
13806 (sh*-*-netbsd*) Likewise.
13807 (sparc-*-netbsdelf*) Likewise.
13808 (sparc64-*-netbsd*) Likewise.
13809 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13811 (vax-*-netbsdelf*) Likewise.
13812 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13813 (UINT_FAST8_TYPE) Likewise.
13814 (INT_FAST16_TYPE) Check CHAR_FAST16.
13815 (UINT_FAST16_TYPE) Likewise.
13817 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13820 * config/i386/i386.md
13821 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13822 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13823 *divmodsi4_zext_2): New define_insn_and_split.
13824 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13825 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13826 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13827 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13828 New define_insn_and_split.
13829 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13830 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13831 operands[1] having DImode when mode is SImode.
13833 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13834 always SImode for DIV and MOD in REG_EQUAL notes.
13836 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13838 PR middle-end/82319
13839 * match.pd: Fix handling of NaNs in pattern.
13841 2017-09-29 Jeff Law <law@redhat.com>
13843 * sbitmap.c (bitmap_bit_in_range_p): New function.
13844 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13845 * tree-ssa-dse.c (live_bytes_read): New function.
13846 (dse_classify_store): Ignore reads of dead bytes.
13848 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13849 typos and whitespace errors.
13850 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13851 * config/s390/s390.c (s390_emit_prologue): Likewise.
13853 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13856 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13857 with a symbol for LRA.
13859 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13861 PR rtl-optimization/82338
13862 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13864 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13866 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13867 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13868 (qsort_chk): Declare.
13869 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13870 (qsort_chk): New function.
13872 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13874 PR tree-optimization/82337
13875 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13876 phi definition if the PHI result appears in an abnormal PHI.
13877 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13878 basis appears in an abnormal PHI.
13880 2017-09-29 Richard Biener <rguenther@suse.de>
13882 * graphite-isl-ast-to-gimple.c
13883 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13884 (binary_op_to_tree): Use it.
13885 (get_rename_from_scev): Likewise.
13886 (copy_loop_phi_nodes): Likewise.
13887 (copy_bb_and_scalar_dependences): Likewise.
13888 (translate_pending_phi_nodes): Likewise.
13890 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13893 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13894 for movabsq $(i32 << shift), r64.
13896 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13898 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13899 index when encoding %esp as %rsp to avoid 0x67 prefix.
13901 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13903 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13904 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13906 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13908 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13909 Extensions with more than 16 double VFP registers.
13910 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13911 to_clear_mask and all code related to it. Replace the remaining
13912 entry by a sbitmap and adapt code accordingly.
13914 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13916 * brig-builtins.def: Change pure attributes to const.
13918 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13920 * config.gcc (default_gnu_indirect_function): Default to yes for
13921 sparc*-*-linux* with glibc.
13923 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13925 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13926 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13927 when creating .init_array and .fini_array sections with priority
13930 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13933 * config/aarch64/aarch64.c
13934 (aarch64_builtin_support_vector_misalignment): Always return false
13935 when misalignment is unknown.
13937 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13939 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13940 this function to return false if the definition used by the swap
13941 instruction is artificial, or if the memory address from which the
13942 constant value is loaded is not represented by a base address held
13943 in a register or if the base address register is a frame or stack
13944 pointer. Additionally, return false if the base address of the
13945 loaded constant is a SYMBOL_REF but is not considered to be a
13947 (replace_swapped_load_constant): New function.
13948 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13949 loaded constant vector with a load of a swapped constant vector.
13951 2017-09-27 Carl Love <cel@us.ibm.com>
13953 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13954 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13955 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13958 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13960 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13961 first, always call autopref_rank_data otherwise.
13963 2017-09-27 Richard Biener <rguenther@suse.de>
13965 * graphite-scop-detection.c (find_scop_parameters): Move
13966 loop bound handling ...
13967 (gather_bbs::before_dom_children): ... here, avoiding the need
13968 to build scop_info->loop_nest.
13969 (record_loop_in_sese): Remove.
13970 * sese.h (sese_info_t::loop_nest): Remove.
13971 * sese.c (new_sese_info): Do not allocate loop_nest.
13972 (free_sese_info): Do not free loop_nest.
13974 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13977 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13978 lhs from calls if the lhs has addressable type.
13980 2017-09-27 Richard Biener <rguenther@suse.de>
13982 * graphite.h (scop::max_alias_set): New member.
13983 * graphite-scop-detection.c: Remove references to non-existing
13984 --param in comments.
13985 (build_alias_sets): Record the maximum alias set used for drs.
13986 (build_scops): Support zero as unlimited for
13987 --param graphite-max-arrays-per-scop.
13988 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13989 and inline into ...
13990 (build_poly_sr_1): ... here. Compute alias set based on the
13991 maximum alias set used for drs rather than
13992 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13994 2017-09-27 Richard Biener <rguenther@suse.de>
13996 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13997 --param loop-block-tile-size=0 to disable tiling.
13999 2017-09-27 Richard Biener <rguenther@suse.de>
14001 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14002 (graphite-max-nb-scop-params): Document special value zero.
14003 * domwalk.h (dom_walker::STOP): New symbolical constant.
14004 (dom_walker::dom_walker): Add optional parameter for bb to
14006 (dom_walker::~dom_walker): Declare.
14007 (dom_walker::before_dom_children): Document STOP return value.
14008 (dom_walker::m_user_bb_to_rpo): New member.
14009 (dom_walker::m_bb_to_rpo): Likewise.
14010 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14011 mapping here if not provided by the user.
14012 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14013 provided by the user.
14014 (dom_walker::STOP): Define.
14015 (dom_walker::walk): Do not compute bb to RPO mapping here.
14016 Support STOP return value from before_dom_children to stop
14018 * graphite-optimize-isl.c (optimize_isl): If the schedule
14019 is the same still generate code if -fgraphite-identity
14020 or -floop-parallelize-all are given.
14021 * graphite-scop-detection.c: Include cfganal.h.
14022 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14024 (gather_bbs::before_dom_children): Return STOP for BBs
14026 (build_scops): Compute bb to RPO mapping and pass it to
14027 the domwalk. Treat --param graphite-max-nb-scop-params=0
14028 as not limiting the number of params.
14029 * graphite.c (graphite_initialize): Remove limit on the
14030 number of basic-blocks in a function.
14031 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14032 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14033 default value of 10.
14035 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14037 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14038 Adjust code to eliminate needing to do the shift right 32-bits
14039 operation after XSCVDPSPN.
14041 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14043 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14044 ((X / Y) != 0 -> X >= Y): Likewise.
14046 2017-09-26 Carl Love <cel@us.ibm.com>
14048 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14049 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14050 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14051 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14052 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14053 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14054 definitions and overloading.
14055 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14056 statement for P9V_BUILTIN_XST_LEN_R.
14057 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14058 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14059 define_expand and define_insn for the instructions and builtins.
14060 * doc/extend.texi: Update the built-in documentation file for the new
14061 built-in functions.
14062 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14063 define_insn for the instructions
14065 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14068 * gcc/config/netbsd-protos.h: New file.
14069 * gcc/config/netbsd.c: New file.
14070 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14071 * gcc/config/t-netbsd: New file.
14072 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14073 (tmake_file) Add t-netbsd.
14074 (extra_objs) Add netbsd.o.
14076 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14080 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14082 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14084 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14085 sign extension from a vector register to a GPR by doing a 32-bit
14086 direct move and then an EXTSW.
14087 (extendsi<mode>2 splitter): Likewise.
14088 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14089 right or vector extract after doing XSCVDPSPN. Use
14090 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14092 (movdi_from_sf_zero_ext): Likewise.
14093 (reload_gpr_from_vsxsf): Likewise.
14094 (p8_mfvsrd_4_disf): Delete, no longer used.
14095 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14096 then needing to move the SFmode to a GPR to use the XSCVDPSP
14097 instruction instead of FRSP and XSCVDPSPN.
14098 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14099 it is adjacent to the other XSCVSPDP insns.
14100 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14101 SFmode to be in traditional Altivec registers.
14102 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14103 (vsx_xscvspdpn): Likewise.
14104 (vsx_xscvspdpn_scalar): Likewise.
14106 2017-09-26 Martin Jambor <mjambor@suse.cz>
14108 * tree-sra.c (compare_access_positions): Put integral types first,
14109 stabilize sorting of integral types, remove conditions putting
14110 non-full-precision integers last.
14111 (sort_and_splice_var_accesses): Disable scalarization if a
14112 non-integert would be represented by a non-full-precision integer.
14114 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14116 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14117 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14118 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14119 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14120 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14121 conditionals inside the function instead of around it. Call
14122 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14123 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14125 2017-09-26 Richard Biener <rguenther@suse.de>
14127 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14129 (scop_detection::build_scop_breadth): ... this. Removed.
14130 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14131 (scop_detection::harmful_stmt_in_bb): Likewise.
14132 (scop_detection::graphite_can_represent_stmt): Likewise.
14133 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14134 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14135 (scop_detection::can_represent_loop_1): ... this. Removed.
14136 (scop_detection::harmful_loop_in_region): Simplify after inlining
14137 the above and remove more quadraticness.
14138 (build_scops): Adjust.
14139 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14142 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14145 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14146 REGNO (base) == SP_REG if base is a REG.
14148 PR middle-end/35691
14149 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14150 if it is different SSA_NAME.
14151 (optimize_range_tests_cmp_bitwise): New function.
14152 (optimize_range_tests): Call it.
14154 2017-09-26 Richard Biener <rguenther@suse.de>
14156 PR tree-optimization/82321
14157 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14158 for the def being inside the loop.
14160 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14162 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14164 * config/s390/s390-builtins.def: Fix constraint on op4.
14166 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14168 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14169 independent expanders.
14170 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14171 ("vec_ordered", "vec_unordered"): New expanders.
14173 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14175 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14178 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14180 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14181 vec_unpacks_lo_v16qi.
14182 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14184 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14186 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14187 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14188 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14190 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14192 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14194 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14196 ("*vec_slb<mode>"): New insn pattern.
14197 ("vec_shr_<mode>"): New expander.
14198 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14199 and force the shift count operand to V16QImode.
14200 ("vec_srb<mode>"): Set shift count mode to V16QI.
14202 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14204 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14205 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14206 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14208 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14211 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14213 2017-09-26 Richard Biener <rguenther@suse.de>
14215 PR tree-optimization/82320
14216 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14219 2017-09-25 Jeff Law <law@redhat.com>
14221 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14222 prototype for new argument.
14223 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14224 mostly extracted from rs6000_emit_allocate_stack.
14225 (rs6000_emit_probe_stack_range_stack_clash): New function.
14226 (rs6000_emit_allocate_stack): Call
14227 rs6000_emit_probe_stack_range_stack_clash as needed.
14228 (rs6000_emit_probe_stack_range): Add additional argument
14229 to call to gen_probe_stack_range{si,di}.
14230 (output_probe_stack_range): New.
14231 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14232 (output_probe_stack_range_stack_clash): New.
14233 (rs6000_emit_prologue): Emit notes into dump file as requested.
14234 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14235 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14236 Add additional operand and pass it to output_probe_stack_range.
14238 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14240 PR tree-optimization/82163
14241 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14242 (checking_verify_loop_closed_ssa): New parameter.
14243 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14244 (check_loop_closed_ssa_stmt): Delete.
14245 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14246 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14247 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14250 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14252 * brig-builtins.def: Treat HSAIL barrier builtins as
14253 setjmp/longjump style functions.
14255 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14257 * target.def (constant_alignment): New hook.
14258 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14259 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14260 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14261 * doc/tm.texi: Regenerate.
14262 * targhooks.h (default_constant_alignment): Declare.
14263 (constant_alignment_word_strings): Likewise.
14264 * targhooks.c (default_constant_alignment): New function.
14265 (constant_alignment_word_strings): Likewise.
14266 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14267 instead of CONSTANT_ALIGNMENT.
14268 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14269 (force_const_mem): Likewise.
14270 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14271 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14272 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14273 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14274 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14276 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14277 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14278 constant_alignment_word_strings.
14279 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14280 (CONSTANT_ALIGNMENT): Likewise.
14281 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14282 (arm_constant_alignment): New function.
14283 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14284 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14285 constant_alignment_word_strings.
14286 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14287 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14288 constant_alignment_word_strings.
14289 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14290 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14291 (cris_constant_alignment): New function.
14292 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14293 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14294 (epiphany_constant_alignment): New function.
14295 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14296 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14297 constant_alignment_word_strings.
14298 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14299 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14300 constant_alignment_word_strings.
14301 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14302 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14303 constant_alignment_word_strings.
14304 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14305 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14306 * config/i386/i386.c (ix86_constant_alignment): Make static.
14307 Use the same interface as the target hook.
14308 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14309 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14310 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14311 constant_alignment_word_strings.
14312 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14313 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14314 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14315 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14316 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14317 constant_alignment_word_strings.
14318 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14319 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14320 constant_alignment_word_strings.
14321 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14322 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14323 constant_alignment_word_strings.
14324 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14325 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14327 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14328 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14329 * config/mips/mips.c (mips_constant_alignment): New function.
14330 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14331 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14332 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14333 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14334 (mmix_constant_alignment): Make static. Use the same interface
14335 as the target hook.
14336 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14337 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14338 constant_alignment_word_strings.
14339 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14340 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14341 constant_alignment_word_strings.
14342 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14343 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14344 constant_alignment_word_strings.
14345 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14346 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14347 (rs6000_constant_alignment): New function.
14348 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14349 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14350 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14351 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14352 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14353 (rs6000_constant_alignment): New function.
14354 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14355 * config/s390/s390.c (s390_constant_alignment): New function.
14356 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14357 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14358 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14359 constant_alignment_word_strings.
14360 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14361 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14362 (sparc_constant_alignment): New function.
14363 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14364 * config/spu/spu.c (spu_constant_alignment): New function.
14365 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14366 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14367 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14368 constant_alignment_word_strings.
14369 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14370 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14371 constant_alignment_word_strings.
14372 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14373 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14374 constant_alignment_word_strings.
14375 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14376 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14377 (visium_constant_alignment): New function.
14378 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14379 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14380 (xtensa_constant_alignment): New function.
14381 * system.h (CONSTANT_ALIGNMENT): Poison.
14383 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14385 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14386 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14387 (rs6000_builtin_valid_without_lhs): New helper function.
14388 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14389 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14391 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14393 * target.h (vec_perm_indices): Use unsigned short rather than
14395 (auto_vec_perm_indices): Likewise.
14396 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14397 Use unsigned int rather than unsigned char.
14398 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14400 2017-09-25 Richard Biener <rguenther@suse.de>
14402 * cfgloop.h (sort_sibling_loops): Declare.
14403 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14404 (sort_sibling_loops): New function sorting the sibling loop list
14406 * graphite.c (graphite_transform_loops): Sort sibling loops.
14408 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14410 * target.def (vec_perm_const_ok): Change sel parameter to
14412 * optabs-query.c (can_vec_perm_p): Update accordingly.
14413 * doc/tm.texi: Regenerate.
14414 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14415 auto_vec_perm_indices and remove separate nelt field.
14416 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14417 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14418 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14419 (aarch64_expand_vec_perm_const): Update accordingly.
14420 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14421 to vec_perm_indices.
14422 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14423 auto_vec_perm_indices and remove separate nelt field.
14424 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14425 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14426 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14428 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14429 to vec_perm_indices.
14430 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14431 sel to vec_perm_indices.
14432 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14433 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14434 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14436 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14439 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14442 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14443 on the FUNCTION_DECL function context if it has a DIE that is a
14446 2017-09-25 Richard Biener <rguenther@suse.de>
14448 PR tree-optimization/82285
14449 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14452 2017-09-25 Tom de Vries <tom@codesourcery.com>
14456 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14459 2017-09-25 Richard Biener <rguenther@suse.de>
14461 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14462 ISL errors other than isl_error_quota happen. Dump if the
14463 schedule is the same.
14464 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14465 errors instead of aborting inside ISL.
14467 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14470 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14471 of libgcc_eh for m64.
14472 * config/i386/darwin64.h: Likewise.
14474 2017-09-25 Richard Biener <rguenther@suse.de>
14476 PR middle-end/82144
14477 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14478 attribute for incomplete types nor twice for complete ones.
14480 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14483 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14484 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14487 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14490 * config/i386/i386.opt (mprefer-avx256): Use
14491 ix86_target_flags variable.
14492 * config/i386/i386.c (ix86_target_string): Move
14493 -mprefer-avx256 to flag2_opts.
14495 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14497 PR middle-end/35691
14498 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14499 and x != -1 | y != -1 into (x & y) != -1.
14501 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14503 * config.gcc: Add new case statement to set
14504 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14505 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14506 s390x-*-linux* case statements. Added aarch64 to the list of
14507 supported architectures.
14509 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14511 PR tree-optimization/82289
14512 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14513 STMT_VINFO_RELEVANT_P.
14515 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14516 Alan Hayward <alan.hayward@arm.com>
14517 David Sherwood <david.sherwood@arm.com>
14519 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14520 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14521 (extract_range_from_binary_expr_1): Don't call
14522 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14524 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14525 Alan Hayward <alan.hayward@arm.com>
14526 David Sherwood <david.sherwood@arm.com>
14528 * target.def (preferred_vector_alignment): New hook.
14529 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14531 * doc/tm.texi: Regenerate.
14532 * targhooks.h (default_preferred_vector_alignment): Declare.
14533 * targhooks.c (default_preferred_vector_alignment): New function.
14534 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14536 (DR_TARGET_ALIGNMENT): New macro.
14537 (aligned_access_p): Update commentary.
14538 (vect_known_alignment_in_bytes): New function.
14539 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14541 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14542 Calculate the misalignment based on the target alignment rather than
14544 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14545 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14546 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14547 the target alignment, rather than masking the element misalignment
14548 with the number of elements in a vector. Also use the target
14549 alignment when calculating the maximum number of peels.
14550 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14551 instead of TYPE_ALIGN_UNIT.
14552 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14553 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14554 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14555 (vect_create_data_ref_ptr): Likewise.
14556 (vect_setup_realignment): Realign by ANDing with
14557 -DR_TARGET_MISALIGNMENT.
14558 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14559 the number of peels based on DR_TARGET_ALIGNMENT.
14560 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14561 with the guaranteed alignment boundary when deciding whether
14563 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14564 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14565 (ensure_base_align): Remove stmt_info parameter. Get the
14566 target base alignment from DR_TARGET_ALIGNMENT.
14567 (vectorizable_store): Update call accordingly. Interpret
14568 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14570 (vectorizable_load): Likewise.
14572 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14573 Alan Hayward <alan.hayward@arm.com>
14574 David Sherwood <david.sherwood@arm.com>
14576 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14577 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14578 (vect_enhance_data_refs_alignment): Likewise.
14580 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14582 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14583 error. Only quit immediately if parsing is complete.
14584 (BEGIN): Initialize fatal_err and parse_done.
14585 (begin fpu, end fpu): Check number of arguments.
14586 (begin arch, end arch): Likewise.
14587 (begin cpu, end cpu): Likewise.
14588 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14589 (optalias): Likewise.
14591 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14593 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14594 * config/arm/arm-isa.h: Delete. Move definitions to ...
14595 * arm-cpus.in: ... here. Use new feature and fgroup values.
14596 * config/arm/arm.c (arm_option_override): Use lower case for feature
14598 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14599 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14600 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14602 (print_isa_bits_for): New function.
14603 (gen_isa): New function.
14604 (gen_comm_data): Use print_isa_bits_for.
14605 (define feature): New keyword.
14606 (define fgroup): New keyword.
14607 * config/arm/t-arm (TM_H): Remove.
14608 (GTM_H): Add arm-isa.h.
14609 (arm-isa.h): Add rule to generate file.
14610 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14611 case for feature bit names.
14613 2017-09-22 Richard Biener <rguenther@suse.de>
14615 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14617 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14618 print of no dependency loops ...
14619 * graphite.c (graphite_transform_loops): ... here.
14620 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14622 (same_close_phi_node, remove_duplicate_close_phi,
14623 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14624 (canonicalize_loop_closed_ssa): ... here and simplify.
14625 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14626 (optimize_isl): Use dump_printf_loc to tell when we stopped
14627 optimizing because of an ISL timeout.
14629 2017-09-22 Richard Biener <rguenther@suse.de>
14631 PR tree-optimization/82291
14632 * tree-if-conv.c (predicate_mem_writes): Make sure to
14633 remove writes in blocks predicated with false.
14635 2017-09-22 Richard Biener <rguenther@suse.de>
14637 * sese.c: Include cfganal.h.
14638 (if_region_set_false_region): Remove.
14639 (create_if_region_on_edge): Likewise.
14640 (move_sese_in_condition): Re-implement without destroying
14643 2017-09-22 Richard Biener <rguenther@suse.de>
14645 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14646 Verify both BBs contain loop PHI nodes before dispatching to
14647 copy_loop_phi_args.
14648 (graphite_regenerate_ast_isl): Do not recompute dominators,
14649 do not verify three times. Restructure for clarity.
14650 * graphite-scop-detection.c (same_close_phi_node,
14651 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14652 defined_in_loop_p, canonicalize_loop_closed_ssa,
14653 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14654 checking and SSA rewrite, move to ...
14655 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14656 (graphite_initialize): Do not pass in ctx, do not reset the
14657 SCEV cache, compute only dominators.
14658 (graphite_transform_loops): Allocate ISL ctx after
14659 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14660 Maintain post-dominators only around build_scops.
14661 * sese.c (if_region_set_false_region): Make static. Free
14662 and recompute dominators.
14663 (move_sese_in_condition): Assert we don't get called with
14664 post-dominators computed.
14665 * sese.h (if_region_set_false_region): Remove.
14667 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14669 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14670 mode attribute for TARGET_AVX512VL.
14672 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14674 * config/i386/i386.opt (mprefer-avx256): New option.
14675 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14677 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14678 for TARGET_PREFER_AVX256.
14679 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14681 2017-09-21 Jeff Law <law@redhat.com>
14683 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14684 Fix dump output if the only stack space is for pushed registers.
14686 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14688 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14691 2017-09-21 Martin Sebor <msebor@redhat.com>
14694 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14695 code (in C++) or code that triggers warnings.
14697 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14699 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14701 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14703 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14704 * sched-rgn.c: Ditto.
14705 * sel-sched-ir.c: Ditto.
14707 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14709 * toplev.h (set_random_seed): Adjust return type.
14710 * toplev.c (init_local_tick): Move eager initialization of random_seed
14711 to get_random_seed. Adjust comment.
14712 (init_random_seed): Inline to get_random_seed, delete.
14713 (get_random_seed): Initialize random_seed lazily.
14714 (set_random_seed): Do not return previous value.
14715 (print_switch_value): Do not call get_random_seed.
14717 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14719 * cgraph.c (delete_function_version): New, broken out from...
14720 (cgraph_node::delete_function_version): ...here. Rename to
14721 cgraph_node::delete_function_version_by_decl. Update all uses.
14722 (cgraph_node::remove): Call delete_function_version.
14724 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14727 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14728 VAR_DECLs to which addressable non-volatile parameters are mapped
14729 and for id->retvar after the return value assignment. Clear
14730 id->retval and id->retbnd after inlining.
14732 2017-09-21 Richard Biener <rguenther@suse.de>
14734 PR tree-optimization/82276
14735 PR tree-optimization/82244
14736 * tree-vrp.c (build_assert_expr_for): Set
14737 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14739 (remove_range_assertions): Revert earlier change.
14741 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14744 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14746 2017-09-21 Richard Biener <rguenther@suse.de>
14748 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14749 Restore valid IL after code generation errors.
14750 * graphite.c (graphite_transform_loops): Diagnose code
14751 generation issues as MSG_MISSED_OPTIMIZATION and continue
14752 with processing SCOPs.
14754 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14755 Alan Hayward <alan.hayward@arm.com>
14756 David Sherwood <david.sherwood@arm.com>
14758 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14759 rather than choosing between plus_constant and gen_rtx_<CODE>.
14760 * expr.c (emit_push_insn): Likewise.
14761 (expand_expr_real_2): Likewise.
14763 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14764 Alan Hayward <alan.hayward@arm.com>
14765 David Sherwood <david.sherwood@arm.com>
14767 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14769 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14770 Alan Hayward <alan.hayward@arm.com>
14771 David Sherwood <david.sherwood@arm.com>
14773 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14774 calling tree_to_uhwi.
14776 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14781 INTEGER_CST rather than a negative test for ADDR_EXPR.
14783 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14784 Alan Hayward <alan.hayward@arm.com>
14785 David Sherwood <david.sherwood@arm.com>
14787 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14788 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14790 2017-09-21 Richard Biener <rguenther@suse.de>
14792 PR tree-optimization/71351
14793 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14794 graphite_create_new_loop_guard): Remove, fold remaining parts
14796 (translate_isl_ast_node_for): ... here and simplify.
14798 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14801 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14802 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14803 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14804 alternative always use QI mode, for -Os imov (=R,R) alternative
14805 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14808 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14809 Jeff Law <law@redhat.com>
14811 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14812 (allocate_stack_space): New function, partially extracted from
14813 s390_emit_prologue.
14814 (s390_emit_prologue): Track offset to most recent stack probe.
14815 Code to allocate space moved into allocate_stack_space.
14816 Dump actions when no stack is allocated.
14817 (s390_prologue_plus_offset): New function.
14818 (s390_emit_stack_probe): Likewise.
14820 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14822 * common.opt (Wa, Wl, Wp, g, gz=): Add
14824 (gno-column-info): Remove.
14825 (gcolumn-info): Drop RejectNegative.
14826 (gno-): New prefix.
14827 (gno-record-gcc-switches): Remove.
14828 (grecord-gcc-switches): Drop RejectNegative.
14829 (gno-split-dwarf): Remove.
14830 (gsplit-dwarf): Drop RejectNegative.
14831 (gno-strict-dwarf): Remove.
14832 (gstrict-dwarf): Drop RejectNegative.
14833 * config/darwin.opt (gfull, gused): Add RejectNegative.
14834 * dwarf2out.c (gen_producer_string): Drop
14835 gno-record-gcc-switches handler.
14836 * optc-gen.awk: Add g to prefixes with negative forms.
14837 * opts-common.c (remapping_prefix_p): New.
14838 (find_opt): Check it.
14839 (generate_canonical_option): Test g prefix.
14840 (option_map): Add -gno- mapping.
14841 (add_misspelling_candidates): Check remapping_prefix_p.
14843 2017-09-20 Jeff Law <law@redhat.com>
14845 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14846 thinko in stack clash protection support.
14848 * explow.c (compute_stack_clash_protection_loop_data): Use
14849 CONST_INT_P instead of explicit test. Verify object is a
14850 CONST_INT_P before looking at INTVAL.
14851 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14852 instead of explicit test.
14854 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14857 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14858 address instead of to r1 and r11.
14860 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14862 * config.gcc: Support "knm".
14863 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14864 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14866 * config/i386/i386.c (m_KNM): Define.
14867 (processor_target_table): Add "knm".
14869 (ix86_option_override_internal): Add "knm".
14870 (ix86_issue_rate): Add PROCESSOR_KNM.
14871 (ix86_adjust_cost): Ditto.
14872 (ia32_multipass_dfa_lookahead): Ditto.
14873 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14874 (fold_builtin_cpu): Add M_INTEL_KNM.
14875 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14876 (processor_type): Add PROCESSOR_KNM.
14877 * config/i386/x86-tune.def: Add m_KNM.
14878 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14880 2017-09-20 Richard Biener <rguenther@suse.de>
14882 PR tree-optimization/80213
14883 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14884 are allowed in empty BBs as well.
14885 (canonicalize_loop_closed_ssa): Also look for other complex
14887 (scop_detection::get_sese): Include the loop-closed PHI block
14889 (scop_detection::merge_sese): Remove code adding extra blocks.
14890 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14891 (build_scops): Assert the final returned scop is invalid.
14893 2017-09-20 Richard Biener <rguenther@suse.de>
14895 PR tree-optimization/82264
14896 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14898 (vn_phi_lookup): Likewise.
14899 (vn_phi_insert): Likewise.
14901 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14903 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14904 that fits into uhwi or shwi, add DW_AT_const_value regardless
14905 of early_dwarf without going through RTL, using add_AT_unsigned
14908 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14909 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14910 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14911 (DEBUG_MACRO_SECTION): Likewise.
14912 (DEBUG_LTO_MACRO_SECTION): Likewise.
14913 (DEBUG_STR_DWO_SECTION): Likewise.
14914 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14915 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14916 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14917 (DEBUG_LTO_LINE_STR_SECTION): Define.
14918 (init_sections_and_labels): Initialize debug_line_str_section
14919 variable. Initialize debug_loc_section for -gdwarf-5 to
14920 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14922 2017-09-20 Richard Biener <rguenther@suse.de>
14924 * graphite-sese-to-poly.c (extract_affine): Properly handle
14925 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14927 2017-09-20 Richard Biener <rguenther@suse.de>
14929 PR tree-optimization/81373
14930 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14931 Force SESE live-out defs to be handled even if they are
14934 2017-09-19 Jeff Law <law@redhat.com>
14936 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14937 nothing for stack adjustments with REG_STACK_CHECK.
14938 * sched-deps.c (parse_add_or_inc): Reject insns with
14939 REG_STACK_CHECK from dependency breaking.
14940 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14941 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14942 * reg-notes.def (STACK_CHECK): New note.
14944 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14945 (ix86_expand_prologue): Dump stack clash info as needed.
14946 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14948 * function.c (dump_stack_clash_frame_info): New function.
14949 * function.h (dump_stack_clash_frame_info): Prototype.
14950 (enum stack_clash_probes): New enum.
14952 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14953 flag_stack_clash_protection.
14954 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14955 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14956 (arm_frame_pointer_required): Likewise.
14957 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14958 (ia64_expand_prologue): Likewise.
14959 * config/mips/mips.c (mips_expand_prologue): Likewise.
14960 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14961 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14962 (sparc_flat_expand_prologue): Likewise.
14963 * config/spu/spu.c (spu_expand_prologue): Likewise.
14965 * explow.c: Include "params.h".
14966 (anti_adjust_stack_and_probe_stack_clash): New function.
14967 (get_stack_check_protect): Likewise.
14968 (compute_stack_clash_protection_loop_data): Likewise.
14969 (emit_stack_clash_protection_loop_start): Likewise.
14970 (emit_stack_clash_protection_loop_end): Likewise.
14971 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14972 Use anti_adjust_stack_and_probe_stack_clash.
14973 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14974 (emit_stack_clash_protection_loop_start): Likewise.
14975 (emit_stack_clash_protection_loop_end): Likewise.
14976 * rtl.h (get_stack_check_protect): Prototype.
14977 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14978 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14979 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14981 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14983 * doc/tm.texi: Rebuilt.
14984 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14985 get_stack_check_protect.
14986 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14987 * config/arm/arm.c (arm_expand_prologue): Likewise.
14988 (arm_frame_pointer_required): Likewise.
14989 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14990 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14991 * config/mips/mips.c (mips_expand_prologue): Likewise.
14992 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14993 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14994 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14995 (sparc_flat_expand_prologue): Likewise.
14997 * common.opt (-fstack-clash-protection): New option.
14998 * flag-types.h (enum stack_check_type): Note difference between
14999 -fstack-check= and -fstack-clash-protection.
15000 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15001 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15002 * toplev.c (process_options): Issue warnings/errors for cases
15003 not handled with -fstack-clash-protection.
15004 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15005 (-fstack-check): Note additional problem with -fstack-check=generic.
15006 Note that -fstack-check is primarily for Ada and refer users
15007 to -fstack-clash-protection for stack-clash-protection.
15008 Document new params for stack clash protection.
15010 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15012 * config/i386/i386.c (ix86_split_long_move): Do not handle
15013 address used for LEA in a special way.
15015 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15017 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15019 2017-09-19 Martin Sebor <msebor@redhat.com>
15022 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15023 of incompatible types.
15025 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15027 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15028 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15029 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15030 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15032 2017-09-19 Richard Biener <rguenther@suse.de>
15034 PR tree-optimization/82244
15035 * tree-vrp.c (remove_range_assertions): Do not propagate
15036 a constant to abnormals but replace the assert with a copy.
15038 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15040 PR rtl-optimization/57878
15041 PR rtl-optimization/68988
15042 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15043 avoidance test involving non_reload_pseudos. Move frequency test
15044 below the general fragmentation avoidance test.
15046 2017-09-19 Richard Biener <rguenther@suse.de>
15048 PR tree-optimization/69728
15049 * graphite-sese-to-poly.c (schedule_error): New global.
15050 (add_loop_schedule): Handle empty domain by failing the
15052 (build_original_schedule): Handle schedule_error.
15054 2017-09-19 Richard Biener <rguenther@suse.de>
15056 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15057 Do not iterate to sibling loops but only to siblings of inner
15060 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15063 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15064 that the registers are different.
15066 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15068 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15069 to processor_model and "amdfam17h" to arch_names_table.
15070 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15072 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15075 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15077 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15078 Alan Hayward <alan.hayward@arm.com>
15079 David Sherwood <david.sherwood@arm.com>
15081 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15083 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15085 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15086 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15087 vect_schedule_slp_instance.
15088 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15089 Update call to vect_slp_analyze_node_operations. Simplify return
15091 (vect_slp_analyze_bb_1): Update call accordingly.
15092 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15093 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15094 (vect_schedule_slp): Update call accordingly.
15096 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15097 Alan Hayward <alan.hayward@arm.com>
15098 David Sherwood <david.sherwood@arm.com>
15100 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15101 with types that aren't in fact scalar.
15103 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15105 * tree-vect-slp.c (vect_record_max_nunits): New function,
15107 (vect_build_slp_tree_1): ...here.
15108 (vect_build_slp_tree_2): Call it for phis too.
15110 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15112 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15113 to vect_get_vec_def_for_operand when getting the mask operand.
15115 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15116 Alan Hayward <alan.hayward@arm.com>
15117 David Sherwood <david.sherwood@arm.com>
15119 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15122 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15123 Alan Hayward <alan.hayward@arm.com>
15124 David Sherwood <david.sherwood@arm.com>
15126 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15127 calculation for vector booleans.
15129 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15130 Alan Hayward <alan.hayward@arm.com>
15131 David Sherwood <david.sherwood@arm.com>
15133 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15135 (vect_transform_stmt): ...here.
15136 (vect_analyze_stmt): Use it instead of calling
15137 vectorizable_live_operation directly.
15139 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15141 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15142 non-SIMT targets in acc vector loops.
15144 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15146 * configure.ac: Add arc and check if assembler supports gdwarf2.
15147 * configure: Regenerate.
15149 2017-09-18 Richard Biener <rguenther@suse.de>
15151 PR tree-optimization/82220
15152 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15153 epilogue niters from the min_profitable_iters compute.
15155 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15158 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15159 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15160 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15162 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15165 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15166 switching to a new text section.
15168 2017-09-18 Richard Biener <rguenther@suse.de>
15170 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15172 (build_alias_set): Reject aliases with no access function.
15174 2017-09-18 Richard Biener <rguenther@suse.de>
15176 PR tree-optimization/79622
15177 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15179 (build_cross_bb_scalars_use): Likewise.
15181 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15183 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15185 2017-09-18 Alan Modra <amodra@gmail.com>
15188 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15189 stack_pointer_rtx for count 0. Update comments. Break up
15190 large rtl expression.
15192 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15194 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15195 Increase to 20 bytes.
15196 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15197 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15198 or avx version of the stub.
15200 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15203 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15204 compute the minimum stack alignment. Also update preferred stack
15205 boundary for leaf functions.
15207 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15209 PR tree-optimization/82228
15210 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15213 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15215 * common/config/nds32/nds32-common.c
15216 (nds32_option_optimization_table): Refine formatting.
15217 (nds32_option_optimization_table): Use -fsched-pressure and
15218 -fomit-frame-pointer for specific optimization level.
15220 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15222 * config/nds32/nds32.c: Refine formatting and comments.
15223 * config/nds32/nds32.h: Likewise.
15224 * config/nds32/nds32.md: Likewise.
15225 * config/nds32/nds32-cost.c: Likewise.
15226 * config/nds32/nds32-isr.c: Likewise.
15227 * config/nds32/nds32-md-auxiliary.c: Likewise.
15228 * config/nds32/nds32-multiple.md: Likewise.
15229 * config/nds32/nds32-predicates.c: Likewise.
15231 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15232 Jakub Jelinek <jakub@redhat.com>
15234 Add support for -std=c++2a.
15235 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15237 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15239 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15242 * doc/extend.texi (Common Function Attributes): Add
15243 references to ARM, AArch64, and S/390 specific attributes.
15244 (Function Specific Option Pragmas): Add AArch64 and S/390
15245 to list of back ends that support the target pragma.
15247 2017-09-15 Nathan Sidwell <nathan@acm.org>
15249 * doc/standards.texi: Fix C++17 description. Update URLs for
15252 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15254 * common.opt (Wcast-align=strict): New warning option.
15255 * doc/invoke.texi: Document -Wcast-align=strict.
15257 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15259 * cgraph.h (cgraph_thunk_info): Add comments.
15260 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15261 assert for VIRTUAL_* arguments stricter.
15263 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15265 PR tree-optimization/71026
15266 * match.pd: Move RDIV patterns from fold-const.c
15267 * fold-const.c (distribute_real_division): Removed.
15268 (fold_binary_loc): Remove calls to distribute_real_divison.
15270 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15272 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15273 c++1z and gnu++1z as deprecated. Change other references to
15274 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15275 Change -Wc++1z-compat to -Wc++17-compat.
15276 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15277 * dwarf2out.c (highest_c_language): Handle C++17.
15278 (gen_compile_unit_die): Likewise.
15280 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15282 PR rtl-optimization/82192
15283 * combine.c (make_extraction): Don't look through non-paradoxical
15284 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15287 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15288 Alan Hayward <alan.hayward@arm.com>
15289 David Sherwood <david.sherwood@arm.com>
15291 * target.def (function_arg_offset): New hook.
15292 * targhooks.h (default_function_arg_offset): Declare.
15293 * targhooks.c (default_function_arg_offset): New function.
15294 * function.c (locate_and_pad_parm): Use
15295 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15296 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15297 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15298 * doc/tm.texi: Regenerate.
15299 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15300 * config/spu/spu.c (spu_function_arg_offset): New function.
15301 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15302 * system.h (FUNCTION_ARG_OFFSET): Poison.
15304 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15305 Alan Hayard <alan.hayward@arm.com>
15306 David Sherwood <david.sherwood@arm.com>
15308 * target.def (truly_noop_truncation): New hook.
15309 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15310 than TRULY_NOOP_TRUNCATION.
15311 * hooks.h (hook_bool_uint_uint_true): Declare.
15312 * hooks.c (hook_bool_uint_uint_true): New function.
15313 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15314 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15315 * doc/tm.texi: Regenerate.
15316 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15317 rather than TRULY_NOOP_TRUNCATION in comments.
15318 (simplify_comparison): Likewise.
15319 (record_truncated_value): Likewise.
15320 * expmed.c (extract_bit_field_1): Likewise.
15321 (extract_split_bit_field): Likewise.
15322 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15323 instead of TRULY_NOOP_TRUNCATION.
15324 * function.c (assign_parm_setup_block): Likewise.
15325 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15326 * rtlhooks.c: Include target.h.
15327 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15328 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15329 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15330 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15331 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15332 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15333 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15334 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15335 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15336 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15337 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15338 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15339 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15340 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15341 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15342 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15343 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15344 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15345 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15346 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15347 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15348 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15349 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15350 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15351 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15352 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15353 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15354 rather than TRULY_NOOP_TRUNCATION in comments.
15355 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15356 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15357 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15358 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15359 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15360 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15361 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15362 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15363 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15364 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15365 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15366 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15367 rather than TRULY_NOOP_TRUNCATION in comments.
15368 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15369 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15370 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15371 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15372 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15373 TRULY_NOOP_TRUNCATION condition.
15374 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15375 (TRULY_NOOP_TRUNCATION): Delete.
15376 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15377 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15378 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15379 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15380 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15381 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15382 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15383 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15384 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15385 rather than TRULY_NOOP_TRUNCATION in comments.
15386 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15387 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15388 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15389 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15390 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15391 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15393 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15396 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15397 (*cmp_ior): Likewise.
15398 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15399 (*ior_scc_scc_cmp): Likewise.
15400 (*and_scc_scc): Likewise.
15401 (*and_scc_scc_cmp): Likewise.
15403 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15404 Alan Hayard <alan.hayward@arm.com>
15405 David Sherwood <david.sherwood@arm.com>
15407 * target.def (can_change_mode_class): New hook.
15408 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15409 (hard_regno_nregs): Likewise.
15410 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15411 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15412 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15413 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15414 (LOAD_EXTEND_OP): Update accordingly.
15415 * doc/tm.texi: Regenerate.
15416 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15417 CANNOT_CHANGE_MODE_CLASS.
15418 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15419 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15420 * combine.c (simplify_set): Update accordingly.
15421 * emit-rtl.c (validate_subreg): Likewise.
15422 * recog.c (general_operand): Likewise.
15423 * regcprop.c (mode_change_ok): Likewise.
15424 * reload1.c (choose_reload_regs): Likewise.
15425 (inherit_piecemeal_p): Likewise.
15426 * rtlanal.c (simplify_subreg_regno): Likewise.
15427 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15428 instead of CANNOT_CHANGE_MODE_CLASS.
15429 (reload_cse_simplify_operands): Likewise.
15430 * reload.c (push_reload): Use targetm.can_change_mode_class
15431 instead of CANNOT_CHANGE_MODE_CLASS.
15432 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15433 REG_CANNOT_CHANGE_MODE_P.
15434 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15435 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15436 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15437 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15438 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15439 (arm_can_change_mode_class): New function.
15440 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15441 than CANNOT_CHANGE_MODE_CLASS in comments.
15442 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15443 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15444 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15445 (ix86_can_change_mode_class): ...this new function, inverting the
15446 sense of the return value.
15447 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15448 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15449 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15450 (ia64_can_change_mode_class): New function.
15451 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15452 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15453 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15454 (m32c_can_change_mode_class): ...this new function, inverting the
15455 sense of the return value.
15456 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15457 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15458 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15459 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15460 (mips_can_change_mode_class): ...this new function, inverting the
15461 sense of the return value.
15462 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15463 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15464 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15465 (msp430_can_change_mode_class): New function.
15466 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15467 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15468 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15469 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15470 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15471 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15472 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15473 (pa_cannot_change_mode_class): Replace with...
15474 (pa_can_change_mode_class): ...this new function, inverting the
15475 sense of the return value.
15476 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15477 than CANNOT_CHANGE_MODE_CLASS in comments.
15478 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15479 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15480 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15481 (pdp11_cannot_change_mode_class): Replace with...
15482 (pdp11_can_change_mode_class): ...this new function, inverting the
15483 sense of the return value.
15484 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15485 * config/powerpcspe/powerpcspe-protos.h
15486 (rs6000_cannot_change_mode_class_ptr): Delete.
15487 * config/powerpcspe/powerpcspe.c
15488 (rs6000_cannot_change_mode_class_ptr): Delete.
15489 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15490 (rs6000_option_override_internal): Assign to
15491 targetm.can_change_mode_class instead of
15492 rs6000_cannot_change_mode_class_ptr.
15493 (rs6000_cannot_change_mode_class): Replace with...
15494 (rs6000_can_change_mode_class): ...this new function, inverting the
15495 sense of the return value.
15496 (rs6000_debug_cannot_change_mode_class): Replace with...
15497 (rs6000_debug_can_change_mode_class): ...this new function.
15498 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15499 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15500 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15501 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15502 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15504 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15505 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15506 (rs6000_option_override_internal): Assign to
15507 targetm.can_change_mode_class instead of
15508 rs6000_cannot_change_mode_class_ptr.
15509 (rs6000_cannot_change_mode_class): Replace with...
15510 (rs6000_can_change_mode_class): ...this new function, inverting the
15511 sense of the return value.
15512 (rs6000_debug_cannot_change_mode_class): Replace with...
15513 (rs6000_debug_can_change_mode_class): ...this new function.
15514 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15515 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15516 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15517 (s390_can_change_mode_class): ...this new function, inverting the
15518 sense of the return value.
15519 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15520 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15521 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15522 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15523 (sh_cannot_change_mode_class): Replace with...
15524 (sh_can_change_mode_class): ...this new function, inverting the
15525 sense of the return value.
15526 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15527 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15528 (sparc_can_change_mode_class): New function.
15529 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15530 * config/spu/spu.c (spu_can_change_mode_class): New function.
15531 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15532 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15533 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15534 (visium_can_change_mode_class): New function.
15535 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15537 2017-09-15 Richard Biener <rguenther@suse.de>
15539 PR tree-optimization/82217
15540 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15541 but not undefined case.
15543 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15546 * postreload.c (reload_cse_simplify_operands): Skip
15547 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15549 2017-09-15 Richard Biener <rguenther@suse.de>
15551 PR tree-optimization/68823
15552 * graphite-scop-detection.c (build_alias_set): If we have a
15553 possible dependence check whether we can handle them by just
15554 looking at the DRs DR_ACCESS_FNs.
15555 (build_scops): If build_alias_set fails, fail the SCOP.
15557 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15559 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15560 to support float128 built-in functions that require the ISA 3.0
15562 (BU_FLOAT128_3_HW): Likewise.
15563 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15564 built-in functions.
15565 (FMAF128): Likewise.
15567 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15568 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15569 floating point instructions.
15570 (rs6000_invalid_builtin): Likewise.
15571 (rs6000_builtin_mask_names): Likewise.
15572 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15573 (RS6000_BTM_FLOAT128_HW): Likewise.
15574 (RS6000_BTM_COMMON): Likewise.
15575 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15577 * doc/extend.texi (RS/6000 built-in functions): Document the
15578 IEEE 128-bit floating point square root and fused multiply-add
15579 built-in functions.
15581 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15583 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15584 reg (r2) isn't in the set of registers defined in the prologue.
15586 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15587 Alan Hayward <alan.hayward@arm.com>
15588 David Sherwood <david.sherwood@arm.com>
15590 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15591 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15592 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15593 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15594 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15596 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15597 max_vectorization_factor.
15598 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15600 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15601 Alan Hayward <alan.hayward@arm.com>
15602 David Sherwood <david.sherwood@arm.com>
15604 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15605 (vect_worthwhile_without_simd_p): Declare.
15606 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15607 (vectorizable_reduction): Use it.
15608 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15609 (vectorizable_operation): Likewise.
15611 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15612 Alan Hayward <alan.hayward@arm.com>
15613 David Sherwood <david.sherwood@arm.com>
15615 * tree-vectorizer.h (vect_get_num_copies): New function.
15616 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15617 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15618 (vectorizable_induction): Likewise.
15619 (vectorizable_live_operation): Likewise.
15620 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15621 (vectorizable_bswap): Likewise.
15622 (vectorizable_call): Likewise.
15623 (vectorizable_conversion): Likewise.
15624 (vectorizable_assignment): Likewise.
15625 (vectorizable_shift): Likewise.
15626 (vectorizable_operation): Likewise.
15627 (vectorizable_store): Likewise.
15628 (vectorizable_load): Likewise.
15629 (vectorizable_condition): Likewise.
15630 (vectorizable_comparison): Likewise.
15631 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15633 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15634 Alan Hayward <alan.hayward@arm.com>
15635 David Sherwood <david.sherwood@arm.com>
15637 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15638 of vect_init_vector.
15640 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15641 Alan Hayward <alan.hayward@arm.com>
15642 David Sherwood <david.sherwood@arm.com>
15644 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15645 an inline wrapper that provides a location.
15646 (gimple_build_vector): Likewise.
15647 * gimple-fold.c (gimple_build_vector_from_val): New function.
15648 (gimple_build_vector): Likewise.
15649 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15650 functions to build the initial value. Always return a gimple value.
15651 (get_initial_defs_for_reduction): Likewise. Only compute
15653 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15654 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15655 (vectorizable_induction): Use gimple_build_vector rather than
15658 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15659 Alan Hayward <alan.hayward@arm.com>
15660 David Sherwood <david.sherwood@arm.com>
15662 * target.h (vec_perm_indices): New typedef.
15663 (auto_vec_perm_indices): Likewise.
15664 * optabs-query.h: Include target.h
15665 (can_vec_perm_p): Take a vec_perm_indices *.
15666 * optabs-query.c (can_vec_perm_p): Likewise.
15667 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15668 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15669 * tree-vect-generic.c (lower_vec_perm): Likewise.
15670 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15671 (vect_grouped_load_supported): Likewise.
15672 (vect_shift_permute_load_chain): Likewise.
15673 (vect_permute_store_chain): Use auto_vec_perm_indices.
15674 (vect_permute_load_chain): Likewise.
15675 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15676 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15677 Update uses of can_vec_perm_p.
15678 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15679 mode with a number of elements. Take a vec_perm_indices *.
15680 (vect_create_epilog_for_reduction): Update accordingly.
15681 Use auto_vec_perm_indices.
15682 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15683 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15684 (vect_transform_slp_perm_load): Likewise.
15685 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15686 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15687 (vect_gen_perm_mask_checked): Likewise.
15688 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15689 (vect_gen_perm_mask_checked): Likewise.
15690 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15691 (vectorizable_store): Likewise.
15692 (vectorizable_load): Likewise.
15693 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15694 (vectorizable_bswap): Likewise.
15696 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15697 Alan Hayward <alan.hayward@arm.com>
15698 David Sherwood <david.sherwood@arm.com>
15700 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15701 * tree.c (build_vector): Likewise.
15702 (build_vector_from_ctor): Update accordingly.
15703 (build_vector_from_val): Likewise.
15704 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15705 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15706 * tree-vect-generic.c (add_rshift): Likewise.
15707 (expand_vector_divmod): Likewise.
15708 (optimize_vector_constructor): Likewise.
15709 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15710 (vect_transform_slp_perm_load): Likewise.
15711 (vect_schedule_slp_instance): Likewise.
15712 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15713 (vectorizable_call): Likewise.
15714 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15715 * expmed.c (make_tree): Likewise.
15716 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15717 a vector passed to build_vector.
15718 (fold_convert_const): Likewise.
15719 (exact_inverse): Likewise.
15720 (fold_ternary_loc): Likewise.
15721 (fold_relational_const): Likewise.
15722 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15723 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15724 (const_unop): Likewise. Store the reduction accumulator in a
15725 variable rather than an array.
15726 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15727 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15728 the new vector, rather than constructing it after the input arrays.
15729 (native_interpret_vector): Use auto_vec<tree> when building
15730 a vector passed to build_vector. Add elements in order.
15731 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15732 auto_vec<tree> when building a vector passed to build_vector.
15733 (vect_create_epilog_for_reduction): Likewise.
15734 (vectorizable_induction): Likewise.
15735 (get_initial_def_for_reduction): Likewise. Fix indentation of
15737 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15739 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15740 passed to build_vector.
15742 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15743 Alan Hayward <alan.hayward@arm.com>
15744 David Sherwood <david.sherwood@arm.com>
15746 * tree-core.h (tree_base::u): Add an "nelts" field.
15747 (tree_vector): Use VECTOR_CST_NELTS as the length.
15748 * tree.c (tree_size): Likewise.
15749 (make_vector): Initialize VECTOR_CST_NELTS.
15750 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15751 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15752 TYPE_VECTOR_SUBPARTS.
15753 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15754 as the number of units, setting it from VECTOR_CST_NELTS.
15755 (const_vector_from_tree): Likewise.
15756 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15757 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15758 (fold_negate_expr_1): Likewise.
15759 (fold_convert_const): Likewise.
15760 (const_binop): Likewise. Differentiate the number of output and
15762 (const_unop): Likewise.
15763 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15764 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15765 in cases that did the opposite.
15767 2017-09-14 Richard Biener <rguenther@suse.de>
15769 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15772 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15774 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15776 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15779 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15780 if and where to split a bb, except for splitting before debug insn
15781 sequences followed by non-label real insn. Delete debug insns
15782 in between basic blocks.
15784 * combine.c (make_compound_operation_int): Formatting fixes.
15786 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15787 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15788 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15789 * config/sol2.h (LINK_EH_SPEC): Likewise.
15790 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15791 * config/s390/linux.h (LINK_SPEC): Likewise.
15792 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15793 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15794 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15795 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15796 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15797 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15798 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15799 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15801 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15803 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15805 (ENDFILE_LINUX_SPEC): Likewise.
15806 (LINK_EH_SPEC): Likewise.
15807 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15808 (LINK_OS_LINUX_SPEC32): Likewise.
15809 (LINK_OS_LINUX_SPEC64): Likewise.
15810 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15811 (LINK_OS_LINUX_SPEC): Likewise.
15813 2017-09-13 Martin Liska <mliska@suse.cz>
15815 PR middle-end/82154
15816 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15817 CASE_HIGH is NULL_TREE.
15819 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15820 Alan Hayward <alan.hayward@arm.com>
15821 David Sherwood <david.sherwood@arm.com>
15823 * target.def (secondary_memory_needed): New hook.
15824 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15825 instead of SECONDARY_MEMORY_NEEDED.
15826 (secondary_memory_needed_mode): Likewise.
15827 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15828 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15829 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15830 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15831 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15832 * doc/tm.texi: Regenerate.
15833 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15834 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15835 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15836 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15837 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15838 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15839 mode argument first and change the reg_class arguments to reg_class_t.
15840 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15841 Make static. Update the call to inline_secondary_memory_needed.
15842 (ix86_register_move_cost): Update the call to
15843 inline_secondary_memory_needed.
15844 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15845 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15847 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15848 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15850 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15851 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15852 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15853 and match hook interface. Add comment from mips.h.
15854 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15855 * config/mmix/mmix.md (truncdfsf2): Refer to
15856 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15858 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15859 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15860 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15861 (pa_secondary_memory_needed): New function.
15862 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15863 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15864 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15865 (pdp11_secondary_memory_needed): Make static and match hook interface.
15866 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15867 * config/powerpcspe/powerpcspe-protos.h
15868 (rs6000_secondary_memory_needed_ptr): Delete.
15869 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15871 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15872 (rs6000_option_override_internal): Assign to
15873 targetm.secondary_memory_needed rather than
15874 rs6000_secondary_memory_needed_ptr.
15875 (rs6000_secondary_memory_needed): Match hook interface.
15876 (rs6000_debug_secondary_memory_needed): Likewise.
15877 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15878 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15879 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15880 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15881 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15882 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15884 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15885 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15886 (rs6000_option_override_internal): Assign to
15887 targetm.secondary_memory_needed rather than
15888 rs6000_secondary_memory_needed_ptr.
15889 (rs6000_secondary_memory_needed): Match hook interface.
15890 (rs6000_debug_secondary_memory_needed): Likewise.
15891 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15892 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15893 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15894 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15895 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15896 (sparc_secondary_memory_needed): New function.
15897 * lra-constraints.c (check_and_process_move): Refer to
15898 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15900 (curr_insn_transform): Likewise.
15901 (process_alt_operands): Use targetm.secondary_memory_needed
15902 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15903 (check_secondary_memory_needed_p): Likewise.
15904 (choose_split_class): Likewise.
15905 * reload.c: Unconditionally include code that was previously
15906 conditional on SECONDARY_MEMORY_NEEDED.
15907 (push_secondary_reload): Use targetm.secondary_memory_needed
15908 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15909 (push_reload): Likewise.
15910 * reload1.c: Unconditionally include code that was previously
15911 conditional on SECONDARY_MEMORY_NEEDED.
15912 (choose_reload_regs): Use targetm.secondary_memory_needed
15913 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15914 (gen_reload): Likewise.
15915 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15917 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15918 Alan Hayward <alan.hayward@arm.com>
15919 David Sherwood <david.sherwood@arm.com>
15921 * target.def (secondary_memory_needed_mode): New hook:
15922 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15923 * targhooks.h (default_secondary_memory_needed_mode): New function.
15924 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15925 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15926 * doc/tm.texi: Regenerate.
15927 * lra-constraints.c (check_and_process_move): Use
15928 targetm.secondary_memory_needed_mode instead of
15929 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15930 (curr_insn_transform): Likewise.
15931 * reload.c (get_secondary_mem): Likewise.
15932 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15933 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15935 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15936 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15937 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15938 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15939 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15941 * config/powerpcspe/powerpcspe-protos.h
15942 (rs6000_secondary_memory_needed_mode): Delete.
15943 * config/powerpcspe/powerpcspe.c
15944 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15945 (rs6000_secondary_memory_needed_mode): Make static.
15946 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15947 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15949 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15951 (rs6000_secondary_memory_needed_mode): Make static.
15952 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15953 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15954 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15955 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15956 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15958 (sparc_secondary_memory_needed_mode): New function.
15959 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15961 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15963 * config/aarch64/constraints.md (Umq): New constraint.
15964 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15966 (mov<mode>): Update condition.
15968 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15970 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15971 when bitposition is the same.
15973 2017-09-13 Richard Biener <rguenther@suse.de>
15975 * dwarf2out.c (output_die_symbol): Remove.
15976 (output_die): Do not output a DIEs symbol.
15978 2017-09-13 Richard Biener <rguenther@suse.de>
15980 PR middle-end/82128
15981 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15982 default-def to avoid breaking iterator update with the weird
15983 interaction with cgraph_update_edges_for_call_stmt_node.
15985 2017-09-13 Richard Biener <rguenther@suse.de>
15987 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15988 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15989 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15990 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15992 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15994 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15995 Disable pc relative literal load irrespective of
15996 TARGET_FIX_ERR_A53_84341 for default.
15998 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16000 * config/sparc/sparc.c (output_return): Output the source location of
16001 the insn in the delay slot, if any.
16002 (output_sibcall): Likewise.
16004 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16007 * common.opt (-static-pie): New alias.
16008 (shared): Negate static-pie.
16009 (-no-pie): Update help text.
16011 (static-pie): New option.
16012 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16013 -static-pie support.
16014 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16015 (LINK_EH_SPEC): Likewise.
16016 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16017 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16018 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16019 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16020 (init_gcc_specs): Likewise.
16021 (init_spec): Likewise.
16022 (display_help): Update help message for -pie.
16023 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16026 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16028 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16029 (movdi_aarch64): Likewise.
16030 (movti_aarch64): Likewise.
16032 2017-09-12 Simon Wright <simon@pushface.org>
16035 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16036 calculation of the minor version, always output as 0.
16038 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16041 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16042 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16043 on it early, rather than manual conversion late. For
16044 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16045 instead of performing manual conversion.
16047 2017-09-12 Carl Love <cel@us.ibm.com>
16049 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16050 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16052 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16053 VMULOSW): Add definitions.
16054 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16055 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16056 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16057 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16058 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16060 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16062 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16064 (movti_aarch64): Likewise.
16065 (movdf_aarch64): Likewise.
16066 (movtf_aarch64): Likewise.
16067 (load_pairdi): Likewise.
16068 (store_pairdi): Likewise.
16069 (load_pairdf): Likewise.
16070 (store_pairdf): Likewise.
16071 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16072 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16073 (ldr_got_small_<mode>): Likewise.
16074 (ldr_got_small_28k_<mode>): Likewise.
16075 (ldr_got_tiny): Likewise.
16076 * config/aarch64/iterators.md (ldst_sz): New.
16077 (ldpstp_sz): Likewise.
16078 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16080 (thunderx_load): Split load_8 to load_16.
16081 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16083 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16084 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16085 (xgene1_store_pair): Split store_8 to store_16.
16086 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16087 (falkor_st_0_st_sd): Split store_8 to store_16.
16089 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16091 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16092 and store1/2/3/4 to store_4/8/12/16.
16093 * config/aarch64/aarch64.md: Update for rename.
16094 * config/arm/arm.md: Likewise.: Likewise.
16095 * config/arm/arm.c: Likewise.
16096 * config/arm/thumb1.md: Likewise.
16097 * config/arm/thumb2.md: Likewise.
16098 * config/arm/vfp.md: Likewise.
16099 * config/arm/arm-generic.md: Likewise.
16100 * config/arm/arm1020e.md: Likewise.
16101 * config/arm/arm1026ejs.md: Likewise.
16102 * config/arm/arm1136jfs.md: Likewise.
16103 * config/arm/arm926ejs.md: Likewise.
16104 * config/arm/cortex-a15.md: Likewise.
16105 * config/arm/cortex-a17.md: Likewise.
16106 * config/arm/cortex-a5.md: Likewise.
16107 * config/arm/cortex-a53.md: Likewise.
16108 * config/arm/cortex-a57.md: Likewise.
16109 * config/arm/cortex-a7.md: Likewise.
16110 * config/arm/cortex-a8.md: Likewise.
16111 * config/arm/cortex-a9.md: Likewise.
16112 * config/arm/cortex-m4.md: Likewise.
16113 * config/arm/cortex-m7.md: Likewise.
16114 * config/arm/cortex-r4.md: Likewise.
16115 * config/arm/exynos-m1.md: Likewise.
16116 * config/arm/fa526.md: Likewise.
16117 * config/arm/fa606te.md: Likewise.
16118 * config/arm/fa626te.md: Likewise.
16119 * config/arm/fa726te.md: Likewise.
16120 * config/arm/fmp626.md: Likewise.
16121 * config/arm/iwmmxt.md: Likewise.
16122 * config/arm/ldmstm.md: Likewise.
16123 * config/arm/marvell-pj4.md: Likewise.
16124 * config/arm/xgene1.md: Likewise.
16125 * config/aarch64/thunderx.md: Likewise.
16126 * config/aarch64/thunderx2t99.md: Likewise.
16127 * config/aarch64/falkor.md: Likewise.
16129 2017-09-12 Martin Liska <mliska@suse.cz>
16131 * attribs.c (private_lookup_attribute): New function.
16132 * attribs.h (private_lookup_attribute): Declared here.
16133 (lookup_attribute): Called from this place.
16135 2017-09-12 Richard Biener <rguenther@suse.de>
16137 PR tree-optimization/82157
16138 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16139 stmts with side-effects.
16141 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16142 Alan Hayward <alan.hayward@arm.com>
16143 David Sherwood <david.sherwood@arm.com>
16145 * target.def (hard_regno_nregs): New hook.
16146 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16147 * targhooks.h (default_hard_regno_nregs): Declare.
16148 * targhooks.c (default_hard_regno_nregs): New function.
16149 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16150 (TARGET_HARD_REGNO_NREGS): ...this hook.
16151 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16152 (CLASS_MAX_NREGS): Likewise.
16153 * doc/tm.texi: Regenerate.
16154 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16155 instead of HARD_REGNO_NREGS.
16156 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16157 HARD_REGNO_NREGS in the comment.
16158 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16159 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16160 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16161 Return an unsigned int.
16162 (TARGET_HARD_REGNO_NREGS): Redefine.
16163 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16164 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16165 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16166 (arc_hard_regno_nregs): New function.
16167 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16168 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16169 (arm_hard_regno_nregs): New function.
16170 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16171 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16172 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16173 (TARGET_HARD_REGNO_NREGS): Redefine.
16174 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16175 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16176 (HARD_REGNO_NREGS): Delete.
16177 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16178 (cr16_hard_regno_nregs): New function.
16179 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16180 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16181 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16182 (cris_hard_regno_nregs): New function.
16183 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16184 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16185 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16186 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16187 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16188 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16189 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16190 (frv_hard_regno_nregs): Make static. Take and return an
16192 (frv_class_max_nregs): Remove outdated copy of documentation.
16193 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16194 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16195 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16196 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16197 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16198 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16199 (TARGET_HARD_REGNO_NREGS): Redefine.
16200 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16201 (CLASS_MAX_NREGS): Update comment.
16202 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16203 (ia64_hard_regno_nregs): New function.
16204 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16205 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16206 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16207 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16208 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16210 (m32c_hard_regno_nregs): Likewise. Make static.
16211 (TARGET_HARD_REGNO_NREGS): Redefine.
16212 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16213 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16214 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16215 (m68k_hard_regno_nregs): New function.
16216 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16217 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16218 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16219 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16220 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16221 Take and return an unsigned int.
16222 (TARGET_HARD_REGNO_NREGS): Redefine.
16223 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16224 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16225 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16226 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16227 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16228 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16229 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16230 (msp430_hard_regno_nregs): Make static. Take and return an
16232 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16233 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16234 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16235 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16236 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16237 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16238 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16239 (TARGET_HARD_REGNO_NREGS): Redefine.
16240 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16241 (PA_HARD_REGNO_NREGS): ...this.
16242 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16243 (PA_HARD_REGNO_NREGS): ...this.
16244 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16245 (pa_hard_regno_nregs): New function.
16246 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16247 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16248 (pdp11_hard_regno_nregs): New function.
16249 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16250 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16251 (rs6000_hard_regno_nregs_hook): New function.
16252 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16253 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16254 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16255 Take and return an unsigned int. Move earlier in file.
16256 (TARGET_HARD_REGNO_NREGS): Redefine.
16257 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16258 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16259 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16260 (rl78_hard_regno_nregs): Make static. Take and return an
16262 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16263 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16264 (rs6000_hard_regno_nregs_hook): New function.
16265 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16266 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16267 (TARGET_HARD_REGNO_NREGS): Redefine.
16268 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16269 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16270 instead of HARD_REGNO_NREGS.
16271 (s390_hard_regno_nregs): New function.
16272 (s390_hard_regno_mode_ok): Add comment from s390.h.
16273 (TARGET_HARD_REGNO_NREGS): Redefine.
16274 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16275 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16276 (sh_hard_regno_nregs): New function.
16277 (sh_pass_in_reg_p): Use it.
16278 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16279 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16280 (sparc_hard_regno_nregs): New function.
16281 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16282 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16283 (spu_function_arg_advance): Use it, supplying a valid register number.
16284 (TARGET_HARD_REGNO_NREGS): Redefine.
16285 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16286 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16287 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16288 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16289 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16290 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16291 (CLASS_MAX_NREGS): Remove copy of old documentation.
16292 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16293 (visium_hard_regno_nregs): New function.
16294 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16295 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16296 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16297 xtensa_hard_regno_nregs): New function.
16298 * system.h (HARD_REGNO_NREGS): Poison.
16300 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16302 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16303 hard_regno_nregs instead of HARD_REGNO_NREGS.
16304 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16305 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16306 (c6x_expand_epilogue): Likewise.
16307 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16308 (frv_read_iacc_argument): Likewise.
16309 * config/sh/sh.c: Include regs.h.
16310 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16311 (regs_used): Likewise.
16312 (output_stack_adjust): Likewise.
16313 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16314 * expmed.c: Include regs.h.
16315 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16316 * ree.c: Include regs.h.
16317 (combine_reaching_defs): Use hard_regno_nregs instead of
16319 (add_removable_extension): Likewise.
16321 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16323 * regs.h (hard_regno_nregs): Turn into a function.
16324 (end_hard_regno): Update accordingly.
16325 * caller-save.c (setup_save_areas): Likewise.
16326 (save_call_clobbered_regs): Likewise.
16327 (replace_reg_with_saved_mem): Likewise.
16328 (insert_restore): Likewise.
16329 (insert_save): Likewise.
16330 * combine.c (can_change_dest_mode): Likewise.
16331 (move_deaths): Likewise.
16332 (distribute_notes): Likewise.
16333 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16334 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16335 (rs6000_split_multireg_move): Likewise.
16336 (rs6000_register_move_cost): Likewise.
16337 (rs6000_memory_move_cost): Likewise.
16338 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16339 (rs6000_split_multireg_move): Likewise.
16340 (rs6000_register_move_cost): Likewise.
16341 (rs6000_memory_move_cost): Likewise.
16342 * cselib.c (cselib_reset_table): Likewise.
16343 (cselib_lookup_1): Likewise.
16344 * emit-rtl.c (set_mode_and_regno): Likewise.
16345 * function.c (aggregate_value_p): Likewise.
16346 * ira-color.c (setup_profitable_hard_regs): Likewise.
16347 (check_hard_reg_p): Likewise.
16348 (calculate_saved_nregs): Likewise.
16349 (assign_hard_reg): Likewise.
16350 (improve_allocation): Likewise.
16351 (calculate_spill_cost): Likewise.
16352 * ira-emit.c (modify_move_list): Likewise.
16353 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16354 (ira_hard_reg_in_set_p): Likewise.
16355 * ira.c (setup_reg_mode_hard_regset): Likewise.
16356 (clarify_prohibited_class_mode_regs): Likewise.
16357 (check_allocation): Likewise.
16358 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16359 (lra_setup_reg_renumber): Likewise.
16360 (setup_try_hard_regno_pseudos): Likewise.
16361 (spill_for): Likewise.
16362 (assign_hard_regno): Likewise.
16363 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16364 * lra-constraints.c (in_class_p): Likewise.
16365 (lra_constraint_offset): Likewise.
16366 (simplify_operand_subreg): Likewise.
16367 (lra_constraints): Likewise.
16368 (split_reg): Likewise.
16369 (split_if_necessary): Likewise.
16370 (invariant_p): Likewise.
16371 (inherit_in_ebb): Likewise.
16372 * lra-lives.c (process_bb_lives): Likewise.
16373 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16374 (get_hard_regs): Likewise.
16375 (do_remat): Likewise.
16376 * lra-spills.c (assign_spill_hard_regs): Likewise.
16377 * mode-switching.c (create_pre_exit): Likewise.
16378 * postreload.c (reload_combine_recognize_pattern): Likewise.
16379 * recog.c (peep2_find_free_register): Likewise.
16380 * regcprop.c (kill_value_regno): Likewise.
16381 (set_value_regno): Likewise.
16382 (copy_value): Likewise.
16383 (maybe_mode_change): Likewise.
16384 (find_oldest_value_reg): Likewise.
16385 (copyprop_hardreg_forward_1): Likewise.
16386 * regrename.c (check_new_reg_p): Likewise.
16387 (regrename_do_replace): Likewise.
16388 * reload.c (push_reload): Likewise.
16389 (combine_reloads): Likewise.
16390 (find_dummy_reload): Likewise.
16391 (operands_match_p): Likewise.
16392 (find_reloads): Likewise.
16393 (find_equiv_reg): Likewise.
16394 (reload_adjust_reg_for_mode): Likewise.
16395 * reload1.c (count_pseudo): Likewise.
16396 (count_spilled_pseudo): Likewise.
16397 (find_reg): Likewise.
16398 (clear_reload_reg_in_use): Likewise.
16399 (free_for_value_p): Likewise.
16400 (allocate_reload_reg): Likewise.
16401 (choose_reload_regs): Likewise.
16402 (reload_adjust_reg_for_temp): Likewise.
16403 (emit_reload_insns): Likewise.
16404 (delete_output_reload): Likewise.
16405 * rtlanal.c (subreg_get_info): Likewise.
16406 * sched-deps.c (sched_analyze_reg): Likewise.
16407 * sel-sched.c (init_regs_for_mode): Likewise.
16408 (mark_unavailable_hard_regs): Likewise.
16409 (choose_best_reg_1): Likewise.
16410 (verify_target_availability): Likewise.
16411 * valtrack.c (dead_debug_insert_temp): Likewise.
16412 * var-tracking.c (track_loc_p): Likewise.
16413 (emit_note_insn_var_location): Likewise.
16414 * varasm.c (make_decl_rtl): Likewise.
16415 * reginfo.c (choose_hard_reg_mode): Likewise.
16416 (init_reg_modes_target): Refer directly to
16417 this_target_regs->x_hard_regno_nregs.
16419 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16421 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16422 instead of hard_regno_nregs.
16424 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16426 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16427 end_hard_regno instead of hard_regno_nregs.
16428 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16429 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16430 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16431 * ira-color.c (improve_allocation): Likewise.
16432 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16433 * lra-lives.c (mark_regno_live): Likewise.
16434 (mark_regno_dead): Likewise.
16435 * lra-remat.c (operand_to_remat): Likewise.
16436 * lra.c (collect_non_operand_hard_regs): Likewise.
16437 * postreload.c (reload_combine_note_store): Likewise.
16438 (move2add_valid_value_p): Likewise.
16439 * reload.c (regno_clobbered_p): Likewise.
16441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16443 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16445 * config/v850/v850.c (v850_reorg): Likewise.
16446 * reload.c (refers_to_regno_for_reload_p): Likewise.
16447 (find_equiv_reg): Likewise.
16448 * reload1.c (reload_reg_reaches_end_p): Likewise.
16450 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16452 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16454 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16455 * config/arm/arm.c (output_move_neon): Likewise.
16456 (arm_attr_length_move_neon): Likewise.
16457 (neon_split_vcombine): Likewise.
16458 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16459 (c6x_mark_reg_written): Likewise.
16460 (c6x_dwarf_register_span): Likewise.
16461 * config/i386/i386.c (ix86_save_reg): Likewise.
16462 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16463 (rws_access_reg): Likewise.
16464 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16465 * mode-switching.c (create_pre_exit): Likewise.
16466 * ree.c (combine_reaching_defs): Likewise.
16467 (add_removable_extension): Likewise.
16468 * regcprop.c (find_oldest_value_reg): Likewise.
16469 (copyprop_hardreg_forward_1): Likewise.
16470 * reload.c (reload_inner_reg_of_subreg): Likewise.
16471 (push_reload): Likewise.
16472 (combine_reloads): Likewise.
16473 (find_dummy_reload): Likewise.
16474 (reload_adjust_reg_for_mode): Likewise.
16475 * reload1.c (find_reload_regs): Likewise.
16476 (forget_old_reloads_1): Likewise.
16477 (reload_reg_free_for_value_p): Likewise.
16478 (reload_adjust_reg_for_temp): Likewise.
16479 (emit_reload_insns): Likewise.
16480 (delete_output_reload): Likewise.
16481 * sel-sched.c (choose_best_reg_1): Likewise.
16482 (choose_best_pseudo_reg): Likewise.
16484 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16485 Alan Hayward <alan.hayward@arm.com>
16486 David Sherwood <david.sherwood@arm.com>
16488 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16489 * target.def (slow_unaligned_access): New hook.
16490 * targhooks.h (default_slow_unaligned_access): Declare.
16491 * targhooks.c (default_slow_unaligned_access): New function.
16492 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16493 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16494 * doc/tm.texi: Regenerate.
16495 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16496 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16497 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16499 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16500 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16502 (rs6000_slow_unaligned_access): New function.
16503 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16504 (expand_block_compare): Likewise.
16505 (expand_strn_compare): Likewise.
16506 (rs6000_rtx_costs): Likewise.
16507 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16508 (riscv_slow_unaligned_access): Likewise.
16509 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16510 (riscv_slow_unaligned_access_p): ...this and make static.
16511 (riscv_option_override): Update accordingly.
16512 (riscv_slow_unaligned_access): New function.
16513 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16514 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16515 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16516 (rs6000_slow_unaligned_access): New function.
16517 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16518 (rs6000_rtx_costs): Likewise.
16519 * config/rs6000/rs6000-string.c (expand_block_compare)
16520 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16521 of SLOW_UNALIGNED_ACCESS.
16522 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16523 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16524 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16525 of SLOW_UNALIGNED_ACCESS.
16526 * expmed.c (simple_mem_bitfield_p): Likewise.
16527 * expr.c (alignment_for_piecewise_move): Likewise.
16528 (emit_group_load_1): Likewise.
16529 (emit_group_store): Likewise.
16530 (copy_blkmode_from_reg): Likewise.
16531 (emit_push_insn): Likewise.
16532 (expand_assignment): Likewise.
16533 (store_field): Likewise.
16534 (expand_expr_real_1): Likewise.
16535 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16536 * lra-constraints.c (simplify_operand_subreg): Likewise.
16537 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16538 * gimple-ssa-store-merging.c: Likewise in block comment at start
16540 * tree-ssa-strlen.c: Include target.h.
16541 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16542 of SLOW_UNALIGNED_ACCESS.
16543 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16545 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16547 PR rtl-optimization/82185
16548 * expmed.c (emit_store_flag_int): Only test tem if it has been
16551 2017-09-12 Richard Biener <rguenther@suse.de>
16553 PR middle-end/82149
16554 * match.pd ((FTYPE) N CMP CST): Fix typo.
16556 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16558 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16560 (mips_near_type_p): Add 'short_call' attribute as a synonym
16562 * doc/extend.texi (short_call): Document new function attribute.
16564 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16567 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16568 assertion check that in the condition.
16569 (get_atomic_generic_size): Likewise. Before testing if parameter
16570 has pointer type, if it has array type, call for C++
16571 default_conversion to perform array-to-pointer conversion.
16573 2017-09-12 Richard Biener <rguenther@suse.de>
16575 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16576 for operations we cannot scalarize.
16578 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16580 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16581 vectors heap vectors. Clean up comments.
16582 Make visited_bbs a reference.
16583 (profitable_jump_thread_path): Make GC
16584 vectors heap vectors. Clean up comments.
16586 (convert_and_register_jump_thread_path): Make GC vectors heap
16588 (check_subpath_and_update_thread_path): Same. Clean up comments.
16589 Make visited_bbs a reference.
16590 (handle_phi): Abstract common code to to
16591 register_jump_thread_path_if_profitable.
16592 Rename VAR_BB to DEF_BB.
16594 Make GC vectors heap vectors.
16595 Make visited_bbs a reference.
16596 (handle_assignment): Same.
16597 (register_jump_thread_path_if_profitable): New.
16598 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16600 Make GC vectors heap vectors. Clean up comments.
16601 Make visited_bbs a reference.
16602 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16603 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16606 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16609 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16610 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16612 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16614 Revert r251800 and r251799.
16616 2017-09-11 Martin Jambor <mjambor@suse.cz>
16619 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16620 arguments in advance.
16621 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16622 use it to find predecessor edges.
16623 (naive_outof_ssa): Collect vector of predecessors.
16625 2017-09-08 Jason Merrill <jason@redhat.com>
16627 PR c++/70029 - ICE with ref-qualifier and -flto
16628 * langhooks.h (struct lang_hooks_for_types): Add
16629 copy_lang_qualifiers.
16630 * attribs.c (build_type_attribute_qual_variant): Use it.
16631 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16633 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16634 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16636 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16639 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16640 (*mulsi3_sp64): New instruction.
16641 (mulsi3): New expander.
16643 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16645 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16647 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16649 * sancov.c: Include memmodel.h.
16651 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16654 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16657 2017-09-07 Carl Love <cel@us.ibm.com>
16659 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16660 the sldi instruction.
16662 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16664 * sancov.c: Include tm_p.h.
16666 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16669 * output.h (switch_to_other_text_partition): New declaration.
16670 * varasm.c (switch_to_other_text_partition): New function.
16671 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16672 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16673 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16674 to the other text partition before emitting LCL label and switch back
16675 after emitting the word after it.
16677 2017-09-07 Richard Biener <rguenther@suse.de>
16679 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16680 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16681 critical edges here, after loop init.
16682 (pass_data_pre): Remove PROP_no_crit_edges flags.
16683 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16684 for valueization of call args to avoid leaking VN_TOP.
16685 (visit_use): Assert we do not visit default defs.
16686 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16687 Use error_mark_node to more easily detect leaking VN_TOP.
16688 All default-defs are varying, not VN_TOP. Mark them visited.
16689 (run_scc_vn): Make code match comment.
16691 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16693 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16694 OPTION_MASK_FLOAT128_KEYWORD.
16695 (POWERPC_MASKS): Likewise.
16696 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16697 support for the -mfloat128-type option, and make -mfloat128
16698 default on PowerPC Linux systems. Define or undefine
16699 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16700 Define __float128 to be __ieee128 if IEEE 128-bit support is
16701 enabled, or undefine it.
16702 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16703 Delete defining __FLOAT128_TYPE__.
16704 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16705 -mfloat128-type option and make -mfloat128 default on PowerPC
16707 (TARGET_FLOAT128_TYPE): Likewise.
16708 (-mfloat128-type): Likewise.
16709 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16710 Delete the -mfloat128-type option and make -mfloat128 default on
16711 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16712 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16713 128-bit floating point is enabled. Change tests from using
16714 -mfloat128-type to -mfloat128.
16715 (rs6000_mangle_type): Use the correct mangling for the __float128
16716 type even if normal long double is restricted to 64-bits.
16717 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16719 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16720 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16721 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16723 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16724 documentation for -mfloat128.
16726 2017-09-06 Olivier Hainque <hainque@adacore.com>
16728 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16730 2017-09-06 Wish Wu <wishwu007@gmail.com>
16731 Jakub Jelinek <jakub@redhat.com>
16733 * asan.c (initialize_sanitizer_builtins): Add
16734 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16735 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16736 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16737 BT_FN_VOID_UINT64_PTR variables.
16738 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16739 (BT_FN_VOID_UINT16_UINT16): Likewise.
16740 (BT_FN_VOID_UINT32_UINT32): Likewise.
16741 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16742 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16743 (BT_FN_VOID_UINT64_PTR): Likewise.
16744 * common.opt (flag_sanitize_coverage): New variable.
16745 (fsanitize-coverage=trace-pc): Remove.
16746 (fsanitize-coverage=): Add.
16747 * flag-types.h (enum sanitize_coverage_code): New enum.
16748 * fold-const.c (fold_range_test): Disable non-short-circuit
16749 optimization if flag_sanitize_coverage.
16750 (fold_truth_andor): Likewise.
16751 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16752 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16753 (coverage_sanitizer_opts): New array.
16754 (get_closest_sanitizer_option): Add OPTS argument, handle also
16755 OPT_fsanitize_coverage_.
16756 (parse_sanitizer_options): Adjusted to also handle
16757 OPT_fsanitize_coverage_.
16758 (common_handle_option): Add OPT_fsanitize_coverage_.
16759 * sancov.c (instrument_comparison, instrument_switch): New function.
16760 (sancov_pass): Add trace-cmp support.
16761 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16762 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16763 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16764 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16765 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16766 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16767 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16768 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16769 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16770 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16772 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16774 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16775 error. Only quit immediately if parsing is complete.
16776 (BEGIN): Initialize fatal_err and parse_done.
16777 (begin fpu, end fpu): Check number of arguments.
16778 (begin arch, end arch): Likewise.
16779 (begin cpu, end cpu): Likewise.
16780 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16781 (optalias): Likewise.
16783 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16785 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16786 * config/arm/arm-isa.h: Delete. Move definitions to ...
16787 * arm-cpus.in: ... here. Use new feature and fgroup values.
16788 * config/arm/arm.c (arm_option_override): Use lower case for feature
16790 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16791 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16792 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16794 (print_isa_bits_for): New function.
16795 (gen_isa): New function.
16796 (gen_comm_data): Use print_isa_bits_for.
16797 (define feature): New keyword.
16798 (define fgroup): New keyword.
16799 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16800 (arm-isa.h): Add rule to generate file.
16801 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16802 case for feature bit names.
16804 2017-09-06 Richard Biener <rguenther@suse.de>
16806 * tree-ssa-pre.c (NECESSARY): Remove.
16807 (create_expression_by_pieces): Do not touch pass-local flags.
16808 (insert_into_preds_of_block): Likewise.
16809 (do_pre_regular_insertion): Likewise.
16810 (eliminate_insert): Likewise.
16811 (eliminate_dom_walker::before_dom_children): Likewise.
16812 (fini_eliminate): Do not look at inserted_exprs.
16813 (mark_operand_necessary): Remove.
16814 (remove_dead_inserted_code): Replace with simple work-list
16815 algorithm based on inserted_exprs and SSA uses.
16816 (pass_pre::execute): Re-order fini_eliminate and
16817 remove_dead_inserted_code.
16819 2017-09-06 Olivier Hainque <hainque@adacore.com>
16821 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16822 for VxWorks 7. Adjust surrounding comments.
16824 2017-09-06 Richard Biener <rguenther@suse.de>
16826 * gimple-ssa-strength-reduction.c
16827 (find_candidates_dom_walker::before_dom_children): Also allow
16830 2017-09-06 Richard Biener <rguenther@suse.de>
16832 PR tree-optimization/82108
16833 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16834 for gap in the non-permutation SLP case.
16836 2017-09-06 Martin Jambor <mjambor@suse.cz>
16838 PR tree-optimization/82078
16839 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16840 add_access_to_work_queue...
16841 (build_accesses_from_assign): ...here.
16842 (propagate_all_subaccesses): Make sure racc is the group
16843 representative, if there is one.
16845 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16847 PR middle-end/82095
16848 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16851 2017-09-06 Richard Biener <rguenther@suse.de>
16853 * gimple-ssa-strength-reduction.c
16854 (find_candidates_dom_walker::before_doom_children): Use a
16855 type and not a mode check.
16857 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16860 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16862 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16864 2017-09-05 Jeff Law <law@redhat.com>
16866 PR tree-optimization/64910
16867 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16868 cases where we have 3 or more operands.
16870 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16872 PR middle-end/81768
16873 * omp-low.c (lower_omp_for): Recompute tree invariant if
16874 gimple_omp_for_initial/final is ADDR_EXPR.
16876 PR middle-end/81768
16877 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16878 into gimple val before gimplification fo the COND_EXPR.
16880 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16882 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16883 REGION_COPY argument.
16884 (thread_through_all_blocks): Remove unused argument to
16885 duplicate_thread_path.
16887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16888 Alan Hayward <alan.hayward@arm.com>
16889 David Sherwood <david.sherwood@arm.com>
16891 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16892 Take a scalar_mode rather than a machine_mode.
16893 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16894 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16895 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16896 (aarch64_gen_adjusted_ldpstp): Likewise.
16897 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16899 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16900 Alan Hayward <alan.hayward@arm.com>
16901 David Sherwood <david.sherwood@arm.com>
16903 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16904 Take a scalar_int_mode instead of a machine_mode.
16905 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16906 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16907 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16908 (aarch64_simd_attr_length_rglist): Delete.
16909 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16910 a scalar_int_mode instead of a machine_mode.
16911 (aarch64_add_offset): Likewise.
16912 (aarch64_internal_mov_immediate): Likewise
16913 (aarch64_add_constant_internal): Likewise.
16914 (aarch64_add_constant): Likewise.
16915 (aarch64_movw_imm): Likewise.
16916 (aarch64_rtx_arith_op_extract_p): Likewise.
16917 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16918 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16919 Remove assert that the mode isn't a vector.
16920 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16921 (aarch64_expand_mov_immediate): Update calls after above changes.
16922 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16923 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16924 (aarch64_move_imm): Likewise.
16925 (aarch64_can_const_movi_rtx_p): Likewise.
16926 (aarch64_strip_extend): Likewise.
16927 (aarch64_extr_rtx_p): Likewise.
16928 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16929 a CONST_INT when the mode parameter is VOIDmode.
16930 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16932 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16934 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16935 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16936 (bitwise_type_for_mode): Update accordingly.
16938 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16940 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16941 * stor-layout.c (mode_for_size_tree): Likewise.
16942 (mode_for_array): Update accordingly.
16943 (layout_decl): Likewise.
16944 (compute_record_mode): Likewise. Only set the mode once.
16946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16948 * target.def (get_mask_mode): Change return type to opt_mode.
16950 * doc/tm.texi: Regenerate.
16951 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16952 * targhooks.c (default_get_mask_mode): Likewise.
16953 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16954 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16955 targetm.get_mask_mode.
16956 * tree.c (build_truth_vector_type): Likewise.
16958 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16960 * machmode.h (mode_for_vector): Return an opt_mode.
16961 * stor-layout.c (mode_for_vector): Likewise.
16962 (mode_for_int_vector): Update accordingly.
16963 (layout_type): Likewise.
16964 * config/i386/i386.c (emit_memmov): Likewise.
16965 (ix86_expand_set_or_movmem): Likewise.
16966 (ix86_expand_vector_init): Likewise.
16967 (ix86_get_mask_mode): Likewise.
16968 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16970 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16971 * expmed.c (extract_bit_field_1): Likewise.
16972 * expr.c (expand_expr_real_2): Likewise.
16973 * optabs-query.c (can_vec_perm_p): Likewise.
16974 (can_vec_mask_load_store_p): Likewise.
16975 * optabs.c (expand_vec_perm): Likewise.
16976 * targhooks.c (default_get_mask_mode): Likewise.
16977 * tree-vect-stmts.c (vectorizable_store): Likewise.
16978 (vectorizable_load): Likewise.
16979 (get_vectype_for_scalar_type_and_size): Likewise.
16981 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16983 * machmode.h (mode_for_int_vector): New function.
16984 * stor-layout.c (mode_for_int_vector): Likewise.
16985 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16986 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16987 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16988 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16989 (s390_expand_vcond): Likewise.
16991 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16993 * machmode.h (opt_machine_mode): New type.
16994 (opt_mode<T>): Allow construction from anything that can be
16996 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16997 (mode_for_size): Return an opt_machine_mode.
16998 * stor-layout.c (mode_for_size): Likewise.
16999 (mode_for_size_tree): Update call accordingly.
17000 (bitwise_mode_for_mode): Likewise.
17001 (make_fract_type): Likewise.
17002 (make_accum_type): Likewise.
17003 * caller-save.c (replace_reg_with_saved_mem): Update call
17005 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17006 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17007 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17008 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17009 * expmed.c (extract_bit_field_1): Likewise.
17010 * reload.c (get_secondary_mem): Likewise.
17011 * varasm.c (assemble_integer): Likewise.
17012 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17015 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17017 * machmode.h (decimal_float_mode_for_size): New function.
17018 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17019 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17020 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17021 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17022 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17023 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17025 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17027 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17028 (get_builtin_sync_mode): Likewise.
17029 (expand_ifn_atomic_compare_exchange): Likewise.
17030 (expand_builtin_atomic_clear): Likewise.
17031 (expand_builtin_atomic_test_and_set): Likewise.
17032 (fold_builtin_atomic_always_lock_free): Likewise.
17033 * calls.c (compute_argument_addresses): Likewise.
17034 (emit_library_call_value_1): Likewise.
17035 (store_one_arg): Likewise.
17036 * combine.c (combine_instructions): Likewise.
17037 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17038 * config/arm/arm.c (arm_function_value): Likewise.
17039 (aapcs_allocate_return_reg): Likewise.
17040 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17041 * config/i386/i386.c (construct_container): Likewise.
17042 (ix86_gimplify_va_arg): Likewise.
17043 (ix86_expand_sse_cmp): Likewise.
17044 (emit_memmov): Likewise.
17045 (emit_memset): Likewise.
17046 (expand_small_movmem_or_setmem): Likewise.
17047 (ix86_expand_pextr): Likewise.
17048 (ix86_expand_pinsr): Likewise.
17049 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17050 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17052 * config/mips/mips.c (mips_function_value_1) Likewise.
17053 (mips_block_move_straight): Likewise.
17054 (mips_expand_ins_as_unaligned_store): Likewise.
17055 * config/powerpcspe/powerpcspe.c
17056 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17057 (rs6000_darwin64_record_arg_flush): Likewise.
17058 * config/rs6000/rs6000.c
17059 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17060 (rs6000_darwin64_record_arg_flush): Likewise.
17061 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17062 (sparc_function_value_1): Likewise.
17063 * config/spu/spu.c (adjust_operand): Likewise.
17064 (spu_emit_branch_or_set): Likewise.
17065 (arith_immediate_p): Likewise.
17066 * emit-rtl.c (gen_lowpart_common): Likewise.
17067 * expr.c (expand_expr_real_1): Likewise.
17068 * function.c (assign_parm_setup_block): Likewise.
17069 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17070 * reload1.c (alter_reg): Likewise.
17071 * stor-layout.c (mode_for_vector): Likewise.
17072 (layout_type): Likewise.
17074 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17076 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17077 (spu_convert_move): Likewise.
17078 * lower-subreg.c (resolve_simple_move): Likewise.
17080 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17083 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17084 define_insn to a define_expand.
17085 (altivec_vsum2sws_direct): New define_insn.
17086 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17088 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17090 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17093 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17098 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17099 realign stack if stack alignment needed is less than incoming
17102 2017-09-05 Marek Polacek <polacek@redhat.com>
17105 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17108 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17110 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17112 2017-09-05 Richard Biener <rguenther@suse.de>
17114 PR tree-optimization/82084
17115 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17117 2017-09-05 Richard Biener <rguenther@suse.de>
17119 PR tree-optimization/82102
17120 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17122 2017-09-05 Martin Liska <mliska@suse.cz>
17124 PR tree-optimization/82032
17125 * tree-cfg.c (generate_range_test): New function.
17126 * tree-cfg.h (generate_range_test): Declared here.
17127 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17128 (cleanup_control_expr_graph): Use it.
17129 * tree-switch-conversion.c (try_switch_expansion): Remove
17131 (emit_case_nodes): Use generate_range_test.
17133 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17136 * config/i386/i386.md (*<btsc><mode>_mask): Add
17137 TARGET_USE_BT to insn constraint.
17138 (*btr<mode>_mask): Ditto.
17140 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17142 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17143 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17145 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17148 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17149 TARGET_NEON and TARGET_IWMMXT.
17150 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17151 TARGET_NEON and TARGET_IWMMXT.
17152 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17154 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17156 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17157 (ix86_rewrite_tls_address): Ditto.
17158 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17159 (ix86_rewrite_tls_address_1): Ditto.
17160 (ix86_rewrite_tls_address): Ditto.
17161 * config/i386/predicates.md (tls_address_pattern): New predicate.
17162 * config/i386/i386.md (TLS address splitter): New splitter.
17164 2017-09-04 Richard Biener <rguenther@suse.de>
17166 PR tree-optimization/82084
17167 * fold-const.h (can_native_encode_string_p): Declare.
17168 * fold-const.c (can_native_encode_string_p): Factor out from ...
17169 (native_encode_string): ... here.
17170 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17171 vectorizing stores from constants we later cannot handle.
17173 2017-09-04 Marek Polacek <polacek@redhat.com>
17176 * doc/invoke.texi: Update -Wtautological-compare documentation.
17178 2017-09-04 Jeff Law <law@redhat.com>
17180 PR tree-optimization/64910
17181 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17182 swap the first and last operand if the last is a constant.
17184 2017-09-04 Marek Polacek <polacek@redhat.com>
17187 * convert.c (do_narrow): When sanitizing signed integer overflows,
17188 bail out for signed types.
17189 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17191 2017-09-04 Richard Biener <rguenther@suse.de>
17193 PR tree-optimization/82060
17194 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17195 Move devirtualization after stmt folding and before EH/AB/noreturn
17196 cleanup to get the stmt refs canonicalized. Use a bool instead
17197 of gimple_modified_p since that doesn't work for NOPs. Schedule
17198 NOPs generated by folding for removal.
17200 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17201 Alan Hayward <alan.hayward@arm.com>
17202 David Sherwood <david.sherwood@arm.com>
17204 * coretypes.h (pad_direction): New enum.
17205 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17206 (FUNCTION_ARG_PADDING): Likewise.
17207 * target.def (function_arg_padding): New hook.
17208 * targhooks.h (default_function_arg_padding): Declare.
17209 * targhooks.c (default_function_arg_padding): New function.
17210 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17211 (TARGET_FUNCTION_ARG_PADDING): ...this.
17212 * doc/tm.texi: Regenerate.
17213 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17214 instead of direction.
17215 (compute_argument_addresses): Likewise.
17216 (load_register_parameters): Likewise.
17217 (emit_library_call_value_1): Likewise.
17218 (store_one_arg): Use targetm.calls.function_arg_padding instead
17219 of FUNCTION_ARG_PADDING.
17220 (must_pass_in_stack_var_size_or_pad): Likewise.
17221 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17222 (emit_group_store): Likewise.
17223 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17224 instead of FUNCTION_ARG_PADDING.
17225 (emit_push_insn): Likewise, and propagate enum change throughout
17227 * function.h (direction): Delete.
17228 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17230 * function.c (assign_parm_find_stack_rtl): Likewise.
17231 (assign_parm_setup_block_p): Likewise.
17232 (assign_parm_setup_block): Likewise.
17233 (gimplify_parameters): Likewise.
17234 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17235 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17237 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17238 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17239 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17240 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17241 (aarch64_function_arg_padding): ...this new function.
17242 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17243 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17244 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17245 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17246 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17247 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17248 (arm_pad_arg_upward): Replace with...
17249 (arm_function_arg_padding): ...this new function.
17250 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17252 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17253 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17254 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17255 (ia64_hpux_function_arg_padding): Replace with...
17256 (ia64_function_arg_padding): ...this new function. Use pad_direction
17257 instead of direction. Check for TARGET_HPUX.
17258 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17259 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17260 (iq2000_function_arg_padding): New function.
17261 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17262 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17263 (mips_function_arg_padding): ...this new function.
17264 (mips_pad_reg_upward): Update accordingly.
17265 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17266 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17267 targetm.calls.function_arg_padding.
17268 (FUNCTION_ARG_PADDING): Delete.
17269 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17270 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17271 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17272 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17273 (nios2_block_reg_padding): Return pad_direction instead of direction.
17274 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17275 instead of direction.
17276 (nios2_function_arg_padding): Likewise. Make static.
17277 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17278 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17279 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17280 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17281 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17282 (pa_function_arg_padding): Make static. Return pad_direction instead
17284 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17285 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17286 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17287 instead of direction. Use targetm.calls.function_arg_padding.
17288 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17289 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17290 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17291 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17292 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17294 (function_arg_padding): Rename to...
17295 (rs6000_function_arg_padding): ...this. Make static. Return
17296 pad_direction instead of direction.
17297 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17298 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17299 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17300 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17301 instead of direction. Use targetm.calls.function_arg_padding.
17302 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17303 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17304 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17305 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17306 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17307 (function_arg_padding): Rename to...
17308 (rs6000_function_arg_padding): ...this. Make static. Return
17309 pad_direction instead of direction.
17310 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17311 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17312 * config/s390/s390.c (s390_function_arg_padding): New function.
17313 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17314 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17315 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17316 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17317 (function_arg_padding): Rename to...
17318 (sparc_function_arg_padding): ...this. Make static. Return
17319 pad_direction instead of direction.
17320 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17321 * config/spu/spu.c (spu_function_arg_padding): New function.
17322 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17323 * system.h (FUNCTION_ARG_PADDING): Poison.
17325 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17326 Alan Hayward <alan.hayward@arm.com>
17327 David Sherwood <david.sherwood@arm.com>
17329 * target.def (modes_tieable_p): New hook.
17330 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17331 (TARGET_MODES_TIEABLE_P): ...this.
17332 * doc/tm.texi.in: Regenerate.
17333 * hooks.h (hook_bool_mode_mode_true): Declare.
17334 * hooks.c (hook_bool_mode_mode_true): New function.
17335 * combine.c (subst): Use targetm.modes_tieable_p instead of
17337 * dse.c (find_shift_sequence): Likewise.
17338 * expmed.c (extract_low_bits): Likewise.
17339 * lower-subreg.c: Include target.h.
17340 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17342 * rtlanal.c (rtx_cost): Likewise.
17343 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17344 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17345 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17346 (TARGET_MODES_TIEABLE_P): Redefine.
17347 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17348 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17349 (TARGET_MODES_TIEABLE_P): Redefine.
17350 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17351 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17352 (arc_modes_tieable_p): New function.
17353 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17354 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17355 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17356 (arm_modes_tieable_p): Make static.
17357 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17358 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17359 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17360 (TARGET_MODES_TIEABLE_P): Redefine.
17361 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17362 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17363 (TARGET_MODES_TIEABLE_P): Redefine.
17364 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17365 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17366 (cr16_modes_tieable_p): New function.
17367 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17368 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17369 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17370 (TRULY_NOOP_TRUNCATION): Update comment.
17371 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17372 (TRULY_NOOP_TRUNCATION): Update comment.
17373 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17374 (frv_modes_tieable_p): New function.
17375 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17376 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17377 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17378 (TARGET_MODES_TIEABLE_P): Redefine.
17379 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17380 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17381 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17382 (TARGET_MODES_TIEABLE_P): Redefine.
17383 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17384 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17385 (ia64_modes_tieable_p): New function.
17386 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17387 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17388 (iq2000_modes_tieable_p): New function.
17389 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17390 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17391 (lm32_modes_tieable_p): New function.
17392 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17393 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17394 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17395 (TARGET_MODES_TIEABLE_P): Redefine.
17396 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17397 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17398 (m32r_modes_tieable_p): New function.
17399 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17400 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17401 (m68k_modes_tieable_p): New function.
17402 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17403 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17404 (mcore_modes_tieable_p): New function.
17405 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17406 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17408 (TARGET_MODES_TIEABLE_P): Redefine.
17409 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17410 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17411 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17412 (TARGET_MODES_TIEABLE_P): Redefine.
17413 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17414 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17415 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17416 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17417 (mn10300_modes_tieable_p): ...this and make static.
17418 (TARGET_MODES_TIEABLE_P): Redefine.
17419 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17420 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17421 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17422 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17423 (msp430_modes_tieable_p): Make static.
17424 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17425 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17426 (TARGET_MODES_TIEABLE_P): Redefine.
17427 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17428 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17429 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17430 (TARGET_MODES_TIEABLE_P): Redefine.
17431 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17432 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17433 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17434 (TARGET_MODES_TIEABLE_P): Redefine.
17435 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17436 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17437 (pdp11_modes_tieable_p): New function.
17438 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17439 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17440 (rs6000_modes_tieable_p): New function.
17441 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17442 * config/powerpcspe/powerpcspe.md: Update comment.
17443 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17444 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17445 (TARGET_MODES_TIEABLE_P): Redefine.
17446 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17447 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17448 (rl78_modes_tieable_p): New function.
17449 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17450 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17451 (rs6000_modes_tieable_p): New function.
17452 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17453 * config/rs6000/rs6000.md: Update comment.
17454 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17455 * config/rx/rx.c (rx_modes_tieable_p): New function.
17456 (TARGET_MODES_TIEABLE_P): Redefine.
17457 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17458 * config/s390/s390.c (s390_modes_tieable_p): New function.
17459 (TARGET_MODES_TIEABLE_P): Redefine.
17460 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17461 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17462 (sh_modes_tieable_p): New function.
17463 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17464 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17465 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17466 (sparc_modes_tieable_p): Make static.
17467 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17468 * config/spu/spu.c (spu_modes_tieable_p): New function.
17469 (TARGET_MODES_TIEABLE_P): Redefine.
17470 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17471 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17472 (TARGET_MODES_TIEABLE_P): Redefine.
17473 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17474 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17475 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17476 * config/v850/v850.c (v850_modes_tieable_p): New function.
17477 (TARGET_MODES_TIEABLE_P): Redefine.
17478 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17479 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17480 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17481 (visium_modes_tieable_p): New function.
17482 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17483 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17484 (xtensa_modes_tieable_p): New function.
17485 * system.h (MODES_TIEABLE_P): Poison.
17487 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17488 Alan Hayward <alan.hayward@arm.com>
17489 David Sherwood <david.sherwood@arm.com>
17491 * target.def (hard_regno_mode_ok): New hook.
17492 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17493 (TARGET_HARD_REGNO_MODE_OK): ...this.
17494 * doc/tm.texi.in: Regenerate.
17495 * hooks.h (hook_bool_uint_mode_true): Declare.
17496 * hooks.c (hook_bool_uint_mode_true): New function.
17497 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17498 HARD_REGNO_MODE_OK.
17499 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17500 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17501 instead of HARD_REGNO_MODE_OK.
17502 * caller-save.c: Include target.h.
17503 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17504 HARD_REGNO_MODE_OK.
17505 * combine.c (can_combine_p): Likewise.
17506 (combinable_i3pat): Likewise.
17507 (can_change_dest_mode): Likewise.
17508 * expr.c (init_expr_target): Likewise.
17509 (convert_move): Likewise.
17510 (convert_modes): Likewise.
17511 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17512 (setup_prohibited_mode_move_regs): Likewise.
17513 * ira.h (target_ira): Likewise.
17514 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17515 * lra-constraints.c (process_alt_operands): Likewise.
17516 (split_reg): Likewise.
17517 * recog.c (peep2_find_free_register): Likewise.
17518 * ree.c (combine_reaching_defs): Likewise.
17519 * regcprop.c (maybe_mode_change): Likewise.
17520 * reginfo.c (init_reg_sets_1): Likewise.
17521 (choose_hard_reg_mode): Likewise.
17522 (simplifiable_subregs): Likewise.
17523 * regrename.c (check_new_reg_p): Likewise.
17524 * reload.c (find_valid_class): Likewise.
17525 (find_valid_class_1): Likewise.
17526 (reload_inner_reg_of_subreg): Likewise.
17527 (push_reload): Likewise.
17528 (combine_reloads): Likewise.
17529 (find_dummy_reload): Likewise.
17530 (find_reloads): Likewise.
17531 * reload1.c (find_reg): Likewise.
17532 (set_reload_reg): Likewise.
17533 (allocate_reload_reg): Likewise.
17534 (choose_reload_regs): Likewise.
17535 (reload_adjust_reg_for_temp): Likewise.
17536 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17537 (simplify_subreg_regno): Likewise.
17538 * sel-sched.c (init_regs_for_mode): Likewise.
17539 * varasm.c (make_decl_rtl): Likewise.
17540 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17541 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17542 HARD_REGNO_MODE_OK.
17543 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17544 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17545 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17546 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17547 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17548 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17549 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17550 (arc_mode_class): Delete.
17551 (HARD_REGNO_MODE_OK): Delete.
17552 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17553 (arc_hard_regno_mode_ok): Rename old array to...
17554 (arc_hard_regno_mode_ok_modes): ...this.
17555 (arc_conditional_register_usage): Update accordingly.
17556 (arc_mode_class): Make static.
17557 (arc_hard_regno_mode_ok): New function.
17558 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17559 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17560 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17561 (arm_hard_regno_mode_ok): Make static.
17562 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17563 HARD_REGNO_MODE_OK.
17564 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17565 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17566 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17568 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17569 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17570 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17571 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17572 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17574 * config/bfin/predicates.md (valid_reg_operand): Use
17575 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17576 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17577 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17578 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17579 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17580 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17581 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17582 (cr16_hard_regno_mode_ok): Make static and return a bool.
17583 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17584 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17585 (cris_hard_regno_mode_ok): New function.
17586 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17587 (epiphany_mode_class): Delete.
17588 (HARD_REGNO_MODE_OK): Delete.
17589 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17590 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17591 (hard_regno_mode_ok): Rename to...
17592 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17593 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17594 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17595 HARD_REGNO_MODE_OK.
17596 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17597 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17598 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17599 (frv_hard_regno_mode_ok): Make static and return a bool.
17600 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17601 HARD_REGNO_MODE_OK.
17602 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17603 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17604 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17605 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17608 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17609 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17610 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17612 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17613 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17614 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17615 (ia64_hard_regno_mode_ok): New function.
17616 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17617 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17618 (iq2000_hard_regno_mode_ok): New function.
17619 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17620 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17621 (lm32_hard_regno_mode_ok): New function.
17622 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17623 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17624 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17625 instead of HARD_REGNO_MODE_OK.
17626 (m32c_hard_regno_ok): Rename to...
17627 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17628 (m32c_cannot_change_mode_class): Update accordingly.
17629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17630 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17631 (m32r_mode_class): Delete.
17632 (HARD_REGNO_MODE_OK): Delete.
17633 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17634 (m32r_hard_regno_mode_ok): Rename to...
17635 (m32r_hard_regno_modes): ...this.
17636 (m32r_mode_class): Make static.
17637 (m32r_hard_regno_mode_ok): New function.
17638 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17639 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17640 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17641 (m68k_hard_regno_mode_ok): Make static.
17642 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17643 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17644 (mcore_hard_regno_mode_ok): New function.
17645 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17646 (HARD_REGNO_MODE_OK): Delete.
17647 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17649 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17650 (microblaze_hard_regno_mode_ok): New function.
17651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17652 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17653 (mips_hard_regno_mode_ok): Delete.
17654 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17655 (mips_hard_regno_mode_ok_p): ...this and make static.
17656 (mips_hard_regno_mode_ok_p): Rename to...
17657 (mips_hard_regno_mode_ok_uncached): ...this.
17658 (mips_hard_regno_mode_ok): New function.
17659 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17660 of HARD_REGNO_MODE_OK.
17661 (mips_option_override): Update after above name changes.
17662 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17663 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17664 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17665 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17666 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17667 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17668 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17669 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17670 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17671 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17672 (msp430_hard_regno_mode_ok): Make static and return a bool.
17673 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17674 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17675 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17678 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17679 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17680 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17681 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17682 (PA_HARD_REGNO_MODE_OK): ...this
17683 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17684 (PA_HARD_REGNO_MODE_OK): ...this.
17685 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17686 (pa_hard_regno_mode_ok): New function.
17687 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17688 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17689 (pdp11_hard_regno_mode_ok): New function.
17690 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17691 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17693 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17696 (rs6000_hard_regno_mode_ok): Rename to...
17697 (rs6000_hard_regno_mode_ok_uncached): ...this.
17698 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17699 (rs6000_hard_regno_mode_ok): New function.
17700 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17701 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17702 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17703 (riscv_hard_regno_mode_ok): ...this and make static.
17704 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17705 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17706 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17707 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17708 (rl78_hard_regno_mode_ok): Make static and return bool.
17709 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17710 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17712 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17713 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17714 (rs6000_hard_regno_mode_ok): Rename to...
17715 (rs6000_hard_regno_mode_ok_uncached): ...this.
17716 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17717 (rs6000_hard_regno_mode_ok): New function.
17718 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17719 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17720 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17721 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17722 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17723 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17725 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17726 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17727 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17728 (sh_hard_regno_mode_ok): Make static.
17729 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17730 instead of HARD_REGNO_MODE_OK.
17731 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17732 (sparc_mode_class): Delete.
17733 (HARD_REGNO_MODE_OK): Delete.
17734 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17735 (hard_regno_mode_classes): Make static.
17736 (sparc_mode_class): Likewise.
17737 (sparc_hard_regno_mode_ok): New function.
17738 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17739 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17740 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17743 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17744 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17745 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17746 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17747 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17748 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17749 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17750 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17751 (visium_hard_regno_mode_ok): New function.
17752 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17753 instead of HARD_REGNO_MODE_OK.
17754 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17755 (HARD_REGNO_MODE_OK): Delete.
17756 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17757 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17758 (xtensa_option_override): Update accordingly.
17759 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17760 (xtensa_hard_regno_mode_ok): New function.
17761 * system.h (HARD_REGNO_MODE_OK): Poison.
17763 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17764 Alan Hayward <alan.hayward@arm.com>
17765 David Sherwood <david.sherwood@arm.com>
17767 * target.def (hard_regno_call_part_clobbered): New hook.
17768 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17769 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17770 * doc/tm.texi: Regenerate.
17771 * hooks.h (hook_bool_uint_mode_false): Declare.
17772 * hooks.c (hook_bool_uint_mode_false): New function.
17773 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17774 * cselib.c (cselib_process_insn): Use
17775 targetm.hard_regno_call_part_clobbered instead of
17776 HARD_REGNO_CALL_PART_CLOBBERED.
17777 * ira-conflicts.c (ira_build_conflicts): Likewise.
17778 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17779 * lra-constraints.c (need_for_call_save_p): Likewise.
17780 * lra-lives.c: Include target.h.
17781 (check_pseudos_live_through_calls): Use
17782 targetm.hard_regno_call_part_clobbered instead of
17783 HARD_REGNO_CALL_PART_CLOBBERED.
17784 * regcprop.c: Include target.h.
17785 (copyprop_hardreg_forward_1): Use
17786 targetm.hard_regno_call_part_clobbered instead of
17787 HARD_REGNO_CALL_PART_CLOBBERED.
17788 * reginfo.c (choose_hard_reg_mode): Likewise.
17789 * regrename.c (check_new_reg_p): Likewise.
17790 * reload.c (find_equiv_reg): Likewise.
17791 * reload1.c (emit_reload_insns): Likewise.
17792 * sched-deps.c (deps_analyze_insn): Likewise.
17793 * sel-sched.c (init_regs_for_mode): Likewise.
17794 (mark_unavailable_hard_regs): Likewise.
17795 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17796 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17797 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17799 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17800 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17801 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17803 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17805 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17806 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17807 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17809 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17810 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17811 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17813 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17814 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17816 * config/powerpcspe/powerpcspe.c
17817 (rs6000_hard_regno_call_part_clobbered): New function.
17818 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17819 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17820 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17822 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17823 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17824 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17826 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17827 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17828 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17830 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17831 Alan Hayward <alan.hayward@arm.com>
17832 David Sherwood <david.sherwood@arm.com>
17834 * rtl.h (subreg_memory_offset): Declare.
17835 * emit-rtl.c (subreg_memory_offset): New function.
17836 * expmed.c (store_bit_field_1): Use it.
17837 * expr.c (undefined_operand_subword_p): Likewise.
17838 * simplify-rtx.c (simplify_subreg): Likewise.
17840 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17842 PR rtl-optimization/57448
17845 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17846 using atomic_load pattern.
17847 (expand_atomic_store): Likewise.
17849 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17852 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17853 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17854 handling. Use replace_call_with_value with NULL instead of
17855 gsi_replace, unlink_stmt_vdef and release_defs.
17857 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17860 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17862 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17865 * rtl.h (emit_library_call_value_1): Declare.
17866 (emit_library_call): Replace declaration with a series of overloads.
17867 Remove the parameter count argument.
17868 (emit_library_call_value): Likewise.
17869 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17870 with an "rtx_mode_t *".
17871 (emit_library_call_value): Delete.
17872 (emit_library_call): Likewise.
17873 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17874 (asan_emit_allocas_unpoison): Likewise.
17875 * builtins.c (expand_builtin_powi): Likewise.
17876 (expand_asan_emit_allocas_unpoison): Likewise.
17877 * cfgexpand.c (expand_main_function): Likewise.
17878 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17879 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17880 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17881 * config/arm/arm.c (arm_trampoline_init): Likewise.
17882 (arm_call_tls_get_addr): Likewise.
17883 (arm_expand_divmod_libfunc): Likewise.
17884 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17885 (smulsi3_highpart): Likewise.
17886 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17887 (c6x_expand_compare): Likewise.
17888 (c6x_expand_movmem): Likewise.
17889 * config/frv/frv.c (frv_trampoline_init): Likewise.
17890 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17891 (ix86_expand_divmod_libfunc): Likewise.
17892 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17893 (ia64_expand_compare): Likewise.
17894 (ia64_profile_hook): Likewise.
17895 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17896 (nonlocal_goto): Likewise.
17897 (restore_stack_nonlocal): Likewise.
17898 * config/m32r/m32r.c (block_move_call): Likewise.
17899 (m32r_trampoline_init): Likewise.
17900 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17901 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17902 (m68k_call_m68k_read_tp): Likewise.
17903 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17904 (microblaze_expand_divide): Likewise.
17905 * config/mips/mips.h (mips_args): Likewise.
17906 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17907 (MIPS_ICACHE_SYNC): Likewise.
17908 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17909 (nios2_trampoline_init): Likewise.
17910 * config/pa/pa.c (hppa_tls_call): Likewise.
17911 (pa_trampoline_init): Likewise.
17912 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17913 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17914 (expand_strn_compare): Likewise.
17915 (rs6000_generate_compare): Likewise.
17916 (rs6000_expand_float128_convert): Likewise.
17917 (output_profile_hook): Likewise.
17918 (rs6000_trampoline_init): Likewise.
17919 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17920 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17921 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17922 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17923 (rs6000_generate_compare): Likewise.
17924 (rs6000_expand_float128_convert): Likewise.
17925 (output_profile_hook): Likewise.
17926 (rs6000_trampoline_init): Likewise.
17927 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17928 * config/sh/sh.c (sh_trampoline_init): Likewise.
17929 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17930 (sparc_emit_float_lib_cmp): Likewise.
17931 (sparc32_initialize_trampoline): Likewise.
17932 (sparc64_initialize_trampoline): Likewise.
17933 (sparc_profile_hook): Likewise.
17934 * config/spu/spu.c (ea_load_store): Likewise.
17935 * config/spu/spu.md (floatunssidf2): Likewise.
17936 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17937 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17938 * config/visium/visium.c (expand_block_move_4): Likewise.
17939 (expand_block_move_2): Likewise.
17940 (expand_block_move_1): Likewise.
17941 (expand_block_set_4): Likewise.
17942 (expand_block_set_2): Likewise.
17943 (expand_block_set_1): Likewise.
17944 (visium_trampoline_init): Likewise.
17945 (visium_profile_hook): Likewise.
17946 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17947 (xtensa_setup_frame_addresses): Likewise.
17948 (xtensa_trampoline_init): Likewise.
17949 * except.c (sjlj_emit_function_enter): Likewise.
17950 (sjlj_emit_function_exit): Likewise.
17951 * explow.c (allocate_dynamic_stack_space): Likewise.
17952 (probe_stack_range): Likewise.
17953 * expr.c (convert_mode_scalar): Likewise.
17954 * optabs.c (expand_binop): Likewise.
17955 (expand_twoval_binop_libfunc): Likewise.
17956 (expand_unop): Likewise.
17957 (prepare_cmp_insn): Likewise.
17958 (prepare_float_lib_cmp): Likewise.
17959 (expand_float): Likewise.
17960 (expand_fix): Likewise.
17961 (expand_fixed_convert): Likewise.
17962 (maybe_emit_sync_lock_test_and_set): Likewise.
17963 (expand_atomic_compare_and_swap): Likewise.
17964 (expand_mem_thread_fence): Likewise.
17965 (expand_atomic_fetch_op): Likewise.
17967 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17969 * doc/generic.texi (OpenACC): Adjust URL.
17970 * doc/invoke.texi (C Dialect Options): Ditto.
17972 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17974 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17975 predicate for operand 1. Add (m,<S>) constraint.
17976 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17977 Prevent memory operand 1 with register operand 2.
17979 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17981 PR rtl-optimization/82024
17982 * combine.c (try_combine): If the combination result is a PARALLEL,
17983 and we only need to retain the SET in there that would be placed
17984 at I2, check that we can place that at I3 instead, before doing so.
17986 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17989 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17991 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17992 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17995 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17996 Jeff Law <law@redhat.com>
17998 * varasm.c (bss_initializer_p): Do not put constants into .bss
17999 (categorize_decl_for_section): Handle bss_initializer_p returning
18000 false when DECL_INITIAL is NULL.
18002 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18005 * config/s390/s390.c (s390_can_inline_p): New function.
18007 2017-09-01 Jeff Law <law@redhat.com>
18009 PR tree-optimization/82052
18010 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18011 Always initialize the returned slot after a hash table miss
18012 when INSERT is true.
18014 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18016 * config/s390/s390.md (mem_signal_fence): Remove.
18017 * doc/md.texi (mem_signal_fence): Remove.
18018 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18020 * target-insns.def (mem_signal_fence): Remove.
18022 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18025 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18028 * asan.c (create_odr_indicator): Strip name encoding from assembler
18029 name before appending it after __odr_asan_.
18031 2017-09-01 Martin Liska <mliska@suse.cz>
18033 PR tree-optimization/82059
18034 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18035 frequency only when an edge is redirected.
18037 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18039 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18040 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18041 (arc_conditional_register_usage): Remove ARC600 lp_count
18043 (arc_file_start): Emit Tag_ARC_CPU_variation.
18044 (arc_can_use_doloop_p): New conditions to use ZOLs.
18045 (hwloop_fail): New function.
18046 (hwloop_optimize): Likewise.
18047 (hwloop_pattern_reg): Likewise.
18048 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18049 (arc_reorg_loops): New function, calls reorg_loops.
18050 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18051 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18053 (arc_loop_hazard): Remove function, functionality moved into
18055 (arc_hazard): Remove arc_loop_hazard call.
18056 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18057 into hwloop_optimize.
18058 (arc_label_align): Remove ZOL handling.
18059 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18060 * config/arc/arc.md (doloop_begin): Remove pattern.
18061 (doloop_begin_i): Likewise.
18062 (doloop_end_i): Likewise.
18063 (doloop_fallback): Likewise.
18064 (doloop_fallback_m): Likewise.
18065 (doloop_end): Reimplement expand.
18066 (arc_lp): New pattern for LP instruction.
18067 (loop_end): New pattern.
18068 (loop_fail): Likewise.
18069 (decrement_and_branch_until_zero): Likewise.
18070 * config/arc/arc.opt (mlpc-width): New option.
18071 * doc/invoke.texi (mlpc-width): Document option.
18073 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18075 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18076 (arc_ccfsm_advance): Fix checking for delay slots.
18077 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18079 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18081 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18083 (movhi_insn): Update store instruction constraint which are saving
18084 6-bit short immediates.
18085 (movsi_insn): Consider also short scaled load operations.
18086 (zero_extendhisi2_i): Use Usd constraint instead of T.
18087 (extendhisi2_i): Add q constraint.
18088 (arc_clzsi2): Add type and length attributes.
18089 (arc_ctzsi2): Likewise.
18090 * config/arc/constraints.md (Usc): Update constraint, the
18091 assembler can parse two relocations for a single instruction.
18093 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18095 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18096 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18098 2017-08-31 Olivier Hainque <hainque@adacore.com>
18100 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18101 match as powerpc-wrs-vxworks*.
18103 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18105 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18106 register constraint for by-element operand.
18107 (aarch64_mls_elt_merge<mode>): Likewise.
18109 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18111 * config/arc/arc.c (arc_can_follow_jump): Check for short
18114 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18116 * config.gcc: Use g.opt for arc.
18117 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18118 functionality moved to ...
18119 (legitimate_scaled_address_p): New function, ...here.
18120 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18121 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18122 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18124 (arc_override_options): Handle G option.
18125 (arc_output_pic_addr_const): Correct function definition.
18126 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18127 (arc_decl_anon_ns_mem_p): Delete.
18128 (arc_in_small_data_p): Overhaul this function to take into
18129 consideration the value given via G option.
18130 (arc_rewrite_small_data_1): Renamed and corrected old
18131 arc_rewrite_small_data function.
18132 (arc_rewrite_small_data): New function.
18133 (small_data_pattern): Don't use pic_offset_table_rtx.
18134 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18135 * config/arc/simdext.md (movmisalignv2hi): Use
18136 prepare_move_operands function.
18138 (movmisalign*): Likewise.
18139 * doc/invoke.texi (ARC options): Document -G option.
18141 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18143 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18145 * config/arc/arc.c (arc_print_operand): Output scalled address for
18146 sdata whenever is possible.
18147 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18148 load/stores are available.
18149 (compact_sda_memory_operand): Check for the alignment required by
18150 code density instructions.
18151 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18153 * config/arc/constraints.md (Usd): Update constraint.
18154 (Us0): New constraint.
18155 (Usc): Update constraint.
18157 2017-08-31 Richard Biener <rguenther@suse.de>
18159 PR middle-end/82054
18160 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18161 function only once.
18163 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18165 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18166 Resize type_signature.
18168 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18169 Alan Hayward <alan.hayward@arm.com>
18170 David Sherwood <david.sherwood@arm.com>
18172 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18173 subregs whose inner modes can be stored in GPRs.
18174 (aarch64_classify_index): Likewise.
18176 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18177 Alan Hayward <alan.hayward@arm.com>
18178 David Sherwood <david.sherwood@arm.com>
18180 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18181 (V_INT_EQUIV): ...this.
18182 (v_cmp_result): Rename to...
18183 (v_int_equiv): ...this.
18184 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18185 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18186 (copysign<mode>3): Likewise.
18187 (aarch64_simd_bsl<mode>_internal): Likewise.
18188 (aarch64_simd_bsl<mode>): Likewise.
18189 (vec_cmp<mode><mode>): Likewise.
18190 (vcond<mode><mode>): Likewise.
18191 (vcond<v_cmp_mixed><mode>): Likewise.
18192 (vcondu<mode><v_cmp_mixed>): Likewise.
18193 (aarch64_cm<optab><mode>): Likewise.
18194 (aarch64_cmtst<mode>): Likewise.
18195 (aarch64_fac<optab><mode>): Likewise.
18196 (vec_perm_const<mode>): Likewise.
18197 (vcond_mask_<mode><v_cmp_result>): Rename to...
18198 (vcond_mask_<mode><v_int_equiv>): ...this.
18199 (vec_cmp<mode><v_cmp_result>): Rename to...
18200 (vec_cmp<mode><v_int_equiv>): ...this.
18202 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18203 Alan Hayward <alan.hayward@arm.com>
18204 David Sherwood <david.sherwood@arm.com>
18206 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18208 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18209 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18210 (UNSPEC_LD4_DREG): New unspecs.
18211 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18212 (aarch64_ld2<mode>_dreg_be): Replace with...
18213 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18215 (aarch64_ld3<mode>_dreg_le)
18216 (aarch64_ld3<mode>_dreg_be): Replace with...
18217 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18219 (aarch64_ld4<mode>_dreg_le)
18220 (aarch64_ld4<mode>_dreg_be): Replace with...
18221 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18224 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18226 PR tree-optimization/81987
18227 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18228 insert an initializer in a location not dominated by the stride
18231 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18233 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18234 on the entire header of the finally block in the fallthru case.
18236 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18238 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18240 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18242 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18243 rs6000_emit_move_from_cr and call renamed function.
18244 (rs6000_emit_prologue): Call renamed functions.
18245 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18246 movesi_from_cr, remove volatile CRs.
18248 2017-08-30 Jon Beniston <jon@beniston.com>
18249 Richard Biener <rguenther@suse.de>
18251 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18252 of VECTOR_MODE_P check.
18253 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18254 element vector types.
18256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18258 * df.h (df_read_modify_subreg_p): Remove in favor of...
18259 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18260 const_rtx instead of an rtx.
18261 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18262 * df-problems.c (df_word_lr_mark_ref): Likewise.
18263 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18264 (mark_pseudo_reg_dead): Likewise.
18265 (mark_ref_dead): Likewise.
18266 * reginfo.c (init_subregs_of_mode): Likewise.
18267 * sched-deps.c (sched_analyze_1): Likewise.
18268 * df-scan.c (df_def_record_1): Likewise.
18269 (df_uses_record): Likewise.
18270 (df_read_modify_subreg_p): Remove in favor of...
18271 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18272 const_rtx instead of an rtx.
18274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18275 Alan Hayward <alan.hayward@arm.com>
18276 David Sherwood <david.sherwood@arm.com>
18278 * rtl.h (partial_subreg_p): New function.
18279 * caller-save.c (save_call_clobbered_regs): Use it.
18280 * calls.c (expand_call): Likewise.
18281 * combine.c (combinable_i3pat): Likewise.
18282 (simplify_set): Likewise.
18283 (make_extraction): Likewise.
18284 (make_compound_operation_int): Likewise.
18285 (gen_lowpart_or_truncate): Likewise.
18286 (force_to_mode): Likewise.
18287 (make_field_assignment): Likewise.
18288 (reg_truncated_to_mode): Likewise.
18289 (record_truncated_value): Likewise.
18290 (move_deaths): Likewise.
18291 * cse.c (record_jump_cond): Likewise.
18292 (cse_insn): Likewise.
18293 * cselib.c (cselib_lookup_1): Likewise.
18294 * expmed.c (extract_bit_field_using_extv): Likewise.
18295 * function.c (assign_parm_setup_reg): Likewise.
18296 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18297 * ira-build.c (create_insn_allocnos): Likewise.
18298 * lra-coalesce.c (merge_pseudos): Likewise.
18299 * lra-constraints.c (match_reload): Likewise.
18300 (simplify_operand_subreg): Likewise.
18301 (curr_insn_transform): Likewise.
18302 * lra-lives.c (process_bb_lives): Likewise.
18303 * lra.c (new_insn_reg): Likewise.
18304 (lra_substitute_pseudo): Likewise.
18305 * regcprop.c (mode_change_ok): Likewise.
18306 (maybe_mode_change): Likewise.
18307 (copyprop_hardreg_forward_1): Likewise.
18308 * reload.c (push_reload): Likewise.
18309 (find_reloads): Likewise.
18310 (find_reloads_subreg_address): Likewise.
18311 * reload1.c (alter_reg): Likewise.
18312 (eliminate_regs_1): Likewise.
18313 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18315 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18317 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18318 back to if statements, including unpack.
18320 2017-08-30 Martin Liska <mliska@suse.cz>
18322 PR inline-asm/82001
18323 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18325 (func_checker::compare_asm_inputs_outputs): ... this function.
18326 (func_checker::compare_gimple_asm): Use the function to compare
18327 also ASM constrains.
18328 * ipa-icf-gimple.h: Rename the function.
18330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18331 Alan Hayward <alan.hayward@arm.com>
18332 David Sherwood <david.sherwood@arm.com>
18334 * coretypes.h (complex_mode): New type.
18335 * gdbhooks.py (build_pretty_printer): Handle it.
18336 * machmode.h (complex_mode): New class.
18337 (complex_mode::includes_p): New function.
18338 (is_complex_int_mode): Likewise.
18339 (is_complex_float_mode): Likewise.
18340 * genmodes.c (get_mode_class): Handle complex mode classes.
18341 * function.c (expand_function_end): Use is_complex_int_mode.
18343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18344 Alan Hayward <alan.hayward@arm.com>
18345 David Sherwood <david.sherwood@arm.com>
18347 * coretypes.h (scalar_mode_pod): New typedef.
18348 * gdbhooks.py (build_pretty_printer): Handle it.
18349 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18350 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18351 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18352 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18353 as_a <scalar_mode>.
18355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18356 Alan Hayward <alan.hayward@arm.com>
18357 David Sherwood <david.sherwood@arm.com>
18359 * machmode.h (mode_for_vector): Take a scalar_mode instead
18361 * stor-layout.c (mode_for_vector): Likewise.
18362 * explow.c (promote_mode): Use as_a <scalar_mode>.
18363 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18366 Alan Hayward <alan.hayward@arm.com>
18367 David Sherwood <david.sherwood@arm.com>
18369 * target.def (preferred_simd_mode): Take a scalar_mode
18370 instead of a machine_mode.
18371 * targhooks.h (default_preferred_simd_mode): Likewise.
18372 * targhooks.c (default_preferred_simd_mode): Likewise.
18373 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18374 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18375 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18376 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18377 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18378 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18379 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18380 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18382 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18383 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18384 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18385 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18386 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18387 * doc/tm.texi: Regenerate.
18388 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18392 Alan Hayward <alan.hayward@arm.com>
18393 David Sherwood <david.sherwood@arm.com>
18395 * target.def (scalar_mode_supported_p): Take a scalar_mode
18396 instead of a machine_mode.
18397 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18398 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18399 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18400 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18401 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18402 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18403 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18404 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18405 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18406 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18407 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18408 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18409 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18410 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18412 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18413 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18414 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18415 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18416 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18418 * doc/tm.texi: Regenerate.
18420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18421 Alan Hayward <alan.hayward@arm.com>
18422 David Sherwood <david.sherwood@arm.com>
18424 * coretypes.h (opt_scalar_mode): New typedef.
18425 * gdbhooks.py (build_pretty_printers): Handle it.
18426 * machmode.h (mode_iterator::get_2xwider): Add overload for
18428 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18430 * expr.c (convert_mode_scalar): Likewise.
18431 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18432 * optabs.c (expand_float): Likewise.
18433 (expand_fix): Likewise.
18434 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18437 Alan Hayward <alan.hayward@arm.com>
18438 David Sherwood <david.sherwood@arm.com>
18440 * optabs.c (expand_float): Explicitly check for scalars before
18441 using a branching expansion.
18442 (expand_fix): Likewise.
18444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18445 Alan Hayward <alan.hayward@arm.com>
18446 David Sherwood <david.sherwood@arm.com>
18448 * expr.c (convert_mode): Split scalar handling out into...
18449 (convert_mode_scalar): ...this new function. Treat the modes
18452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18453 Alan Hayward <alan.hayward@arm.com>
18454 David Sherwood <david.sherwood@arm.com>
18456 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18458 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18461 Alan Hayward <alan.hayward@arm.com>
18462 David Sherwood <david.sherwood@arm.com>
18464 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18465 rather than a machine_mode.
18466 (fixed_from_string): Likewise.
18467 (fixed_convert): Likewise.
18468 (fixed_convert_from_int): Likewise.
18469 (fixed_convert_from_real): Likewise.
18470 (real_convert_from_fixed): Likewise.
18471 * fixed-value.c (fixed_from_double_int): Likewise.
18472 (fixed_from_string): Likewise.
18473 (fixed_convert): Likewise.
18474 (fixed_convert_from_int): Likewise.
18475 (fixed_convert_from_real): Likewise.
18476 (real_convert_from_fixed): Likewise.
18477 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18480 Alan Hayward <alan.hayward@arm.com>
18481 David Sherwood <david.sherwood@arm.com>
18483 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18484 of separate mode class checks. Do not allow vector modes here.
18485 (immed_wide_int_const): Use as_a <scalar_mode>.
18486 * explow.c (trunc_int_for_mode): Likewise.
18487 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18488 (wi::shwi): Likewise.
18489 (wi::min_value): Likewise.
18490 (wi::max_value): Likewise.
18491 * dwarf2out.c (loc_descriptor): Likewise.
18492 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18493 for CONST_WIDE_INT.
18495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18496 Alan Hayward <alan.hayward@arm.com>
18497 David Sherwood <david.sherwood@arm.com>
18499 * tree.h (SCALAR_TYPE_MODE): New macro.
18500 * expr.c (expand_expr_addr_expr_1): Use it.
18501 (expand_expr_real_2): Likewise.
18502 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18503 (fold_convert_const_fixed_from_int): Likewise.
18504 (fold_convert_const_fixed_from_real): Likewise.
18505 (native_encode_fixed): Likewise
18506 (native_encode_complex): Likewise
18507 (native_encode_vector): Likewise.
18508 (native_interpret_fixed): Likewise.
18509 (native_interpret_real): Likewise.
18510 (native_interpret_complex): Likewise.
18511 (native_interpret_vector): Likewise.
18512 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18513 (simd_clone_adjust_argument_types): Likewise.
18514 (simd_clone_init_simd_arrays): Likewise.
18515 (simd_clone_adjust): Likewise.
18516 * stor-layout.c (layout_type): Likewise.
18517 * tree.c (build_minus_one_cst): Likewise.
18518 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18519 * tree-inline.c (estimate_move_cost): Likewise.
18520 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18521 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18522 (vectorizable_reduction): Likewise.
18523 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18524 (vect_recog_mixed_size_cond_pattern): Likewise.
18525 (check_bool_pattern): Likewise.
18526 (adjust_bool_pattern): Likewise.
18527 (search_type_for_mask_1): Likewise.
18528 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18529 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18530 (vectorizable_load): Likewise.
18531 (vectorizable_store): Likewise.
18532 * ubsan.c (ubsan_encode_value): Likewise.
18533 * varasm.c (output_constant): Likewise.
18535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18536 Alan Hayward <alan.hayward@arm.com>
18537 David Sherwood <david.sherwood@arm.com>
18539 * coretypes.h (scalar_mode): New class.
18540 * machmode.h (scalar_mode): Likewise.
18541 (scalar_mode::includes_p): New function.
18542 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18543 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18544 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18545 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18546 * expmed.c (store_bit_field_1): Likewise.
18547 (extract_bit_field_1): Likewise.
18548 * expr.c (write_complex_part): Likewise.
18549 (read_complex_part): Likewise.
18550 (emit_move_complex_push): Likewise.
18551 (expand_expr_real_2): Likewise.
18552 * function.c (assign_parm_setup_reg): Likewise.
18553 (assign_parms_unsplit_complex): Likewise.
18554 * optabs.c (expand_binop): Likewise.
18555 * rtlanal.c (subreg_get_info): Likewise.
18556 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18557 * varasm.c (output_constant_pool_2): Likewise.
18559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18560 Alan Hayward <alan.hayward@arm.com>
18561 David Sherwood <david.sherwood@arm.com>
18563 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18565 (expmed_mult_highpart_optab): Likewise.
18566 (expmed_mult_highpart): Likewise.
18568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18569 Alan Hayward <alan.hayward@arm.com>
18570 David Sherwood <david.sherwood@arm.com>
18572 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18573 instead of a machine_mode.
18574 (builtin_memset_read_str): Likewise.
18575 * builtins.c (c_readstr): Likewise.
18576 (builtin_memcpy_read_str): Likewise.
18577 (builtin_strncpy_read_str): Likewise.
18578 (builtin_memset_read_str): Likewise.
18579 (builtin_memset_gen_str): Likewise.
18580 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18581 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18582 instead of a machine_mode.
18583 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18585 (make_extraction): Likewise.
18586 (try_widen_shift_mode): Take and return scalar_int_modes instead
18588 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18589 a scalar_int_mode instead of a machine_mode.
18590 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18591 (avr_addr_space_pointer_mode): Likewise.
18592 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18593 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18594 (msp430_unwind_word_mode): Likewise.
18595 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18596 (spu_addr_space_pointer_mode): Likewise.
18597 (spu_addr_space_address_mode): Likewise.
18598 (spu_libgcc_cmp_return_mode): Likewise.
18599 (spu_libgcc_shift_count_mode): Likewise.
18600 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18601 (rl78_addr_space_pointer_mode): Likewise.
18602 (fl78_unwind_word_mode): Likewise.
18603 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18605 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18606 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18607 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18608 (mips_valid_pointer_mode): Likewise.
18609 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18610 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18611 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18613 (ft32_addr_space_address_mode): Likewise.
18614 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18615 scalar_int_mode instead of a machine_mode.
18616 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18618 (m32c_addr_space_address_mode): Likewise.
18619 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18620 (rs6000_eh_return_filter_mode): Likewise.
18621 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18622 (rs6000_eh_return_filter_mode): Likewise.
18623 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18624 (s390_libgcc_shift_count_mode): Likewise.
18625 (s390_unwind_word_mode): Likewise.
18626 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18628 * target.def (mode_rep_extended): Likewise.
18629 (valid_pointer_mode): Likewise.
18630 (addr_space.valid_pointer_mode): Likewise.
18631 (eh_return_filter_mode): Return a scalar_int_mode rather than
18633 (libgcc_cmp_return_mode): Likewise.
18634 (libgcc_shift_count_mode): Likewise.
18635 (unwind_word_mode): Likewise.
18636 (addr_space.pointer_mode): Likewise.
18637 (addr_space.address_mode): Likewise.
18638 * doc/tm.texi: Regenerate.
18639 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18641 (do_jump): Use scalar_int_mode for local variables.
18642 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18643 rather than a machine_mode.
18644 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18645 (scompare_loc_descriptor_wide): Likewise.
18646 (scompare_loc_descriptor_narrow): Likewise.
18647 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18649 * except.c (sjlj_emit_dispatch_table): Likewise.
18650 (expand_builtin_eh_copy_values): Likewise.
18651 * explow.c (convert_memory_address_addr_space_1): Likewise.
18652 Take a scalar_int_mode rather than a machine_mode.
18653 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18654 than a machine_mode.
18655 (memory_address_addr_space): Use scalar_int_mode for local variables.
18656 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18657 rather than a machine_mode.
18658 * expmed.c (mask_rtx): Likewise.
18659 (init_expmed_one_conv): Likewise.
18660 (expand_mult_highpart_adjust): Likewise.
18661 (extract_high_half): Likewise.
18662 (expmed_mult_highpart_optab): Likewise.
18663 (expmed_mult_highpart): Likewise.
18664 (expand_smod_pow2): Likewise.
18665 (expand_sdiv_pow2): Likewise.
18666 (emit_store_flag_int): Likewise.
18667 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18669 (extract_low_bits): Likewise.
18670 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18672 * expr.c (pieces_addr::adjust): Likewise.
18673 (can_store_by_pieces): Likewise.
18674 (store_by_pieces): Likewise.
18675 (clear_by_pieces_1): Likewise.
18676 (expand_expr_addr_expr_1): Likewise.
18677 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18678 (expand_expr_real_1): Likewise.
18679 (try_casesi): Likewise.
18680 * final.c (shorten_branches): Likewise.
18681 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18682 type of "mode" to machine_mode.
18683 * internal-fn.c (expand_arith_overflow_result_store): Take a
18684 scalar_int_mode rather than a machine_mode.
18685 (expand_mul_overflow): Use scalar_int_mode for local variables.
18686 * loop-doloop.c (doloop_modify): Likewise.
18687 (doloop_optimize): Likewise.
18688 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18689 than a machine_mode.
18690 (expand_doubleword_shift_condmove): Likewise.
18691 (expand_doubleword_shift): Likewise.
18692 (expand_doubleword_clz): Likewise.
18693 (expand_doubleword_popcount): Likewise.
18694 (expand_doubleword_parity): Likewise.
18695 (expand_absneg_bit): Use scalar_int_mode for local variables.
18696 (prepare_float_lib_cmp): Likewise.
18697 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18698 rather than a machine_mode.
18699 (convert_memory_address_addr_space): Likewise.
18700 (get_mode_bounds): Likewise.
18701 (get_address_mode): Return a scalar_int_mode rather than a
18703 * rtlanal.c (get_address_mode): Likewise.
18704 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18705 than a machine_mode.
18706 * targhooks.c (default_mode_rep_extended): Likewise.
18707 (default_valid_pointer_mode): Likewise.
18708 (default_addr_space_valid_pointer_mode): Likewise.
18709 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18710 than a machine_mode.
18711 (default_libgcc_cmp_return_mode): Likewise.
18712 (default_libgcc_shift_count_mode): Likewise.
18713 (default_unwind_word_mode): Likewise.
18714 (default_addr_space_pointer_mode): Likewise.
18715 (default_addr_space_address_mode): Likewise.
18716 * targhooks.h (default_eh_return_filter_mode): Likewise.
18717 (default_libgcc_cmp_return_mode): Likewise.
18718 (default_libgcc_shift_count_mode): Likewise.
18719 (default_unwind_word_mode): Likewise.
18720 (default_addr_space_pointer_mode): Likewise.
18721 (default_addr_space_address_mode): Likewise.
18722 (default_mode_rep_extended): Take a scalar_int_mode rather than
18724 (default_valid_pointer_mode): Likewise.
18725 (default_addr_space_valid_pointer_mode): Likewise.
18726 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18728 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18729 rather than a machine_mode.
18730 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18731 for local variables.
18732 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18733 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18734 than a machine_mode.
18736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18737 Alan Hayward <alan.hayward@arm.com>
18738 David Sherwood <david.sherwood@arm.com>
18740 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18741 the mode argument to scalar_int_mode.
18742 (do_jump_by_parts_zero_rtx): Likewise.
18743 (do_jump_by_parts_equality_rtx): Likewise.
18744 (do_jump_by_parts_greater): Take a mode argument.
18745 (do_jump_by_parts_equality): Likewise.
18746 (do_jump_1): Update calls accordingly.
18748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18749 Alan Hayward <alan.hayward@arm.com>
18750 David Sherwood <david.sherwood@arm.com>
18752 * is-a.h (safe_dyn_cast): New function.
18753 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18754 (jump_table_for_label): Likewise.
18755 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18756 instead of an rtx_insn *.
18757 (shorten_branches): Use dyn_cast instead of LABEL_P and
18758 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18759 rtx_jump_table_data::get_data_mode.
18760 (final_scan_insn): Likewise.
18762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18763 Alan Hayward <alan.hayward@arm.com>
18764 David Sherwood <david.sherwood@arm.com>
18766 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18767 trying to combine a full-register integer set with a subreg
18770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18771 Alan Hayward <alan.hayward@arm.com>
18772 David Sherwood <david.sherwood@arm.com>
18774 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18775 that is always either address_mode or pointer_mode.
18777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18778 Alan Hayward <alan.hayward@arm.com>
18779 David Sherwood <david.sherwood@arm.com>
18781 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18782 when the two are known to be equal.
18784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18785 Alan Hayward <alan.hayward@arm.com>
18786 David Sherwood <david.sherwood@arm.com>
18788 * simplify-rtx.c (simplify_const_unary_operation): Use
18789 is_a <scalar_int_mode> instead of checking for a nonzero
18790 precision. Forcibly convert op_mode to a scalar_int_mode
18791 in that case. More clearly differentiate the operand and
18792 result modes and use the former when deciding what the value
18793 of a count-bits operation should be. Use is_int_mode instead
18794 of checking for a MODE_INT. Remove redundant check for whether
18795 this mode has a zero precision.
18797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18798 Alan Hayward <alan.hayward@arm.com>
18799 David Sherwood <david.sherwood@arm.com>
18801 * optabs.c (widen_leading): Change the type of the mode argument
18802 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18803 (widen_bswap): Likewise.
18804 (expand_parity): Likewise.
18805 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18806 (expand_ffs): Likewise.
18807 (epand_unop): Check for scalar integer modes before calling the
18810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18811 Alan Hayward <alan.hayward@arm.com>
18812 David Sherwood <david.sherwood@arm.com>
18814 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18816 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18823 value_mode arguments. Use scalar_int_mode internally.
18824 (store_bit_field_1): Rename the new integer mode from imode
18825 to op0_mode and use it instead of GET_MODE (op0). Update calls
18826 to store_split_bit_field, store_bit_field_using_insv and
18827 store_fixed_bit_field.
18828 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18829 Use scalar_int_mode internally. Use a bit count rather than a mode
18830 when calculating the largest bit size for get_best_mode.
18831 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18832 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18833 Remove assertion that OP0 has a scalar integer mode.
18834 (store_split_bit_field): Add op0_mode and value_mode arguments.
18835 Update calls to extract_fixed_bit_field.
18836 (extract_bit_field_using_extv): Add an op0_mode argument.
18837 Use scalar_int_mode internally.
18838 (extract_bit_field_1): Rename the new integer mode from imode to
18839 op0_mode and use it instead of GET_MODE (op0). Update calls to
18840 extract_split_bit_field, extract_bit_field_using_extv and
18841 extract_fixed_bit_field.
18842 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18843 to extract_split_bit_field and extract_fixed_bit_field_1.
18844 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18845 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18846 on the target mode.
18847 (extract_split_bit_field): Add an op0_mode argument. Update call
18848 to extract_fixed_bit_field.
18850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18851 Alan Hayward <alan.hayward@arm.com>
18852 David Sherwood <david.sherwood@arm.com>
18854 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18855 * explow.c (hard_function_value): Likewise.
18856 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18857 convert_to_mode call outside the loop.
18858 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18859 for the mode iterator. Require the mode specified by max_pieces
18861 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18863 (copy_blkmode_to_reg): Likewise.
18864 (set_storage_via_setmem): Likewise.
18865 * optabs.c (prepare_cmp_insn): Likewise.
18866 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18867 * stor-layout.c (finish_bitfield_representative): Likewise.
18869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18870 Alan Hayward <alan.hayward@arm.com>
18871 David Sherwood <david.sherwood@arm.com>
18873 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18874 * expr.c (convert_move): Use them.
18875 (convert_modes): Likewise.
18876 (store_expr_with_bounds): Likewise.
18878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18879 Alan Hayward <alan.hayward@arm.com>
18880 David Sherwood <david.sherwood@arm.com>
18882 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18883 parameter for the mode of "x". Remove the "known_x", "known_mode"
18884 and "known_ret" arguments. Change the type of the mode argument
18885 to scalar_int_mode.
18886 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18887 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18888 (reg_num_sign_bit_copies_for_combine): Likewise.
18889 * rtlanal.c (nonzero_bits1): Likewise.
18890 (num_sign_bit_copies1): Likewise.
18891 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18892 (reg_num_sign_bit_copies_general): Likewise.
18893 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18894 (reg_nonzero_bits_general): Likewise.
18896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18897 Alan Hayward <alan.hayward@arm.com>
18898 David Sherwood <david.sherwood@arm.com>
18900 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18901 than in subroutines. Return 1 for non-integer modes.
18902 (cached_num_sign_bit_copies): Change the type of the mode parameter
18903 to scalar_int_mode.
18904 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18905 classes. Handle CONST_INT_P first and then check whether X also
18906 has a scalar integer mode. Check the same thing for inner registers
18907 of a SUBREG and for values that are being extended or truncated.
18909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18910 Alan Hayward <alan.hayward@arm.com>
18911 David Sherwood <david.sherwood@arm.com>
18913 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18914 in subroutines. Return the mode mask for non-integer modes.
18915 (cached_nonzero_bits): Change the type of the mode parameter
18916 to scalar_int_mode.
18917 (nonzero_bits1): Likewise. Remove early exit for other mode
18918 classes. Handle CONST_INT_P first and then check whether X
18919 also has a scalar integer mode.
18921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18922 Alan Hayward <alan.hayward@arm.com>
18923 David Sherwood <david.sherwood@arm.com>
18925 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18926 Return a scalar_int_mode and assert that the size is greater than
18928 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18929 (op_by_pieces_d::op_by_pieces_d): Likewise.
18930 (op_by_pieces_d::run): Likewise.
18931 (can_store_by_pieces): Likewise.
18933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18934 Alan Hayward <alan.hayward@arm.com>
18935 David Sherwood <david.sherwood@arm.com>
18937 * combine.c (extract_left_shift): Add a mode argument and update
18939 (make_compound_operation_int): Change the type of the mode parameter
18940 to scalar_int_mode and update the call to extract_left_shift.
18942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18943 Alan Hayward <alan.hayward@arm.com>
18944 David Sherwood <david.sherwood@arm.com>
18946 * combine.c (simplify_and_const_int): Change the type of the mode
18947 parameter to scalar_int_mode.
18948 (simplify_and_const_int_1): Likewise. Update recursive call.
18950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18951 Alan Hayward <alan.hayward@arm.com>
18952 David Sherwood <david.sherwood@arm.com>
18954 * combine.c (simplify_compare_const): Check that the mode is a
18955 scalar_int_mode (rather than VOIDmode) before testing its
18957 (simplify_comparison): Move COMPARISON_P handling out of the
18958 loop and restrict the latter part of the loop to scalar_int_modes.
18959 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18960 and when considering SUBREG_REGs. Use is_int_mode instead of
18961 checking GET_MODE_CLASS against MODE_INT.
18963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18964 Alan Hayward <alan.hayward@arm.com>
18965 David Sherwood <david.sherwood@arm.com>
18967 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18968 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18969 shift_unit_mode and for modes involved in scalar shifts.
18971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18972 Alan Hayward <alan.hayward@arm.com>
18973 David Sherwood <david.sherwood@arm.com>
18975 * combine.c (force_int_to_mode): New function, split out from...
18976 (force_to_mode): ...here. Keep xmode up-to-date and use it
18977 instead of GET_MODE (x).
18979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18980 Alan Hayward <alan.hayward@arm.com>
18981 David Sherwood <david.sherwood@arm.com>
18983 * optabs-query.h (extraction_insn::struct_mode): Change type to
18984 opt_scalar_int_mode and update comment.
18985 (extraction_insn::field_mode): Change type to scalar_int_mode.
18986 (extraction_insn::pos_mode): Likewise.
18987 * combine.c (make_extraction): Update accordingly.
18988 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18989 (get_optab_extraction_insn): Likewise.
18990 * recog.c (simplify_while_replacing): Likewise.
18991 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18992 parameter to opt_scalar_int_mode.
18994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18995 Alan Hayward <alan.hayward@arm.com>
18996 David Sherwood <david.sherwood@arm.com>
18998 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18999 to a scalar_int_mode instead of a machine_mode.
19000 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19001 (get_best_mode): Return a boolean and use a pointer argument to store
19002 the selected mode. Replace the limit mode parameter with a bit limit.
19003 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19004 for the values returned by bit_field_mode_iterator::next_mode.
19005 (store_bit_field): Update call to get_best_mode.
19006 (store_fixed_bit_field): Likewise.
19007 (extract_fixed_bit_field): Likewise.
19008 * expr.c (optimize_bitfield_assignment_op): Likewise.
19009 * fold-const.c (optimize_bit_field_compare): Likewise.
19010 (fold_truth_andor_1): Likewise.
19011 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19012 Update for new type of m_mode.
19013 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19020 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19021 (store_bit_field): Check is_a <scalar_int_mode> before calling
19022 strict_volatile_bitfield_p.
19023 (extract_bit_field): Likewise.
19025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19026 Alan Hayward <alan.hayward@arm.com>
19027 David Sherwood <david.sherwood@arm.com>
19029 * target.def (cstore_mode): Return a scalar_int_mode.
19030 * doc/tm.texi: Regenerate.
19031 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19032 * targhooks.h (default_cstore_mode): Likewise.
19033 * targhooks.c (default_cstore_mode): Likewise, using a forced
19035 * expmed.c (emit_cstore): Expect the target of the cstore to be
19038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19039 Alan Hayward <alan.hayward@arm.com>
19040 David Sherwood <david.sherwood@arm.com>
19042 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19044 (niter_desc): Likewise mode.
19045 (iv_analyze): Add a mode parameter.
19047 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19048 and change its type to scalar_int_mode.
19049 * loop-iv.c: Update commentary at head of file.
19050 (iv_constant): Pass the mode paraeter before the rtx it describes
19051 and change its type to scalar_int_mode. Remove VOIDmode handling.
19052 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19053 (iv_extend): Likewise.
19054 (shorten_into_mode): Likewise.
19055 (iv_add): Use scalar_int_mode.
19056 (iv_mult): Likewise.
19057 (iv_shift): Likewise.
19058 (canonicalize_iv_subregs): Likewise.
19059 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19060 it describes and change its mode to scalar_int_mode. Also change
19061 the type of the returned inner_mode to scalar_int_mode.
19062 (get_biv_step): Likewise, turning outer_mode from a pointer
19063 into a direct parameter. Update call to get_biv_step_1.
19064 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19065 iv_constant and get_biv_step.
19066 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19067 and change its type to scalar_int_mode. Don't initialise iv->mode
19068 to VOIDmode and remove later checks for its still being VOIDmode.
19069 Update calls to iv_analyze_op and iv_analyze_expr. Check
19070 is_a <scalar_int_mode> when changing the mode under consideration.
19071 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19072 Update call to iv_analyze_expr.
19073 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19074 inner register is not also a scalar_int_mode. Update call to
19076 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19077 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19078 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19079 separate mode class checks. Update calls to iv_analyze. Remove
19080 fix-up of VOIDmodes after iv_analyze_biv.
19081 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19082 don't have a scalar_int_mode. Update call to biv_p.
19084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19085 Alan Hayward <alan.hayward@arm.com>
19086 David Sherwood <david.sherwood@arm.com>
19088 * cfgexpand.c (convert_debug_memory_address): Use
19089 as_a <scalar_int_mode>.
19090 * combine.c (expand_compound_operation): Likewise.
19091 (make_extraction): Likewise.
19092 (change_zero_ext): Likewise.
19093 (simplify_comparison): Likewise.
19094 * cse.c (cse_insn): Likewise.
19095 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19096 (mem_loc_descriptor): Likewise.
19097 (loc_descriptor): Likewise.
19098 * expmed.c (init_expmed_one_mode): Likewise.
19099 (synth_mult): Likewise.
19100 (emit_store_flag_1): Likewise.
19101 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19102 of a comparison with size.
19103 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19104 (reduce_to_bit_field_precision): Likewise.
19105 * function.c (expand_function_end): Likewise.
19106 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19107 * loop-doloop.c (doloop_modify): Likewise.
19108 * optabs.c (expand_binop): Likewise.
19109 (expand_unop): Likewise.
19110 (expand_copysign_absneg): Likewise.
19111 (prepare_cmp_insn): Likewise.
19112 (maybe_legitimize_operand): Likewise.
19113 * recog.c (const_scalar_int_operand): Likewise.
19114 * rtlanal.c (get_address_mode): Likewise.
19115 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19116 (simplify_cond_clz_ctz): Likewise.
19117 * tree-nested.c (get_nl_goto_field): Likewise.
19118 * tree.c (build_vector_type_for_mode): Likewise.
19119 * var-tracking.c (use_narrower_mode): Likewise.
19121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19122 Alan Hayward <alan.hayward@arm.com>
19123 David Sherwood <david.sherwood@arm.com>
19125 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19126 * builtins.c (expand_builtin_signbit): Use it.
19127 * cfgexpand.c (expand_debug_expr): Likewise.
19128 * dojump.c (do_jump): Likewise.
19129 (do_compare_and_jump): Likewise.
19130 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19131 * expmed.c (make_tree): Likewise.
19132 * expr.c (expand_expr_real_2): Likewise.
19133 (expand_expr_real_1): Likewise.
19134 (try_casesi): Likewise.
19135 * fold-const-call.c (fold_const_call_ss): Likewise.
19136 * fold-const.c (unextend): Likewise.
19137 (extract_muldiv_1): Likewise.
19138 (fold_single_bit_test): Likewise.
19139 (native_encode_int): Likewise.
19140 (native_encode_string): Likewise.
19141 (native_interpret_int): Likewise.
19142 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19143 * internal-fn.c (expand_addsub_overflow): Likewise.
19144 (expand_neg_overflow): Likewise.
19145 (expand_mul_overflow): Likewise.
19146 (expand_arith_overflow): Likewise.
19147 * match.pd: Likewise.
19148 * stor-layout.c (layout_type): Likewise.
19149 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19150 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19151 * tree-ssanames.c (get_range_info): Likewise.
19152 * tree-switch-conversion.c (array_value_type) Likewise.
19153 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19154 (vect_recog_divmod_pattern): Likewise.
19155 (vect_recog_mixed_size_cond_pattern): Likewise.
19156 * tree-vrp.c (extract_range_basic): Likewise.
19157 (simplify_float_conversion_using_ranges): Likewise.
19158 * tree.c (int_fits_type_p): Likewise.
19159 * ubsan.c (instrument_bool_enum_load): Likewise.
19160 * varasm.c (mergeable_string_section): Likewise.
19161 (narrowing_initializer_constant_valid_p): Likewise.
19162 (output_constant): Likewise.
19164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19165 Alan Hayward <alan.hayward@arm.com>
19166 David Sherwood <david.sherwood@arm.com>
19168 * machmode.h (NARROWEST_INT_MODE): New macro.
19169 * expr.c (alignment_for_piecewise_move): Use it instead of
19170 GET_CLASS_NARROWEST_MODE (MODE_INT).
19171 (push_block): Likewise.
19172 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19174 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19177 Alan Hayward <alan.hayward@arm.com>
19178 David Sherwood <david.sherwood@arm.com>
19180 * postreload.c (move2add_valid_value_p): Change the type of the
19181 mode parameter to scalar_int_mode.
19182 (move2add_use_add2_insn): Add a mode parameter and use it instead
19184 (move2add_use_add3_insn): Likewise.
19185 (reload_cse_move2add): Update accordingly.
19187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19188 Alan Hayward <alan.hayward@arm.com>
19189 David Sherwood <david.sherwood@arm.com>
19191 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19193 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19194 * optabs.c (expand_unop): Likewise.
19196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19197 Alan Hayward <alan.hayward@arm.com>
19198 David Sherwood <david.sherwood@arm.com>
19200 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19201 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19202 (popcount_loc_descriptor): Likewise.
19203 (bswap_loc_descriptor): Likewise.
19204 (rotate_loc_descriptor): Likewise.
19205 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19206 calling the functions above.
19208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19209 Alan Hayward <alan.hayward@arm.com>
19210 David Sherwood <david.sherwood@arm.com>
19212 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19214 (try_combine): Likewise.
19215 (simplify_if_then_else): Likewise.
19216 * cse.c (cse_insn): Likewise.
19217 * dwarf2out.c (mem_loc_descriptor): Likewise.
19218 * emit-rtl.c (gen_lowpart_common): Likewise.
19219 * simplify-rtx.c (simplify_truncation): Likewise.
19220 (simplify_binary_operation_1): Likewise.
19221 (simplify_const_relational_operation): Likewise.
19222 (simplify_ternary_operation): Likewise.
19223 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19226 Alan Hayward <alan.hayward@arm.com>
19227 David Sherwood <david.sherwood@arm.com>
19229 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19230 * reload.c (push_reload): Likewise.
19231 (find_reloads): Likewise.
19233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19234 Alan Hayward <alan.hayward@arm.com>
19235 David Sherwood <david.sherwood@arm.com>
19237 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19238 (make_compound_operation_int): Likewise.
19239 (change_zero_ext): Likewise.
19240 * expr.c (convert_move): Likewise.
19241 (convert_modes): Likewise.
19242 * fwprop.c (forward_propagate_subreg): Likewise.
19243 * loop-iv.c (get_biv_step_1): Likewise.
19244 * optabs.c (widen_operand): Likewise.
19245 * postreload.c (move2add_valid_value_p): Likewise.
19246 * recog.c (simplify_while_replacing): Likewise.
19247 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19248 (simplify_binary_operation_1): Likewise. Remove redundant
19249 mode equality check.
19251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19252 Alan Hayward <alan.hayward@arm.com>
19253 David Sherwood <david.sherwood@arm.com>
19255 * combine.c (combine_simplify_rtx): Add checks for
19256 is_a <scalar_int_mode>.
19257 (simplify_if_then_else): Likewise.
19258 (make_field_assignment): Likewise.
19259 (simplify_comparison): Likewise.
19260 * ifcvt.c (noce_try_bitop): Likewise.
19261 * loop-invariant.c (canonicalize_address_mult): Likewise.
19262 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19265 Alan Hayward <alan.hayward@arm.com>
19266 David Sherwood <david.sherwood@arm.com>
19268 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19269 is_a <scalar_int_mode> instead of != BLKmode.
19271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19272 Alan Hayward <alan.hayward@arm.com>
19273 David Sherwood <david.sherwood@arm.com>
19275 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19276 instead of != VOIDmode.
19277 * combine.c (if_then_else_cond): Likewise.
19278 (change_zero_ext): Likewise.
19279 * dwarf2out.c (mem_loc_descriptor): Likewise.
19280 (loc_descriptor): Likewise.
19281 * rtlanal.c (canonicalize_condition): Likewise.
19282 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19285 Alan Hayward <alan.hayward@arm.com>
19286 David Sherwood <david.sherwood@arm.com>
19288 * simplify-rtx.c (simplify_binary_operation_1): Use
19289 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19292 Alan Hayward <alan.hayward@arm.com>
19293 David Sherwood <david.sherwood@arm.com>
19295 * wide-int.h (int_traits<unsigned char>) New class.
19296 (int_traits<unsigned short>) Likewise.
19297 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19298 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19300 * combine.c (set_nonzero_bits_and_sign_copies): Use
19301 is_a <scalar_int_mode>.
19302 (find_split_point): Likewise.
19303 (combine_simplify_rtx): Likewise.
19304 (simplify_logical): Likewise.
19305 (expand_compound_operation): Likewise.
19306 (expand_field_assignment): Likewise.
19307 (make_compound_operation): Likewise.
19308 (extended_count): Likewise.
19309 (change_zero_ext): Likewise.
19310 (simplify_comparison): Likewise.
19311 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19312 (ucompare_loc_descriptor): Likewise.
19313 (minmax_loc_descriptor): Likewise.
19314 (mem_loc_descriptor): Likewise.
19315 (loc_descriptor): Likewise.
19316 * expmed.c (init_expmed_one_mode): Likewise.
19317 * lra-constraints.c (lra_constraint_offset): Likewise.
19318 * optabs.c (prepare_libcall_arg): Likewise.
19319 * postreload.c (move2add_note_store): Likewise.
19320 * reload.c (operands_match_p): Likewise.
19321 * rtl.h (load_extend_op): Likewise.
19322 * rtlhooks.c (gen_lowpart_general): Likewise.
19323 * simplify-rtx.c (simplify_truncation): Likewise.
19324 (simplify_unary_operation_1): Likewise.
19325 (simplify_binary_operation_1): Likewise.
19326 (simplify_const_binary_operation): Likewise.
19327 (simplify_const_relational_operation): Likewise.
19328 (simplify_subreg): Likewise.
19329 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19330 * var-tracking.c (adjust_mems): Likewise.
19331 (prepare_call_arguments): Likewise.
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 * machmode.h (is_int_mode): New fuction.
19338 * combine.c (find_split_point): Use it.
19339 (combine_simplify_rtx): Likewise.
19340 (simplify_if_then_else): Likewise.
19341 (simplify_set): Likewise.
19342 (simplify_shift_const_1): Likewise.
19343 (simplify_comparison): Likewise.
19344 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19345 * cse.c (notreg_cost): Likewise.
19346 (cse_insn): Likewise.
19347 * cselib.c (cselib_lookup_1): Likewise.
19348 * dojump.c (do_jump_1): Likewise.
19349 (do_compare_rtx_and_jump): Likewise.
19350 * dse.c (get_call_args): Likewise.
19351 * dwarf2out.c (rtl_for_decl_init): Likewise.
19352 (native_encode_initializer): Likewise.
19353 * expmed.c (emit_store_flag_1): Likewise.
19354 (emit_store_flag): Likewise.
19355 * expr.c (convert_modes): Likewise.
19356 (store_field): Likewise.
19357 (expand_expr_real_1): Likewise.
19358 * fold-const.c (fold_read_from_constant_string): Likewise.
19359 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19360 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19361 * optabs.c (expand_binop): Likewise.
19362 (expand_unop): Likewise.
19363 (expand_abs_nojump): Likewise.
19364 (expand_one_cmpl_abs_nojump): Likewise.
19365 * simplify-rtx.c (mode_signbit_p): Likewise.
19366 (val_signbit_p): Likewise.
19367 (val_signbit_known_set_p): Likewise.
19368 (val_signbit_known_clear_p): Likewise.
19369 (simplify_relational_operation_1): Likewise.
19370 * tree.c (vector_type_mode): Likewise.
19372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19373 Alan Hayward <alan.hayward@arm.com>
19374 David Sherwood <david.sherwood@arm.com>
19376 * machmode.h (smallest_mode_for_size): Fix formatting.
19377 (smallest_int_mode_for_size): New function.
19378 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19379 instead of smallest_mode_for_size.
19380 * combine.c (make_extraction): Likewise.
19381 * config/arc/arc.c (arc_expand_movmem): Likewise.
19382 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19383 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19384 * config/s390/s390.c (s390_expand_insv): Likewise.
19385 * config/sparc/sparc.c (assign_int_registers): Likewise.
19386 * config/spu/spu.c (spu_function_value): Likewise.
19387 (spu_function_arg): Likewise.
19388 * coverage.c (get_gcov_type): Likewise.
19389 (get_gcov_unsigned_t): Likewise.
19390 * dse.c (find_shift_sequence): Likewise.
19391 * expmed.c (store_bit_field_1): Likewise.
19392 * expr.c (convert_move): Likewise.
19393 (store_field): Likewise.
19394 * internal-fn.c (expand_arith_overflow): Likewise.
19395 * optabs-query.c (get_best_extraction_insn): Likewise.
19396 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19397 * stor-layout.c (layout_type): Likewise.
19398 (initialize_sizetypes): Likewise.
19399 * targhooks.c (default_get_mask_mode): Likewise.
19400 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19403 Alan Hayward <alan.hayward@arm.com>
19404 David Sherwood <david.sherwood@arm.com>
19406 * machmode.h (opt_mode::else_blk): New function.
19407 (int_mode_for_mode): Declare.
19408 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19409 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19411 * cfgexpand.c (expand_debug_expr): Likewise.
19412 * combine.c (gen_lowpart_or_truncate): Likewise.
19413 (gen_lowpart_for_combine): Likewise.
19414 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19415 * config/avr/avr.c (avr_to_int_mode): Likewise.
19416 (avr_out_plus_1): Likewise.
19417 (avr_out_plus): Likewise.
19418 (avr_out_round): Likewise.
19419 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19420 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19421 (s390_expand_vcond): Likewise.
19422 * config/spu/spu.c (spu_split_immediate): Likewise.
19423 (spu_expand_mov): Likewise.
19424 * dse.c (get_stored_val): Likewise.
19425 * expmed.c (store_bit_field_1): Likewise.
19426 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19428 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19429 (extract_low_bits): Likewise.
19430 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19431 handling rather than repeating the check.
19432 (emit_group_store): Likewise.
19433 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19434 * optabs.c (expand_absneg_bit): Likewise.
19435 (expand_copysign_absneg): Likewise.
19436 (expand_copysign_bit): Likewise.
19437 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19438 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19439 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19440 * var-tracking.c (prepare_call_arguments): Likewise.
19441 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19442 int_mode_for_mode instead of mode_for_size.
19443 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19446 Alan Hayward <alan.hayward@arm.com>
19447 David Sherwood <david.sherwood@arm.com>
19449 * machmode.h (int_mode_for_size): New function.
19450 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19451 instead of mode_for_size.
19452 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19454 * combine.c (expand_field_assignment): Use int_mode_for_size
19455 instead of mode_for_size.
19456 (make_extraction): Likewise.
19457 (simplify_shift_const_1): Likewise.
19458 (simplify_comparison): Likewise.
19459 * dojump.c (do_jump): Likewise.
19460 * dwarf2out.c (mem_loc_descriptor): Likewise.
19461 * emit-rtl.c (init_derived_machine_modes): Likewise.
19462 * expmed.c (flip_storage_order): Likewise.
19463 (convert_extracted_bit_field): Likewise.
19464 * expr.c (copy_blkmode_from_reg): Likewise.
19465 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19466 * internal-fn.c (expand_mul_overflow): Likewise.
19467 * lower-subreg.c (simple_move): Likewise.
19468 * optabs-libfuncs.c (init_optabs): Likewise.
19469 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19470 * tree.c (vector_type_mode): Likewise.
19471 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19472 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19473 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19474 * tree-vect-stmts.c (vectorizable_load): Likewise.
19475 (vectorizable_store): Likewise.
19477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19478 Alan Hayward <alan.hayward@arm.com>
19479 David Sherwood <david.sherwood@arm.com>
19481 * coretypes.h (pod_mode): New type.
19482 (scalar_int_mode_pod): New typedef.
19483 * machmode.h (pod_mode): New class.
19484 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19485 * genmodes.c (emit_mode_int_n): Update accordingly.
19486 * lower-subreg.h (target_lower_subreg): Change type to
19487 scalar_int_mode_pod.
19488 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19489 scalar_int_mode_pod.
19491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19492 Alan Hayward <alan.hayward@arm.com>
19493 David Sherwood <david.sherwood@arm.com>
19495 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19496 machine_mode to scalar_int_mode.
19497 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19498 (rs6000_option_override_internal): Remove cast to int.
19499 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19500 machine_mode to scalar_int_mode.
19501 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19502 (rs6000_option_override_internal): Remove cast to int.
19503 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19504 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19506 * config/s390/s390.c (s390_expand_builtin): Likewise.
19507 * coretypes.h (scalar_int_mode): New type.
19508 (opt_scalar_int_mode): New typedef.
19509 * machmode.h (scalar_int_mode): New class.
19510 (scalar_int_mode::includes_p): New function.
19511 (byte_mode): Change type to scalar_int_mode.
19512 (word_mode): Likewise.
19513 (ptr_mode): Likewise.
19514 * emit-rtl.c (byte_mode): Likewise.
19515 (word_mode): Likewise.
19516 (ptr_mode): Likewise.
19517 (init_derived_machine_modes): Update accordingly.
19518 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19519 and MODE_PARTIAL_INT.
19520 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19521 opt_scalar_int_mode.
19523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19524 Alan Hayward <alan.hayward@arm.com>
19525 David Sherwood <david.sherwood@arm.com>
19527 * target.def (libgcc_floating_mode_supported_p): Take a
19529 * doc/tm.texi: Regenerate.
19530 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19532 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19533 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19537 Alan Hayward <alan.hayward@arm.com>
19538 David Sherwood <david.sherwood@arm.com>
19540 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19541 * doc/tm.texi: Regenerate.
19542 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19543 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19544 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19545 * targhooks.h (default_floatn_mode): Likewise.
19546 * targhooks.c (default_floatn_mode): Likewise.
19547 * tree.c (build_common_tree_nodes): Update accordingly.
19549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19550 Alan Hayward <alan.hayward@arm.com>
19551 David Sherwood <david.sherwood@arm.com>
19553 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19554 (mode_iterator::iterate_p): Likewise.
19555 (mode_iterator::get_wider): Likewise.
19556 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19559 Alan Hayward <alan.hayward@arm.com>
19560 David Sherwood <david.sherwood@arm.com>
19562 * coretypes.h (opt_scalar_float_mode): New typedef.
19563 * machmode.h (float_mode_for_size): New function.
19564 * emit-rtl.c (double_mode): Delete.
19565 (init_emit_once): Use float_mode_for_size.
19566 * stor-layout.c (layout_type): Likewise.
19567 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19570 Alan Hayward <alan.hayward@arm.com>
19571 David Sherwood <david.sherwood@arm.com>
19573 * output.h (assemble_real): Take a scalar_float_mode.
19574 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19575 * config/arm/arm.md (consttable_4): Likewise.
19576 (consttable_8): Likewise.
19577 (consttable_16): Likewise.
19578 * config/mips/mips.md (consttable_float): Likewise.
19579 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19580 * varasm.c (assemble_real): Take a scalar_float_mode.
19581 (output_constant_pool_2): Update accordingly.
19582 (output_constant): Likewise.
19584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19585 Alan Hayward <alan.hayward@arm.com>
19586 David Sherwood <david.sherwood@arm.com>
19588 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19589 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19590 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19591 (native_encode_real): Likewise.
19592 (native_interpret_real): Likewise.
19593 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19594 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19597 Alan Hayward <alan.hayward@arm.com>
19598 David Sherwood <david.sherwood@arm.com>
19600 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19601 <scalar_float_mode>. Simplify.
19602 (gen_extend_conv_libfunc): Likewise.
19604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19605 Alan Hayward <alan.hayward@arm.com>
19606 David Sherwood <david.sherwood@arm.com>
19608 * coretypes.h (scalar_float_mode): New type.
19609 * machmode.h (mode_traits::from_int): Use machine_mode if
19610 USE_ENUM_MODES is defined.
19611 (is_a): New function.
19613 (dyn_cast): Likewise.
19614 (scalar_float_mode): New class.
19615 (scalar_float_mode::includes_p): New function.
19616 (is_float_mode): Likewise.
19617 * gdbhooks.py (MachineModePrinter): New class.
19618 (build_pretty_printer): Use it for scalar_float_mode.
19619 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19620 (format_helper::format_helper): Turn into a template.
19621 * genmodes.c (get_mode_class): New function.
19622 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19623 or machine_mode if none.
19624 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19625 as_a <scalar_float_mode>.
19626 * dwarf2out.c (mem_loc_descriptor): Likewise.
19627 (insert_float): Likewise.
19628 (add_const_value_attribute): Likewise.
19629 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19630 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19631 (expand_unop): Update accordingly.
19632 (expand_abs_nojump): Likewise.
19633 (expand_copysign_absneg): Take a scalar_float_mode.
19634 (expand_copysign_bit): Likewise.
19635 (expand_copysign): Update accordingly.
19637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19638 Alan Hayward <alan.hayward@arm.com>
19639 David Sherwood <david.sherwood@arm.com>
19641 * coretypes.h (opt_mode): New class.
19642 * machmode.h (opt_mode): Likewise.
19643 (opt_mode::else_void): New function.
19644 (opt_mode::require): Likewise.
19645 (opt_mode::exists): Likewise.
19646 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19647 (GET_MODE_2XWIDER_MODE): Likewise.
19648 (mode_iterator::get_wider): Update accordingly.
19649 (mode_iterator::get_2xwider): Likewise.
19650 (mode_iterator::get_known_wider): Likewise, turning into a template.
19651 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19652 forcing a wider mode to exist.
19653 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19654 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19655 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19656 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19657 * lower-subreg.c (init_lower_subreg): Likewise.
19658 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19659 on the final iteration.
19660 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19661 a wider mode exists before asking for a move pattern.
19662 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19663 forcing a wider mode to exist.
19664 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19665 returning false if no such mode exists.
19666 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19667 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19668 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19669 Avoid checking for a MODE_INT if we already know the mode is not a
19671 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19672 forcing a wider mode to exist.
19673 (expmed_mult_highpart_optab): Likewise.
19674 (expmed_mult_highpart): Likewise.
19675 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19677 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19678 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19679 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19680 * internal-fn.c (expand_mul_overflow): Update use of
19681 GET_MODE_2XWIDER_MODE.
19682 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19683 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19684 GET_MODE_WIDER_MODE.
19685 (convert_plusminus_to_widen): Likewise.
19686 * tree-switch-conversion.c (array_value_type): Likewise.
19687 * var-tracking.c (emit_note_insn_var_location): Likewise.
19688 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19689 Return false inside rather than outside the loop if no wider mode
19691 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19692 and GET_MODE_2XWIDER_MODE
19693 (can_compare_p): Use else_void.
19694 * gdbhooks.py (OptMachineModePrinter): New class.
19695 (build_pretty_printer): Use it for opt_mode.
19697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19698 Alan Hayward <alan.hayward@arm.com>
19699 David Sherwood <david.sherwood@arm.com>
19701 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19702 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19705 Alan Hayward <alan.hayward@arm.com>
19706 David Sherwood <david.sherwood@arm.com>
19708 * machmode.h (mode_traits): New structure.
19709 (get_narrowest_mode): New function.
19710 (mode_iterator::start): Likewise.
19711 (mode_iterator::iterate_p): Likewise.
19712 (mode_iterator::get_wider): Likewise.
19713 (mode_iterator::get_known_wider): Likewise.
19714 (mode_iterator::get_2xwider): Likewise.
19715 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19716 (FOR_EACH_MODE): Likewise.
19717 (FOR_EACH_MODE_FROM): Likewise.
19718 (FOR_EACH_MODE_UNTIL): Likewise.
19719 (FOR_EACH_WIDER_MODE): Likewise.
19720 (FOR_EACH_2XWIDER_MODE): Likewise.
19721 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19722 * combine.c (simplify_comparison): Likewise
19723 * config/i386/i386.c (type_natural_mode): Likewise.
19724 * cse.c (cse_insn): Likewise.
19725 * dse.c (find_shift_sequence): Likewise.
19726 * emit-rtl.c (init_derived_machine_modes): Likewise.
19727 (init_emit_once): Likewise.
19728 * explow.c (hard_function_value): Likewise.
19729 * expmed.c (extract_fixed_bit_field_1): Likewise.
19730 (extract_bit_field_1): Likewise.
19731 (expand_divmod): Likewise.
19732 (emit_store_flag_1): Likewise.
19733 * expr.c (init_expr_target): Likewise.
19734 (convert_move): Likewise.
19735 (alignment_for_piecewise_move): Likewise.
19736 (widest_int_mode_for_size): Likewise.
19737 (emit_block_move_via_movmem): Likewise.
19738 (copy_blkmode_to_reg): Likewise.
19739 (set_storage_via_setmem): Likewise.
19740 (compress_float_constant): Likewise.
19741 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19742 * optabs-query.c (get_best_extraction_insn): Likewise.
19743 * optabs.c (expand_binop): Likewise.
19744 (expand_twoval_unop): Likewise.
19745 (expand_twoval_binop): Likewise.
19746 (widen_leading): Likewise.
19747 (widen_bswap): Likewise.
19748 (expand_parity): Likewise.
19749 (expand_unop): Likewise.
19750 (prepare_cmp_insn): Likewise.
19751 (prepare_float_lib_cmp): Likewise.
19752 (expand_float): Likewise.
19753 (expand_fix): Likewise.
19754 (expand_sfix_optab): Likewise.
19755 * postreload.c (move2add_use_add2_insn): Likewise.
19756 * reg-stack.c (reg_to_stack): Likewise.
19757 * reginfo.c (choose_hard_reg_mode): Likewise.
19758 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19759 * stor-layout.c (mode_for_size): Likewise.
19760 (smallest_mode_for_size): Likewise.
19761 (mode_for_vector): Likewise.
19762 (finish_bitfield_representative): Likewise.
19763 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19764 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19765 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19766 * var-tracking.c (prepare_call_arguments): Likewise.
19768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19769 Alan Hayward <alan.hayward@arm.com>
19770 David Sherwood <david.sherwood@arm.com>
19772 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19773 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19774 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19775 * machmode.h (mode_size): Move earlier in file.
19776 (mode_precision): Likewise.
19777 (mode_inner): Likewise.
19778 (mode_nunits): Likewise.
19779 (mode_unit_size): Likewise.
19780 (unit_unit_precision): Likewise.
19781 (mode_wider): Likewise.
19782 (mode_2xwider): Likewise.
19783 (machine_mode): New class.
19784 (mode_to_bytes): New function.
19785 (mode_to_bits): Likewise.
19786 (mode_to_precision): Likewise.
19787 (mode_to_inner): Likewise.
19788 (mode_to_unit_size): Likewise.
19789 (mode_to_unit_precision): Likewise.
19790 (mode_to_nunits): Likewise.
19791 (GET_MODE_SIZE): Use mode_to_bytes.
19792 (GET_MODE_BITSIZE): Use mode_to_bits.
19793 (GET_MODE_PRECISION): Use mode_to_precision.
19794 (GET_MODE_INNER): Use mode_to_inner.
19795 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19796 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19797 (GET_MODE_NUNITS): Use mode_to_nunits.
19798 * system.h (ALWAYS_INLINE): New macro.
19799 * config/powerpcspe/powerpcspe-c.c
19800 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19801 int for arg1_mode and arg2_mode.
19803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19804 Alan Hayward <alan.hayward@arm.com>
19805 David Sherwood <david.sherwood@arm.com>
19807 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19808 Prefix mode names with E_ in case statements.
19809 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19810 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19811 (aarch64_split_simd_move): Likewise.
19812 (aarch64_gen_storewb_pair): Likewise.
19813 (aarch64_gen_loadwb_pair): Likewise.
19814 (aarch64_gen_store_pair): Likewise.
19815 (aarch64_gen_load_pair): Likewise.
19816 (aarch64_get_condition_code_1): Likewise.
19817 (aarch64_constant_pool_reload_icode): Likewise.
19818 (get_rsqrte_type): Likewise.
19819 (get_rsqrts_type): Likewise.
19820 (get_recpe_type): Likewise.
19821 (get_recps_type): Likewise.
19822 (aarch64_gimplify_va_arg_expr): Likewise.
19823 (aarch64_simd_container_mode): Likewise.
19824 (aarch64_emit_load_exclusive): Likewise.
19825 (aarch64_emit_store_exclusive): Likewise.
19826 (aarch64_expand_compare_and_swap): Likewise.
19827 (aarch64_gen_atomic_cas): Likewise.
19828 (aarch64_emit_bic): Likewise.
19829 (aarch64_emit_atomic_swap): Likewise.
19830 (aarch64_emit_atomic_load_op): Likewise.
19831 (aarch64_evpc_trn): Likewise.
19832 (aarch64_evpc_uzp): Likewise.
19833 (aarch64_evpc_zip): Likewise.
19834 (aarch64_evpc_ext): Likewise.
19835 (aarch64_evpc_rev): Likewise.
19836 (aarch64_evpc_dup): Likewise.
19837 (aarch64_gen_ccmp_first): Likewise.
19838 (aarch64_gen_ccmp_next): Likewise.
19839 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19840 (alpha_emit_xfloating_libcall): Likewise.
19841 (emit_insxl): Likewise.
19842 (alpha_arg_type): Likewise.
19843 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19844 (arc_preferred_simd_mode): Likewise.
19845 (arc_secondary_reload): Likewise.
19846 (get_arc_condition_code): Likewise.
19847 (arc_print_operand): Likewise.
19848 (arc_legitimate_constant_p): Likewise.
19849 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19850 * config/arc/arc.md (casesi_load): Likewise.
19851 (casesi_compact_jump): Likewise.
19852 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19853 (cc_use_register): Likewise.
19854 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19855 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19856 (arm_init_iwmmxt_builtins): Likewise.
19857 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19858 (neon_expand_vector_init): Likewise.
19859 (arm_attr_length_move_neon): Likewise.
19860 (maybe_get_arm_condition_code): Likewise.
19861 (arm_emit_vector_const): Likewise.
19862 (arm_preferred_simd_mode): Likewise.
19863 (arm_output_iwmmxt_tinsr): Likewise.
19864 (thumb1_output_casesi): Likewise.
19865 (thumb2_output_casesi): Likewise.
19866 (arm_emit_load_exclusive): Likewise.
19867 (arm_emit_store_exclusive): Likewise.
19868 (arm_expand_compare_and_swap): Likewise.
19869 (arm_evpc_neon_vuzp): Likewise.
19870 (arm_evpc_neon_vzip): Likewise.
19871 (arm_evpc_neon_vrev): Likewise.
19872 (arm_evpc_neon_vtrn): Likewise.
19873 (arm_evpc_neon_vext): Likewise.
19874 (arm_validize_comparison): Likewise.
19875 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19876 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19877 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19878 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19879 (c6x_preferred_simd_mode): Likewise.
19880 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19881 (epiphany_rtx_costs): Likewise.
19882 * config/epiphany/predicates.md (proper_comparison_operator):
19884 * config/frv/frv.c (condexec_memory_operand): Likewise.
19885 (frv_emit_move): Likewise.
19886 (output_move_single): Likewise.
19887 (output_condmove_single): Likewise.
19888 (frv_hard_regno_mode_ok): Likewise.
19889 (frv_matching_accg_mode): Likewise.
19890 * config/h8300/h8300.c (split_adds_subs): Likewise.
19891 (h8300_rtx_costs): Likewise.
19892 (h8300_print_operand): Likewise.
19893 (compute_mov_length): Likewise.
19894 (output_logical_op): Likewise.
19895 (compute_logical_op_length): Likewise.
19896 (compute_logical_op_cc): Likewise.
19897 (h8300_shift_needs_scratch_p): Likewise.
19898 (output_a_shift): Likewise.
19899 (compute_a_shift_length): Likewise.
19900 (compute_a_shift_cc): Likewise.
19901 (expand_a_rotate): Likewise.
19902 (output_a_rotate): Likewise.
19903 * config/i386/i386.c (classify_argument): Likewise.
19904 (function_arg_advance_32): Likewise.
19905 (function_arg_32): Likewise.
19906 (function_arg_64): Likewise.
19907 (function_value_64): Likewise.
19908 (ix86_gimplify_va_arg): Likewise.
19909 (ix86_legitimate_constant_p): Likewise.
19910 (put_condition_code): Likewise.
19911 (split_double_mode): Likewise.
19912 (ix86_avx256_split_vector_move_misalign): Likewise.
19913 (ix86_expand_vector_logical_operator): Likewise.
19914 (ix86_split_idivmod): Likewise.
19915 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19916 (ix86_build_const_vector): Likewise.
19917 (ix86_build_signbit_mask): Likewise.
19918 (ix86_match_ccmode): Likewise.
19919 (ix86_cc_modes_compatible): Likewise.
19920 (ix86_expand_branch): Likewise.
19921 (ix86_expand_sse_cmp): Likewise.
19922 (ix86_expand_sse_movcc): Likewise.
19923 (ix86_expand_int_sse_cmp): Likewise.
19924 (ix86_expand_vec_perm_vpermi2): Likewise.
19925 (ix86_expand_vec_perm): Likewise.
19926 (ix86_expand_sse_unpack): Likewise.
19927 (ix86_expand_int_addcc): Likewise.
19928 (ix86_split_to_parts): Likewise.
19929 (ix86_vectorize_builtin_gather): Likewise.
19930 (ix86_vectorize_builtin_scatter): Likewise.
19931 (avx_vpermilp_parallel): Likewise.
19932 (inline_memory_move_cost): Likewise.
19933 (ix86_tieable_integer_mode_p): Likewise.
19934 (x86_maybe_negate_const_int): Likewise.
19935 (ix86_expand_vector_init_duplicate): Likewise.
19936 (ix86_expand_vector_init_one_nonzero): Likewise.
19937 (ix86_expand_vector_init_one_var): Likewise.
19938 (ix86_expand_vector_init_concat): Likewise.
19939 (ix86_expand_vector_init_interleave): Likewise.
19940 (ix86_expand_vector_init_general): Likewise.
19941 (ix86_expand_vector_set): Likewise.
19942 (ix86_expand_vector_extract): Likewise.
19943 (emit_reduc_half): Likewise.
19944 (ix86_emit_i387_round): Likewise.
19945 (ix86_mangle_type): Likewise.
19946 (ix86_expand_round_sse4): Likewise.
19947 (expand_vec_perm_blend): Likewise.
19948 (canonicalize_vector_int_perm): Likewise.
19949 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19950 (expand_vec_perm_1): Likewise.
19951 (expand_vec_perm_interleave3): Likewise.
19952 (expand_vec_perm_even_odd_pack): Likewise.
19953 (expand_vec_perm_even_odd_1): Likewise.
19954 (expand_vec_perm_broadcast_1): Likewise.
19955 (ix86_vectorize_vec_perm_const_ok): Likewise.
19956 (ix86_expand_vecop_qihi): Likewise.
19957 (ix86_expand_mul_widen_hilo): Likewise.
19958 (ix86_expand_sse2_abs): Likewise.
19959 (ix86_expand_pextr): Likewise.
19960 (ix86_expand_pinsr): Likewise.
19961 (ix86_preferred_simd_mode): Likewise.
19962 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19963 * config/i386/sse.md (*andnot<mode>3): Likewise.
19964 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19965 (*<code><mode>3): Likewise.
19966 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19967 (ia64_expand_atomic_op): Likewise.
19968 (ia64_arg_type): Likewise.
19969 (ia64_mode_to_int): Likewise.
19970 (ia64_scalar_mode_supported_p): Likewise.
19971 (ia64_vector_mode_supported_p): Likewise.
19972 (expand_vec_perm_broadcast): Likewise.
19973 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19974 (iq2000_function_arg_advance): Likewise.
19975 (iq2000_function_arg): Likewise.
19976 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19977 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19978 (m68k_libcall_value): Likewise.
19979 (m68k_function_value): Likewise.
19980 (sched_attr_op_type): Likewise.
19981 * config/mcore/mcore.c (mcore_output_move): Likewise.
19982 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19984 (microblaze_function_arg): Likewise.
19985 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19986 (mips_print_operand): Likewise.
19987 (mips_mode_ok_for_mov_fmt_p): Likewise.
19988 (mips_vector_mode_supported_p): Likewise.
19989 (mips_preferred_simd_mode): Likewise.
19990 (mips_expand_vpc_loongson_even_odd): Likewise.
19991 (mips_expand_vec_unpack): Likewise.
19992 (mips_expand_vi_broadcast): Likewise.
19993 (mips_expand_vector_init): Likewise.
19994 (mips_expand_vec_reduc): Likewise.
19995 (mips_expand_msa_cmp): Likewise.
19996 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19997 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19998 (cc_flags_for_mode): Likewise.
19999 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20000 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20001 (nds32_output_casesi_pc_relative): Likewise.
20002 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20003 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20004 (nvptx_gen_unpack): Likewise.
20005 (nvptx_gen_pack): Likewise.
20006 (nvptx_gen_shuffle): Likewise.
20007 (nvptx_gen_wcast): Likewise.
20008 (nvptx_preferred_simd_mode): Likewise.
20009 * config/pa/pa.c (pa_secondary_reload): Likewise.
20010 * config/pa/predicates.md (base14_operand): Likewise.
20011 * config/powerpcspe/powerpcspe-c.c
20012 (altivec_resolve_overloaded_builtin): Likewise.
20013 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20015 (rs6000_preferred_simd_mode): Likewise.
20016 (output_vec_const_move): Likewise.
20017 (rs6000_expand_vector_extract): Likewise.
20018 (rs6000_split_vec_extract_var): Likewise.
20019 (reg_offset_addressing_ok_p): Likewise.
20020 (rs6000_legitimate_offset_address_p): Likewise.
20021 (rs6000_legitimize_address): Likewise.
20022 (rs6000_emit_set_const): Likewise.
20023 (rs6000_const_vec): Likewise.
20024 (rs6000_emit_move): Likewise.
20025 (spe_build_register_parallel): Likewise.
20026 (rs6000_darwin64_record_arg_recurse): Likewise.
20027 (swap_selector_for_mode): Likewise.
20028 (spe_init_builtins): Likewise.
20029 (paired_init_builtins): Likewise.
20030 (altivec_init_builtins): Likewise.
20031 (do_load_for_compare): Likewise.
20032 (rs6000_generate_compare): Likewise.
20033 (rs6000_expand_float128_convert): Likewise.
20034 (emit_load_locked): Likewise.
20035 (emit_store_conditional): Likewise.
20036 (rs6000_output_function_epilogue): Likewise.
20037 (rs6000_handle_altivec_attribute): Likewise.
20038 (rs6000_function_value): Likewise.
20039 (emit_fusion_gpr_load): Likewise.
20040 (emit_fusion_p9_load): Likewise.
20041 (emit_fusion_p9_store): Likewise.
20042 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20043 (fusion_gpr_mem_load): Likewise.
20044 (fusion_addis_mem_combo_load): Likewise.
20045 (fusion_addis_mem_combo_store): Likewise.
20046 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20047 (fusion_gpr_mem_load): Likewise.
20048 (fusion_addis_mem_combo_load): Likewise.
20049 (fusion_addis_mem_combo_store): Likewise.
20050 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20052 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20053 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20054 (rs6000_preferred_simd_mode): Likewise.
20055 (output_vec_const_move): Likewise.
20056 (rs6000_expand_vector_extract): Likewise.
20057 (rs6000_split_vec_extract_var): Likewise.
20058 (reg_offset_addressing_ok_p): Likewise.
20059 (rs6000_legitimate_offset_address_p): Likewise.
20060 (rs6000_legitimize_address): Likewise.
20061 (rs6000_emit_set_const): Likewise.
20062 (rs6000_const_vec): Likewise.
20063 (rs6000_emit_move): Likewise.
20064 (rs6000_darwin64_record_arg_recurse): Likewise.
20065 (swap_selector_for_mode): Likewise.
20066 (paired_init_builtins): Likewise.
20067 (altivec_init_builtins): Likewise.
20068 (rs6000_expand_float128_convert): Likewise.
20069 (emit_load_locked): Likewise.
20070 (emit_store_conditional): Likewise.
20071 (rs6000_output_function_epilogue): Likewise.
20072 (rs6000_handle_altivec_attribute): Likewise.
20073 (rs6000_function_value): Likewise.
20074 (emit_fusion_gpr_load): Likewise.
20075 (emit_fusion_p9_load): Likewise.
20076 (emit_fusion_p9_store): Likewise.
20077 * config/rx/rx.c (rx_gen_move_template): Likewise.
20078 (flags_from_mode): Likewise.
20079 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20080 (s390_slb_comparison): Likewise.
20081 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20082 (s390_vector_mode_supported_p): Likewise.
20083 (s390_cc_modes_compatible): Likewise.
20084 (s390_match_ccmode_set): Likewise.
20085 (s390_canonicalize_comparison): Likewise.
20086 (s390_emit_compare_and_swap): Likewise.
20087 (s390_branch_condition_mask): Likewise.
20088 (s390_rtx_costs): Likewise.
20089 (s390_secondary_reload): Likewise.
20090 (__SECONDARY_RELOAD_CASE): Likewise.
20091 (s390_expand_cs): Likewise.
20092 (s390_preferred_simd_mode): Likewise.
20093 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20094 * config/sh/sh.c (sh_print_operand): Likewise.
20095 (dump_table): Likewise.
20096 (sh_secondary_reload): Likewise.
20097 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20098 * config/sh/sh.md (casesi_worker_1): Likewise.
20099 (casesi_worker_2): Likewise.
20100 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20101 (fcc_comparison_operator): Likewise.
20102 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20103 (emit_soft_tfmode_cvt): Likewise.
20104 (sparc_preferred_simd_mode): Likewise.
20105 (output_cbranch): Likewise.
20106 (sparc_print_operand): Likewise.
20107 (sparc_expand_vec_perm_bmask): Likewise.
20108 (vector_init_bshuffle): Likewise.
20109 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20110 (spu_vector_mode_supported_p): Likewise.
20111 (spu_expand_insv): Likewise.
20112 (spu_emit_branch_or_set): Likewise.
20113 (spu_handle_vector_attribute): Likewise.
20114 (spu_builtin_splats): Likewise.
20115 (spu_builtin_extract): Likewise.
20116 (spu_builtin_promote): Likewise.
20117 (spu_expand_sign_extend): Likewise.
20118 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20119 (tilegx_simd_int): Likewise.
20120 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20121 (tilepro_simd_int): Likewise.
20122 * config/v850/v850.c (const_double_split): Likewise.
20123 (v850_print_operand): Likewise.
20124 (ep_memory_offset): Likewise.
20125 * config/vax/vax.c (vax_rtx_costs): Likewise.
20126 (vax_output_int_move): Likewise.
20127 (vax_output_int_add): Likewise.
20128 (vax_output_int_subtract): Likewise.
20129 * config/visium/predicates.md (visium_branch_operator): Likewise.
20130 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20131 (visium_print_operand_address): Likewise.
20132 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20133 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20134 (xtensa_expand_conditional_branch): Likewise.
20135 (xtensa_copy_incoming_a7): Likewise.
20136 (xtensa_output_literal): Likewise.
20137 * dfp.c (decimal_real_maxval): Likewise.
20138 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20141 Alan Hayward <alan.hayward@arm.com>
20142 David Sherwood <david.sherwood@arm.com>
20144 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20145 (mode_nunits_inline): Likewise.
20146 (mode_inner_inline): Likewise.
20147 (mode_unit_size_inline): Likewise.
20148 (mode_unit_precision_inline): Likewise.
20149 (emit_insn_modes_h): Likewise. Also emit a #define of the
20151 (emit_mode_wider): Add an E_ prefix to mode names.
20152 (emit_mode_complex): Likewise.
20153 (emit_mode_inner): Likewise.
20154 (emit_mode_adjustments): Likewise.
20155 (emit_mode_int_n): Likewise.
20156 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20157 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20158 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20159 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20160 (CRC32_BUILTIN, ENTRY): Likewise.
20161 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20162 (aarch64_pop_regs): Likewise.
20163 (aarch64_process_components): Likewise.
20164 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20165 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20166 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20167 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20168 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20169 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20171 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20172 (pcmpistr): Likewise.
20173 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20174 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20175 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20177 * config/rl78/rl78.c (mduc_regs): Likewise.
20178 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20179 (htm_expand_builtin): Likewise.
20180 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20181 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20182 * config/xtensa/xtensa.c (print_operand): Likewise.
20183 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20184 (NUM_MODE_VECTOR_INT): Likewise.
20185 * genoutput.c (null_operand): Likewise.
20186 (output_operand_data): Likewise.
20187 * genrecog.c (print_parameter_value): Likewise.
20188 * lra.c (debug_operand_data): Likewise.
20190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20192 * dwarf2out.c (scompare_loc_descriptor_wide)
20193 (scompare_loc_descriptor_narrow): New functions, split out from...
20194 (scompare_loc_descriptor): ...here.
20195 * expmed.c (emit_store_flag_int): New function, split out from...
20196 (emit_store_flag): ...here.
20198 2017-08-30 Richard Biener <rguenther@suse.de>
20200 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20201 (dwarf2out_early_finish): Move setting of AT_pubnames from
20202 early debug output to early finish.
20204 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20206 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20207 and -mdata-region to the assembler.
20209 2017-08-30 Richard Biener <rguenther@suse.de>
20211 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20213 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20214 (dwarf2out_early_global_decl): Only generate a DIE for the
20215 abstract origin if it doesn't already exist or is a declaration DIE.
20216 (resolve_addr): Do not add the linkage name twice when
20217 generating a stub DIE for the DW_TAG_GNU_call_site target.
20219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20222 Use machine_mode rather than int for arg1_mode.
20224 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20227 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20228 that the second argument of the built-in functions to unpack
20229 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20230 switch statement instead a lot of if statements.
20231 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20232 Allow 64-bit values to be in Altivec registers as well as
20233 traditional floating point registers.
20234 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20236 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20238 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20239 MAX_REGS_PER_ADDRESS == 1.
20241 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20243 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20244 * config/i386/i386.c (ix86_option_override_internal): Simplify
20245 setting of opts->x_flag_entry.
20247 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20248 Jakub Jelinek <jakub@redhat.com>
20249 Richard Biener <rguenther@suse.de>
20251 PR tree-optimization/81503
20252 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20253 folded constant fits in the target type; reorder tests for clarity.
20255 2017-08-29 Martin Liska <mliska@suse.cz>
20257 * passes.def: Include pass_lower_switch.
20258 * stmt.c (dump_case_nodes): Remove and move to
20259 tree-switch-conversion.
20260 (case_values_threshold): Likewise.
20261 (expand_switch_as_decision_tree_p): Likewise.
20262 (emit_case_decision_tree): Likewise.
20263 (expand_case): Likewise.
20264 (balance_case_nodes): Likewise.
20265 (node_has_low_bound): Likewise.
20266 (node_has_high_bound): Likewise.
20267 (node_is_bounded): Likewise.
20268 (emit_case_nodes): Likewise.
20269 (struct simple_case_node): New struct.
20270 (add_case_node): Remove.
20271 (emit_case_dispatch_table): Use vector instead of case_list.
20272 (reset_out_edges_aux): Remove.
20273 (compute_cases_per_edge): Likewise.
20274 (expand_case): Build list of simple_case_node.
20275 (expand_sjlj_dispatch_table): Use it.
20276 * tree-switch-conversion.c (struct case_node): Moved from
20277 stmt.c and adjusted.
20278 (emit_case_nodes): Likewise.
20279 (node_has_low_bound): Likewise.
20280 (node_has_high_bound): Likewise.
20281 (node_is_bounded): Likewise.
20282 (case_values_threshold): Likewise.
20283 (reset_out_edges_aux): Likewise.
20284 (compute_cases_per_edge): Likewise.
20285 (add_case_node): Likewise.
20286 (dump_case_nodes): Likewise.
20287 (balance_case_nodes): Likewise.
20288 (expand_switch_as_decision_tree_p): Likewise.
20289 (emit_jump): Likewise.
20290 (emit_case_decision_tree): Likewise.
20291 (try_switch_expansion): Likewise.
20292 (do_jump_if_equal): Likewise.
20293 (emit_cmp_and_jump_insns): Likewise.
20294 (fix_phi_operands_for_edge): New function.
20295 (record_phi_operand_mapping): Likewise.
20296 (class pass_lower_switch): New pass.
20297 (pass_lower_switch::execute): New function.
20298 (make_pass_lower_switch): Likewise.
20299 (conditional_probability):
20300 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20301 * tree-pass.h: Add make_pass_lower_switch.
20303 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20306 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20309 2017-08-29 Richard Biener <rguenther@suse.de>
20311 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20312 we do not add a DW_AT_inline attribute twice.
20313 (gen_subprogram_die): Remove code setting DW_AT_inline on
20314 DECL_ABSTRACT_P nodes.
20316 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20318 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20319 calls to internal functions.
20320 (gimplify_modify_expr): Likewise.
20321 * tree-call-cdce.c (use_internal_fn): Likewise.
20322 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20323 (convert_to_divmod): Set the nothrow flag.
20324 * tree-if-conv.c (predicate_mem_writes): Likewise.
20325 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20326 (vectorizable_call): Likewise.
20327 (vectorizable_store): Likewise.
20328 (vectorizable_load): Likewise.
20329 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20330 (vect_recog_mask_conversion_pattern): Likewise.
20332 2017-08-29 Martin Liska <mliska@suse.cz>
20335 * gcc.c (driver_handle_option): Add new argument.
20336 * opts-common.c (handle_option): Pass
20337 target_option_override_hook.
20338 * opts-global.c (lang_handle_option): Add new option.
20339 (set_default_handlers): Add new argument.
20340 (decode_options): Likewise.
20341 * opts.c (target_handle_option): Likewise.
20342 (common_handle_option): Call target_option_override_hook.
20343 * opts.h (struct cl_option_handler_func): Add hook for
20344 target option override.
20345 (struct cl_option_handlers): Likewise.
20346 (set_default_handlers): Add new argument.
20347 (decode_options): Likewise.
20348 (common_handle_option): Likewise.
20349 (target_handle_option): Likewise.
20350 * toplev.c (toplev::main): Pass targetm.target_option.override
20353 2017-08-29 Richard Biener <rguenther@suse.de>
20354 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20356 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20357 life to the active subtree.
20359 2017-08-28 Jeff Law <law@redhat.com>
20361 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20362 derive_equivalences.
20363 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20364 Code moved into....
20365 (edge_info::derive_equivalences): New private member function
20367 * tree-ssa-dom.c (class edge_info): Changed from a struct
20368 to a class. Add ctor/dtor, methods and data members.
20369 (edge_info::edge_info): Renamed from allocate_edge_info.
20370 Initialize additional members.
20371 (edge_info::~edge_info): New.
20372 (free_dom_edge_info): Delete the edge info.
20373 (record_edge_info): Use new class & associated member functions.
20374 Tighten forms for testing for edge equivalences.
20375 (record_temporary_equivalences): Iterate over the simple
20376 equivalences rather than assuming there's only one per edge.
20377 (cprop_into_successor_phis): Iterate over the simple
20378 equivalences rather than assuming there's only one per edge.
20379 (optimize_stmt): Use operand_equal_p rather than pointer
20380 equality for mini-DSE code.
20382 2017-08-28 Nathan Sidwell <nathan@acm.org>
20384 * gcc.c (execute): Fold SIGPIPE handling into switch
20385 statement. Adjust internal error message.
20387 2017-08-28 Richard Biener <rguenther@suse.de>
20390 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20391 Do nothing for removed DIEs.
20393 2017-08-28 Richard Biener <rguenther@suse.de>
20395 PR tree-optimization/81977
20396 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20399 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20402 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20403 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20404 using targetm.gen_mem_thread_fence.
20406 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20409 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20410 predicate to register_operand. Reorder operands.
20411 (*btr<mode>): Ditto.
20412 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20413 (*btr<mode>_mask): Ditto.
20415 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20417 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20418 * config/rs6000/xmmintrin.h: New file.
20419 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20421 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20424 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20425 parameter and_insn and return it.
20426 (recombine_lvx_pattern): Insert a copy to ensure availability of
20427 the base register of the copied masking operation at the point of
20428 the instruction replacement.
20429 (recombine_stvx_pattern): Likewise.
20431 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20433 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20434 undocumented switches.
20435 (-mpower9-dform-vector): Likewise.
20436 (-mpower9-dform): Likewise.
20437 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20438 comments to delete references to -mpower9-dform* switches.
20439 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20440 Delete reference to -mpower9-dform* switches, test for
20441 -mpower9-vector instead.
20442 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20443 (OTHER_P9_VECTOR_MASKS): Likewise.
20444 (POWERPC_MASKS): Likewise.
20445 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20446 tests against -mpower9-dform* to -mpower9-vector. Delete code
20447 that checked for -mpower9-dform* consistancy with other options.
20448 Add test for -mpower9-misc to enable other power9 switches.
20449 (rs6000_init_hard_regno_mode_ok): Likewise.
20450 (rs6000_option_override_internal): Likewise.
20451 (rs6000_emit_prologue): Likewise.
20452 (rs6000_emit_epilogue): Likewise.
20453 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20454 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20455 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20457 (emit_fusion_p9_store): Likewise.
20458 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20459 resetting these macros if the assembler does not support ISA 3.0
20461 (TARGET_P9_DFORM_VECTOR): Likewise.
20462 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20463 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20465 2017-08-25 Alan Modra <amodra@gmail.com>
20467 PR rtl-optimization/81747
20468 * cse.c (cse_extended_basic_block): Don't attempt to record
20469 equivalences for degenerate conditional branches that branch
20470 to their fall-through.
20472 2017-08-24 Martin Sebor <msebor@redhat.com>
20474 PR middle-end/81908
20475 * gimple-fold.c (size_must_be_zero_p): New function.
20476 (gimple_fold_builtin_memory_op): Call it.
20478 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20480 * config/rs6000/mm_malloc.h: New file.
20482 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20484 PR tree-optimization/81913
20485 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20486 analysis when either IVs in condition can wrap.
20488 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20490 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20491 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20493 2017-08-24 Richard Biener <rguenther@suse.de>
20496 * targhooks.c (default_target_can_inline_p): Properly
20497 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20498 is present and always compare.
20499 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20500 infer -mfpmath=sse from TARGET_SSE_P.
20501 (ix86_can_inline_p): Properly use target_option_default_node when
20502 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20504 2017-08-24 Richard Biener <rguenther@suse.de>
20507 * dwarf2out.c (output_die): Handle flag_generate_offload like
20509 (output_comp_unit): Likewise.
20510 (gen_array_type_die): Likewise.
20511 (dwarf2out_early_finish): Likewise.
20512 (note_variable_value_in_expr): Likewise.
20513 (dwarf2out_finish): Likewise. Adjust assert.
20514 * cgraphunit.c (symbol_table::compile): Move setting of
20515 flag_generate_offload earlier ...
20516 (symbol_table::finalize_compilation_unit): ... here, before
20517 early debug finalization.
20519 2017-08-24 Richard Biener <rguenther@suse.de>
20521 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20522 and ipa-fnsummary.h.
20523 (ix86_can_inline_p): When ix86_fpmath flags do not match
20524 check whether the callee uses FP math at all.
20526 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20528 PR middle-end/81931
20529 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20530 instead of TYPE_PRECISION.
20532 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20533 Alan Hayward <alan.hayward@arm.com>
20534 David Sherwood <david.sherwood@arm.com>
20536 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20538 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20540 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20542 (ix86_frame::stack_realign_allocate): New field.
20543 (struct machine_frame_state): Modify comments.
20544 (machine_frame_state::sp_realigned_fp_end): New field.
20545 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20546 layout calculation.
20547 (sp_valid_at): Add assertion to assure no attempt to access invalid
20548 offset of a realigned stack.
20549 (fp_valid_at): Likewise.
20550 (choose_baseaddr): Modify comments.
20551 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20552 ix86_expand_prologue.
20553 (ix86_expand_prologue): Modify stack realignment and allocation.
20554 (ix86_expand_epilogue): Modify comments.
20555 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20556 avx2_runtime, avx512f, and avx512f_runtime.
20558 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20560 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20561 (mstackrealign): Do not init to -1.
20562 * config/i386/i386.c (ix86_option_override_internal):
20563 Check opts_set, not opts when setting default value of
20564 opts->x_ix86_force_align_arg_pointer.
20566 2017-08-23 Richard Biener <rguenther@suse.de>
20568 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20569 lang_hooks.decl_printable_name.
20570 * print-rtl-function.c (print_rtx_function): Likewise.
20571 * tree-pretty-print.c (dump_function_header): Likewise.
20573 2017-08-23 Richard Biener <rguenther@suse.de>
20576 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20577 -g0 at compile-time.
20579 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20581 PR middle-end/19706
20582 * doc/sourcebuild.texi (Other hardware attributes):
20585 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20587 PR middle-end/19706
20588 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20589 Fix single-use check.
20591 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20593 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20595 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20597 * doc/install.texi: Modify to add more details on running selected
20600 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20602 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20603 is combined with -mabi=ms.
20604 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20607 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20609 PR tree-optimization/81488
20610 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20611 and cached_basis fields.
20612 (MAX_SPREAD): New constant.
20613 (alloc_cand_and_find_basis): Initialize new fields.
20614 (clear_visited): New function.
20615 (create_phi_basis_1): Rename from create_phi_basis, set visited
20616 and cached_basis fields.
20617 (create_phi_basis): New wrapper function.
20618 (phi_add_costs_1): Rename from phi_add_costs, add spread
20619 parameter, set visited field, short-circuit when limits reached.
20620 (phi_add_costs): New wrapper function.
20621 (record_phi_increments_1): Rename from record_phi_increments, set
20623 (record_phi_increments): New wrapper function.
20624 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20625 (phi_incr_cost): New wrapper function.
20626 (all_phi_incrs_profitable_1): Rename from
20627 all_phi_incrs_profitable, set visited field.
20628 (all_phi_incrs_profitable): New wrapper function.
20630 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20631 Alan Hayward <alan.hayward@arm.com>
20632 David Sherwood <david.sherwood@arm.com>
20634 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20635 that takes the outer and inner modes.
20636 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20637 comparison as the canonical test for a paradoxical subreg.
20638 * combine.c (simplify_set): Use paradoxical_subreg_p.
20639 (make_extraction): Likewise.
20640 (force_to_mode): Likewise.
20641 (rtx_equal_for_field_assignment_p): Likewise.
20642 (gen_lowpart_for_combine): Likewise.
20643 (simplify_comparison): Likewise.
20644 * cse.c (equiv_constant): Likewise.
20645 * expmed.c (store_bit_field_1): Likewise.
20646 * final.c (alter_subreg): Likewise.
20647 * fwprop.c (propagate_rtx): Likewise.
20648 (forward_propagate_subreg): Likewise.
20649 * ira-conflicts.c (ira_build_conflicts): Likewise.
20650 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20651 * lra-constraints.c (curr_insn_transform): Likewise.
20652 (split_reg): Likewise.
20653 * lra-eliminations.c (move_plus_up): Likewise.
20654 (lra_eliminate_regs_1): Likewise.
20655 * recog.c (general_operand): Likewise.
20656 * ree.c (combine_reaching_defs): Likewise.
20657 * reload.c (push_reload): Likewise.
20658 (find_reloads): Likewise.
20659 * reload1.c (elimination_effects): Likewise.
20660 (compute_reload_subreg_offset): Likewise.
20661 (choose_reload_regs): Likewise.
20662 * rtlanal.c (subreg_lsb_1): Likewise.
20663 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20664 (simplify_subreg): Likewise.
20665 * var-tracking.c (track_loc_p): Likewise.
20666 * emit-rtl.c (byte_lowpart_offset): Likewise.
20667 (paradoxical_subreg_p): Delete out-of-line definition.
20669 2017-08-22 Jeff Law <law@redhat.com>
20671 PR tree-optimization/81741
20672 PR tree-optimization/71947
20673 * tree-ssa-dom.c: Include tree-inline.h.
20674 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20675 equivalences if one is more expensive to compute than the other.
20676 * tree-ssa-scopedtables.h (class const_or_copies): Make
20677 record_const_or_copy_raw method private.
20678 (class avail_exprs_stack): New method simplify_binary_operation.
20679 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20680 avail_exprs_stack::simplify_binary_operation as needed.
20681 (avail_exprs_stack::simplify_binary_operation): New function.
20683 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20685 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20686 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20687 (DOT_SYMBOLS): Likewise.
20688 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20689 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20690 (RS6000_ABI_NAME): Likewise.
20691 (TARGET_CMODEL): Likewise.
20692 (TOC_SECTION_ASM_OP): Likewise.
20693 (SET_CMODEL): New macro.
20694 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20696 2017-08-22 Richard Biener <rguenther@suse.de>
20698 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20699 to allow for free-lang-data replacements similar to verify_type_variant.
20701 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20702 Alan Hayward <alan.hayward@arm.com>
20703 David Sherwood <david.sherwood@arm.com>
20705 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20706 VOIDmode for the LABEL_REF.
20708 2017-08-22 Richard Biener <rguenther@suse.de>
20710 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20712 2017-08-22 Richard Biener <rguenther@suse.de>
20714 * common.opt (feliminate-dwarf2-dups): Ignore.
20715 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20716 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20717 same_die_p_wrap, compute_section_prefix,
20718 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20719 (comdat_symbol_id, comdat_symbol_number): Likewise.
20720 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20722 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20723 (output_die): Mark unreachable path unreachable.
20724 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20725 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20726 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20727 (dwarf2out_early_finish): Likewise.
20729 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20731 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20733 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20736 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20737 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20738 (avr_attribute_table) <io, io_low, address>: Initialize
20739 .decl_required with true.
20741 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20743 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20744 undocumented debugging options.
20745 (-mvsx-scalar-double): Likewise.
20746 (-mallow-df-permute): Likewise.
20747 (-mvectorize-builtins): Likewise.
20748 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20749 (rs6000_builtin_vectorized_function): Likewise.
20750 (rs6000_builtin_md_vectorized_function): Likewise.
20751 (rs6000_opt_vars): Likewise.
20753 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20756 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20757 (*btrq_imm): Rename from *btrq.
20758 (*btcq_imm): Rename from *btcq.
20759 (btsc): New code attribute.
20760 (*<btsc><mode>): New insn pattern.
20761 (*btr<mode>): Ditto.
20762 (*<btsc><mode>_mask): New insn_and_split pattern.
20763 (*btr<mode>_mask): Ditto.
20765 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20766 Alan Hayward <alan.hayward@arm.com>
20767 David Sherwood <david.sherwood@arm.com>
20769 * function.c (pad_below): Simplify padding calculation.
20771 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20772 Alan Hayward <alan.hayward@arm.com>
20773 David Sherwood <david.sherwood@arm.com>
20775 * target.def (function_prologue): Remove frame size argument.
20776 (function_epilogue): Likewise.
20777 * doc/tm.texi: Regenerate.
20778 * final.c (final_start_function): Update call to function_prologue.
20779 (final_end_function): Update call to function_epilogue.
20780 (default_function_pro_epilogue): Remove frame size argument.
20781 * output.h (default_function_pro_epilogue): Likewise.
20782 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20783 (arm_output_function_prologue): Likewise.
20784 * config/frv/frv.c (frv_function_prologue): Likewise.
20785 (frv_function_epilogue): Likewise.
20786 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20787 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20788 (ia64_output_function_epilogue): Likewise.
20789 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20790 (m32r_output_function_epilogue): Likewise.
20791 * config/microblaze/microblaze.c (microblaze_function_prologue)
20792 (microblaze_function_epilogue): Likewise.
20793 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20794 (mips_output_function_epilogue): Likewise.
20795 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20796 (mmix_target_asm_function_epilogue): Likewise.
20797 * config/msp430/msp430.c (msp430_start_function): Likewise.
20798 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20799 (nds32_asm_function_epilogue): Likewise.
20800 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20801 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20802 (pa_output_function_epilogue): Likewise.
20803 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20804 (rs6000_output_function_epilogue): Likewise.
20805 * config/rl78/rl78.c (rl78_start_function): Likewise.
20806 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20807 (rs6000_output_function_epilogue): Likewise.
20808 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20809 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20810 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20811 (sparc_asm_function_epilogue): Likewise.
20813 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20815 * tree.h (type_has_mode_precision_p): New function.
20816 * convert.c (convert_to_integer_1): Use it.
20817 * expr.c (expand_expr_real_2): Likewise.
20818 (expand_expr_real_1): Likewise.
20819 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20820 * match.pd: Likewise.
20821 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20822 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20823 * tree-tailcall.c (process_assignment): Likewise.
20824 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20825 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20826 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20827 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20828 (vectorizable_assignment): Likewise.
20829 (vectorizable_shift): Likewise.
20830 (vectorizable_operation): Likewise.
20831 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20833 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20835 * match.pd: Add pow (C, x) simplification.
20837 2017-08-21 Richard Biener <rguenther@suse.de>
20839 PR tree-optimization/81900
20840 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20841 for blocks with abnormal predecessors.
20842 (compute_antic): Do not set visited flag prematurely.
20844 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20847 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20849 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20851 * stor-layout.h (vector_type_mode): Move to...
20852 * tree.h (vector_type_mode): ...here.
20853 * stor-layout.c (vector_type_mode): Move to...
20854 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20856 2017-08-21 Richard Biener <rguenther@suse.de>
20858 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20859 register_external_die hooks.
20860 (debug_false_tree_charstarstar_uhwistar): Declare.
20861 (debug_nothing_tree_charstar_uhwi): Likewise.
20862 * debug.c (do_nothing_debug_hooks): Adjust.
20863 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20864 (debug_nothing_tree_charstar_uhwi): Likewise.
20865 * dbxout.c (dbx_debug_hooks): Adjust.
20866 (xcoff_debug_hooks): Likewise.
20867 * sdbout.c (sdb_debug_hooks): Likewise.
20868 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20869 * dwarf2out.c (macinfo_label_base): New global.
20870 (dwarf2out_register_external_die): New function for the
20871 register_external_die hook.
20872 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20873 (dwarf2_debug_hooks): Use them.
20874 (dwarf2_lineno_debug_hooks): Adjust.
20875 (struct die_struct): Add with_offset flag.
20876 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20877 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20878 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20879 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20880 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20881 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20882 defining section names for the early LTO debug variants.
20883 (reset_indirect_string): New helper.
20884 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20885 (print_dw_val): Add support for offsetted symbol references.
20886 (get_ultimate_context): Split out from is_cxx.
20887 (is_cxx): Use get_ultimate_context.
20888 (is_fortran): Add decl overload.
20889 (compute_comp_unit_symbol): Split out worker from
20890 compute_section_prefix.
20891 (compute_section_prefix): Call compute_comp_unit_symbol and
20892 set comdat_type_p here.
20893 (output_die): Skip DIE symbol output for the LTO added one.
20894 Handle DIE symbol references with offset.
20895 (output_comp_unit): Guard section name mangling properly.
20896 For LTO debug sections emit a symbol at the section beginning
20897 which we use to refer to its DIEs.
20898 (add_abstract_origin_attribute): For DIEs registered via
20899 dwarf2out_register_external_die directly refer to the early
20900 DIE rather than indirectly through the shadow one we created.
20901 Remove obsolete call to dwarf2out_abstract_function for
20902 non-function/block origins.
20903 (gen_array_type_die): When generating early LTO debug do
20904 not emit DW_AT_string_length.
20905 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20906 late when in LTO. As suggested place a gcc_unreachable for
20907 the DECL_ABSTRACT_P case.
20908 (gen_subprogram_die): Avoid another specification DIE
20909 for early built declarations/definitions for the late LTO case.
20910 (gen_variable_die): Add type references for late duplicated VLA dies
20912 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20913 we have the abstract instance already.
20914 (process_scope_var): Adjust decl DIE contexts in LTO which
20915 first puts them in limbo.
20916 (gen_decl_die): Do not generate type DIEs late apart from
20917 types for VLAs or for decls we do not yet have a DIE. Do not
20918 call dwarf2out_abstract_function late.
20919 (dwarf2out_early_global_decl): Make sure to create DIEs
20920 for abstract instances of a decl first.
20921 (dwarf2out_late_global_decl): Adjust comment.
20922 (output_macinfo_op): With multiple macro sections use
20923 macinfo_label_base to distinguish labels.
20924 (output_macinfo): Likewise. Update macinfo_label_base.
20925 Pass in the line info label.
20926 (note_variable_value_in_expr): When generating LTO resolve
20927 all variable values here by generating DIEs as needed.
20928 (init_sections_and_labels): Add early LTO debug flag parameter
20929 and generate different sections and names if set. Add generation
20930 counter for the labels so we can have multiple of them.
20931 (reset_dies): Helper to allow DIEs to be output multiple times.
20932 (dwarf2out_finish): When outputting DIEs to the fat part of an
20933 LTO object first reset DIEs.
20934 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20935 (modified_type_die): Check for decl_ultimate_origin being self
20937 (gen_type_die_with_usage): Likewise.
20938 (gen_typedef_die): Allow decl_ultimate_origin being self.
20939 (set_decl_abstract_flags): Remove.
20940 (set_block_abstract_flags): Likewise.
20941 (dwarf2out_abstract_function): Treat the early generated DIEs
20942 as the abstract copy and only add DW_AT_inline and
20943 DW_AT_artificial here and call set_decl_origin_self.
20944 If the DIE has an abstract origin don't do anything.
20945 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20946 if we have none yet (Go fails to build one, PR78628).
20947 (variably_modified_type_p): Prevent endless recursion for Ada
20948 cyclic pointer types.
20949 * lto-streamer-in.c: Include debug.h.
20950 (dref_queue): New global.
20951 (lto_read_tree_1): Stream in DIE references.
20952 (lto_input_tree): Register DIE references.
20953 (input_function): Stream DECL_DEBUG_ARGS.
20954 * lto-streamer-out.c: Include debug.h.
20955 (lto_write_tree_1): Output DIE references.
20956 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20957 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20958 (output_function): Stream DECL_DEBUG_ARGS.
20959 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20960 Stream DECL_ABSTRACT_ORIGIN.
20961 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20962 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20963 DECL_CONTEXT for file-scope decls.
20964 * lto-streamer.h (struct dref_entry): Declare.
20965 (dref_queue): Likewise.
20966 * cfgexpand.c (pass_expand::execute): Do not call the
20967 outlining_inline_function hook here.
20968 * lto-wrapper.c (debug_obj): New global.
20969 (tool_cleanup): Unlink it if required.
20970 (debug_objcopy): New function.
20971 (run_gcc): Handle early debug sections in the IL files by
20972 extracting them to separate files, partially linkin them and
20973 feeding the result back as result to the linker.
20974 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20975 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20976 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20977 sections into a separate segment.
20978 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20980 (darwin_asm_dwarf_section): Likewise.
20981 (darwin_asm_output_dwarf_offset): Likewise.
20982 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20984 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20985 Alan Hayward <alan.hayward@arm.com>
20986 David Sherwood <david.sherwood@arm.com>
20988 * read-md.h (md_reader::record_potential_iterator_use): Replace
20989 pointer argument with an rtx and an index.
20990 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20991 (apply_mode_iterator): Likewise.
20992 (apply_code_iterator): Likewise.
20993 (apply_int_iterator): Likewise.
20994 (apply_subst_iterator): Likewise.
20995 (record_iterator_use): Likewise.
20996 (record_attribute_use): Likewise.
20997 (md_reader::record_potential_iterator_use): Likewise. Update calls
20998 to record_iterator_use and apply_iterator.
20999 (iterator_use): Replace ptr with x and index.
21000 (attribute_use): Likewise.
21001 (apply_attribute_uses): Update calls to apply_iterator.
21002 (apply_iterators): Likewise. Update initialization of iterator_use.
21003 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21004 and record_potential_iterator_use.
21005 (rtx_reader::read_rtx_operand): Likewise.
21007 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21008 Alan Hayward <alan.hayward@arm.com>
21009 David Sherwood <david.sherwood@arm.com>
21011 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21014 2017-08-21 Richard Biener <rguenther@suse.de>
21016 PR middle-end/81884
21017 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21018 at struct end conservatively when comparing common bases.
21020 2017-08-21 Richard Biener <rguenther@suse.de>
21022 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21023 (mem_ref_in_stmt): Remove.
21024 (determine_max_movement): Use ref index to get at the reference.
21025 (invariantness_dom_walker::before_dom_children): Deal with
21026 lim data already initialized.
21027 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21029 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21031 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21032 (TARGET_ISA_ROUND): Ditto.
21033 (TARGET_ROUND): Ditto.
21034 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21035 * config/i386/i386.md: Ditto.
21036 * config/i386/sse.md: Ditto.
21037 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21038 with OPTION_MASK_ISA_SSE4_1.
21040 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21043 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21044 __builtin_ia32_lzcnt_u16.
21046 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21049 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21050 (rs6000_set_current_function): Rewrite function to use it.
21052 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21055 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21056 and TYPE_WARN_IF_NOT_ALIGN.
21057 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21058 (handle_warn_if_not_align): New.
21059 (place_union_field): Call handle_warn_if_not_align.
21060 (place_field): Call handle_warn_if_not_align.
21061 Copy TYPE_WARN_IF_NOT_ALIGN.
21062 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21063 (layout_type): Likewise.
21064 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21066 (tree_decl_common): Add warn_if_not_align.
21067 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21068 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21069 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21070 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21071 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21072 * doc/extend.texi: Document warn_if_not_aligned attribute.
21073 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21075 2017-08-17 Martin Liska <mliska@suse.cz>
21078 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21079 (get_data_dependence): Use it as pointer type.
21080 (distribute_loop): Likewise.
21082 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21084 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21085 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21086 (p8_vmrgow_<mode>_direct): New define_insn.
21087 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21088 handle endianness for vmrgew and vmrgow permute patterns.
21090 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21092 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21093 * config/rs6000/rs6000-cpus.def: Remove comment.
21094 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21095 (POWERPC_MASKS): Likewise.
21096 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21097 use of TARGET_VSX_TIMODE.
21098 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21099 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21100 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21101 (rs6000_option_override_internal): Remove dead code.
21102 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21103 (rs6000_legitimize_reload_address): Likewise.
21104 (rs6000_legitimate_address_p): Likewise.
21105 (rs6000_opt_masks): Delete "vsx-timode".
21106 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21107 from function comment.
21108 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21109 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21110 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21112 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21113 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21114 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21115 (VSX_TI): Likewise.
21117 (define_peephole2): Likewise.
21119 2017-08-17 Martin Sebor <msebor@redhat.com>
21122 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21123 past the end of an array.
21124 (test_pp_format): Add test cases.
21126 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21128 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21129 missing ECF_NOTHROW flags.
21131 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21134 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21135 operands residing in integer registers.
21136 (*vsx_le_perm_load_<mode>): Likewise.
21137 (*vsx_le_perm_store_<mode>): Likewise.
21138 (define_peephole2): Add peepholes to optimize the above.
21140 2017-08-17 Marek Polacek <polacek@redhat.com>
21142 PR middle-end/81814
21143 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21144 to mimic what shorten_compare did. Change the return type to bool.
21145 (fold_cond_expr_with_comparison): Update call to
21146 operand_equal_for_comparison_p.
21147 (fold_ternary_loc): Likewise.
21149 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21151 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21153 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21155 2017-08-17 Richard Biener <rguenther@suse.de>
21157 * tree-ssa-structalias.c (solve_graph): When propagating
21158 to successors update the graphs succ edges and avoid duplicate work.
21160 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21163 * config/i386/i386.c (ix86_option_override_internal): Save target
21164 specific options after ix86_stack_protector_guard_reg was changed.
21166 2017-08-17 Richard Biener <rguenther@suse.de>
21168 PR tree-optimization/81827
21169 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21170 (new_var_info): Initialize it conservatively.
21171 (get_call_vi): Mark register vars.
21172 (new_scalar_tmp_constraint_exp): Likewise.
21173 (handle_rhs_call): Likewise.
21174 (handle_const_call): Likewise.
21175 (create_function_info_for): Likewise.
21176 (solve_constraints): Sort varinfos to separate register from
21177 non-register vars to pack points-to solution bitmaps during
21180 2017-08-17 Marek Polacek <polacek@redhat.com>
21182 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21184 2017-08-17 Richard Biener <rguenther@suse.de>
21186 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21187 to true when overflow is undefined and we saturated the result.
21189 2017-08-17 Alan Modra <amodra@gmail.com>
21192 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21193 Don't use store multiple if only one reg needs saving.
21194 (interesting_frame_related_regno): New function.
21195 (rs6000_frame_related): Don't emit frame info for regs that
21197 (rs6000_emit_epilogue): Likewise.
21199 2017-08-16 Nathan Sidwell <nathan@acm.org>
21201 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21202 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21203 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21204 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21205 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21206 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21207 (verify_type): Adjust for TYPE_BINFO move.
21208 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21209 process TYPE_BINFO directly.
21210 (hash_tree): Likewise.
21211 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21213 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21216 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21218 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21220 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21223 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21224 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21225 Add ix86_binary_operator_ok to insn constraint.
21226 (*iordi_1_bts): Ditto.
21227 (*xordi_1_btc): Ditto.
21228 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21229 Update corresponding peephole2 pattern.
21233 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21235 PR tree-optimization/81832
21236 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21237 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21239 2017-08-16 Marek Polacek <polacek@redhat.com>
21242 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21243 regarding size_zero_node.
21245 2017-08-16 Martin Liska <mliska@suse.cz>
21248 * config.gcc: Respect previously set extra_objs in case
21251 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21253 PR tree-optimization/81835
21254 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21255 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21256 not depend on the phi.
21258 2017-08-16 Alan Modra <amodra@gmail.com>
21260 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21263 2017-08-16 Alan Modra <amodra@gmail.com>
21265 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21266 (save_reg_p): ..into this. Update all callers.
21267 (first_reg_to_save): Simplify.
21269 2017-08-16 Alan Modra <amodra@gmail.com>
21271 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21274 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21278 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21279 constant count if that count is less than 32.
21281 2017-08-15 Nathan Sidwell <nathan@acm.org>
21283 * gcc.c (execute): Emit friendlier message if inferior is killed
21284 by an external cause.
21286 2017-08-15 Richard Biener <rguenther@suse.de>
21288 PR tree-optimization/81790
21289 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21290 CONSTRUCTORs from simplifying and VN.
21292 2017-08-14 Martin Sebor <msebor@redhat.com>
21294 * builtin-attrs.def: Add comments.
21296 2017-08-14 Martin Sebor <msebor@redhat.com>
21299 * doc/extend.texi (attribute nonstring): Document new attribute.
21301 2017-08-14 Martin Sebor <msebor@redhat.com>
21304 * tree-diagnostic.c (default_tree_printer): Handle %G.
21305 * gimple-pretty-print.h (percent_G_format): Declare new function.
21306 * gimple-pretty-print.c (percent_G_format): Define.
21307 * tree-pretty-print.c (percent_K_format): Add argument.
21309 2017-08-14 Martin Sebor <msebor@redhat.com>
21311 PR translation/79998
21312 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21313 Remove a stray space.
21315 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21318 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21319 (*iordi_1_bts): Ditto.
21320 (*xordi_1_btc): Ditto.
21322 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21325 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21326 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21328 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21329 quoted strings, and make more translator-friendly.
21330 (darwin_rs6000_override_options): Likewise.
21331 (rs6000_option_override_internal): Likewise.
21332 (rs6000_return_in_memory): Fix overlong line.
21333 (init_cmulative_args): Use quoted strings, and make more
21334 translator-friendly.
21335 (rs6000_pass_by_reference): Fix overlong line.
21336 (def_builtin): Use quoted strings.
21337 (altivec_expand_predicate_builtin): Use quoted strings, and make
21338 more translator-friendly.
21339 (htm_expand_builtin): Use quoted strings.
21340 (cpu_expand_builtin): Use quoted strings, and make more
21341 translator-friendly.
21342 (altivec_expand_builtin): Likewise.
21343 (paired_expand_predicate_builtin): Likewise.
21344 (rs6000_invalid_builtin): Likewise.
21345 (builtin_function_type): Use quoted strings.
21346 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21347 more translator-friendly.
21348 (rs6000_trampoline_init): Likewise.
21349 (rs6000_handle_altivec_attribute): Likewise.
21350 (rs6000_inner_target_options): Use quoted strings.
21351 (rs6000_disable_incompatible_switches): Likewise.
21352 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21353 strings, and make more translator-friendly.
21354 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21356 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21358 PR tree-optimization/81799
21359 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21360 cond_expr to simple gimple operand.
21362 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21364 PR middle-end/46932
21365 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21367 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21371 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21372 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21373 TARGET_GASISR_PROLOGUES.
21374 * config/avr/avr.c (avr_option_override): Same.
21375 (avr_pass_pre_proep::execute): Same.
21377 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21380 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21381 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21383 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21385 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21386 define_insn_and_split. Split to a memory load from 0 in
21387 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21388 using PTR mode iterator.
21389 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21390 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21391 (*add_tp_<mode>): Redefine as define_insn_and_split.
21392 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21393 address space. Merge with *add_tp_x32 using PTR mode iterator.
21394 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21395 Split to an add with a memory load from 0 in
21396 DEFAULT_TLS_SEG_REG address space.
21398 2017-08-12 Andrew Pinski <apinski@cavium.com>
21400 * config/aarch64/aarch64-option-extensions.def (rdma):
21401 Fix feature string to what Linux prints out in /proc/cpuinfo.
21403 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21406 * dwarf2out.c (modified_type_die): For C typedef types that have
21407 an ultimate origin, process the ultimate origin instead of the
21409 (gen_typedef_die): Assert that input DECLs have no ultimate
21411 (gen_type_die_with_usage): For typedef variants that have an
21412 ultimate origin, just call gen_decl_die on the original DECL.
21413 (process_scope_var): Avoid creating DIEs for local typedefs and
21414 concrete static variables.
21416 2017-08-12 Alan Modra <amodra@gmail.com>
21420 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21421 match gnu-user.h startfile.
21422 (ENDFILE_LINUX_SPEC): Similarly.
21424 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21427 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21429 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21431 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21432 * config/aarch64/aarch64.md (mov<mode>): Change.
21433 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21434 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21435 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21437 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21439 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21440 for storage order barriers.
21442 2017-08-11 Martin Liska <mliska@suse.cz>
21445 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21446 variables of void type.
21448 2017-08-11 Martin Liska <mliska@suse.cz>
21450 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21451 TARGET_SUPPORTS_ALIASES.
21452 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21453 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21454 (optimize_weakref): Likewise.
21455 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21456 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21457 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21459 2017-08-11 Martin Liska <mliska@suse.cz>
21462 * config/i386/i386.c (make_resolver_func): Do complete
21463 refactoring of the function.
21465 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21468 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21469 * config/i386/i386.c (ix86_stack_protect_guard): Use
21470 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21471 * doc/invoke.texi (x86 Options): Document
21472 -mstack-protector-guard-symbol= option.
21474 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21476 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21477 * config/i386/i386.c (ix86_split_stack_guard): New function.
21478 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21479 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21480 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21481 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21482 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21483 (split_stack_space_check): Call ix86_split_stack_guard.
21485 2017-08-10 Martin Sebor <msebor@redhat.com>
21487 * print-tree.c (print_node): Print location using the established
21489 Replace spaces with colons.
21490 (debug_raw, debug): Ditto.
21492 2017-08-10 Martin Sebor <msebor@redhat.com>
21495 * pretty-print.c (pp_format): Correct the handling of %s precision.
21497 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21500 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21502 (ix86_finalize_stack_frame_flags): This. Also clear
21503 frame_pointer_needed if -fno-omit-frame-pointer is used without
21505 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21506 with ix86_finalize_stack_frame_flags.
21507 (ix86_expand_epilogue): Likewise.
21508 (ix86_expand_split_stack_prologue): Likewise.
21509 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21511 2017-08-10 Martin Liska <mliska@suse.cz>
21514 * c-attribs.c (handle_target_attribute):
21515 Report warning for an empty string argument of target attribute.
21517 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21520 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21522 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21523 or DECL_NONLOCAL labels.
21524 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21525 or DECL_NONLOCAL labels here.
21527 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21529 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21530 to indicate when early gimple folding has been disabled.
21531 (rs6000_gimple_fold_builtin): Add debug content.
21532 (rs6000_invalid_builtin): Fix whitespace.
21533 (rs6000_expand_builtin): Fix whitespace.
21534 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21536 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21539 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21540 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21542 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21544 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21545 (qdf24xx): Likewise.
21546 * config/aarch64/aarch64.md: Include falkor.md.
21547 * config/aarch64/falkor.md: New.
21549 2017-08-09 Marek Polacek <polacek@redhat.com>
21552 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21553 * diagnostic.c (emit_diagnostic): Add a comment.
21554 (emit_diagnostic_valist): New function.
21556 2017-08-09 Marek Polacek <polacek@redhat.com>
21559 * input.c (make_location): New overload.
21560 * input.h (make_location): Declare.
21562 2017-08-08 Alan Modra <amodra@gmail.com>
21563 H.J. Lu <hongjiu.lu@intel.com>
21566 * gcc.c (NO_PIE_SPEC): Delete.
21567 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21569 (LINK_PIE_SPEC): ..to here.
21570 (LINK_COMMAND_SPEC): Support -no-pie.
21571 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21572 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21574 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21575 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21576 (ENDFILE_CRTEND_SPEC): Similarly.
21578 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21581 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21582 (mstack-protector-guard-offset=): Ditto.
21583 * config/i386/i386.c (ix86_option_override): Handle
21584 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21586 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21587 ix86_stack_protect_guard_offset variables.
21588 (TARGET_STACK_PROTECT_GUARD): Always define.
21589 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21590 and -mstack-protector-guard-offset= options.
21592 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21594 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21595 boundary case for the last candidate.
21597 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21599 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21600 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21602 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21604 PR middle-end/19706
21605 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21606 * config/aarch64/aarch64-builtins.c
21607 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21608 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21609 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21611 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21612 Andrew Pinski <pinskia@gmail.com>
21614 PR middle-end/19706
21615 * internal-fn.def (XORSIGN): New.
21616 * optabs.def (xorsign_optab): New.
21617 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21618 (convert_expand_mult_copysign): New.
21619 (pass_optimize_widening_mul::execute): Call
21620 convert_expand_mult_copysign.
21622 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21624 PR tree-optimization/81354
21625 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21626 Insert on edges rather than explicitly creating landing pads.
21627 (analyze_candidates_and_replace): Commit edge inserts.
21629 2017-08-08 Richard Biener <rguenther@suse.de>
21631 PR middle-end/81719
21632 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21633 (expand_simple_operations): Also look through ADDR_EXPRs with
21634 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21636 2017-08-08 Richard Biener <rguenther@suse.de>
21638 PR tree-optimization/81723
21639 * tree-vect-slp.c (struct bst_traits): New hash traits.
21640 (bst_fail): New global.
21641 (vect_build_slp_tree_2): New worker, split out from ...
21642 (vect_build_slp_tree): ... this now wrapping it with using
21643 bst_fail set to cache SLP tree build fails. Properly handle
21645 (vect_analyze_slp_instance): Allocate and free bst_fail.
21647 2017-08-08 Martin Liska <mliska@suse.cz>
21650 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21651 LABEL_DECLs that can be from a different function.
21653 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21655 PR tree-optimization/81744
21656 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21657 loop's number of iterations.
21659 2017-08-08 Martin Liska <mliska@suse.cz>
21661 * asan.c: Include header files.
21662 * attribs.c (build_decl_attribute_variant): New function moved
21664 (build_type_attribute_qual_variant): Likewise.
21665 (cmp_attrib_identifiers): Likewise.
21666 (simple_cst_list_equal): Likewise.
21667 (omp_declare_simd_clauses_equal): Likewise.
21668 (attribute_value_equal): Likewise.
21669 (comp_type_attributes): Likewise.
21670 (build_type_attribute_variant): Likewise.
21671 (lookup_ident_attribute): Likewise.
21672 (remove_attribute): Likewise.
21673 (merge_attributes): Likewise.
21674 (merge_type_attributes): Likewise.
21675 (merge_decl_attributes): Likewise.
21676 (merge_dllimport_decl_attributes): Likewise.
21677 (handle_dll_attribute): Likewise.
21678 (attribute_list_equal): Likewise.
21679 (attribute_list_contained): Likewise.
21680 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21681 (lookup_attribute_by_prefix): Likewise.
21682 * bb-reorder.c: Include header files.
21683 * builtins.c: Likewise.
21684 * calls.c: Likewise.
21685 * cfgexpand.c: Likewise.
21686 * cgraph.c: Likewise.
21687 * cgraphunit.c: Likewise.
21688 * convert.c: Likewise.
21689 * dwarf2out.c: Likewise.
21690 * final.c: Likewise.
21691 * fold-const.c: Likewise.
21692 * function.c: Likewise.
21693 * gimple-expr.c: Likewise.
21694 * gimple-fold.c: Likewise.
21695 * gimple-pretty-print.c: Likewise.
21696 * gimple.c: Likewise.
21697 * gimplify.c: Likewise.
21698 * hsa-common.c: Likewise.
21699 * hsa-gen.c: Likewise.
21700 * internal-fn.c: Likewise.
21701 * ipa-chkp.c: Likewise.
21702 * ipa-cp.c: Likewise.
21703 * ipa-devirt.c: Likewise.
21704 * ipa-fnsummary.c: Likewise.
21705 * ipa-inline.c: Likewise.
21706 * ipa-visibility.c: Likewise.
21708 * lto-cgraph.c: Likewise.
21709 * omp-expand.c: Likewise.
21710 * omp-general.c: Likewise.
21711 * omp-low.c: Likewise.
21712 * omp-offload.c: Likewise.
21713 * omp-simd-clone.c: Likewise.
21714 * opts-global.c: Likewise.
21715 * passes.c: Likewise.
21716 * predict.c: Likewise.
21717 * sancov.c: Likewise.
21718 * sanopt.c: Likewise.
21719 * symtab.c: Likewise.
21720 * toplev.c: Likewise.
21721 * trans-mem.c: Likewise.
21722 * tree-chkp.c: Likewise.
21723 * tree-eh.c: Likewise.
21724 * tree-into-ssa.c: Likewise.
21725 * tree-object-size.c: Likewise.
21726 * tree-parloops.c: Likewise.
21727 * tree-profile.c: Likewise.
21728 * tree-ssa-ccp.c: Likewise.
21729 * tree-ssa-live.c: Likewise.
21730 * tree-ssa-loop.c: Likewise.
21731 * tree-ssa-sccvn.c: Likewise.
21732 * tree-ssa-structalias.c: Likewise.
21733 * tree-ssa.c: Likewise.
21734 * tree-streamer-in.c: Likewise.
21735 * tree-vectorizer.c: Likewise.
21736 * tree-vrp.c: Likewise.
21737 * tsan.c: Likewise.
21738 * ubsan.c: Likewise.
21739 * varasm.c: Likewise.
21740 * varpool.c: Likewise.
21741 * tree.c: Remove functions moved to attribs.[ch].
21742 * tree.h: Likewise.
21743 * config/aarch64/aarch64.c: Add attrs.h header file.
21744 * config/alpha/alpha.c: Likewise.
21745 * config/arc/arc.c: Likewise.
21746 * config/arm/arm.c: Likewise.
21747 * config/avr/avr.c: Likewise.
21748 * config/bfin/bfin.c: Likewise.
21749 * config/c6x/c6x.c: Likewise.
21750 * config/cr16/cr16.c: Likewise.
21751 * config/cris/cris.c: Likewise.
21752 * config/darwin.c: Likewise.
21753 * config/epiphany/epiphany.c: Likewise.
21754 * config/fr30/fr30.c: Likewise.
21755 * config/frv/frv.c: Likewise.
21756 * config/ft32/ft32.c: Likewise.
21757 * config/h8300/h8300.c: Likewise.
21758 * config/i386/winnt.c: Likewise.
21759 * config/ia64/ia64.c: Likewise.
21760 * config/iq2000/iq2000.c: Likewise.
21761 * config/lm32/lm32.c: Likewise.
21762 * config/m32c/m32c.c: Likewise.
21763 * config/m32r/m32r.c: Likewise.
21764 * config/m68k/m68k.c: Likewise.
21765 * config/mcore/mcore.c: Likewise.
21766 * config/microblaze/microblaze.c: Likewise.
21767 * config/mips/mips.c: Likewise.
21768 * config/mmix/mmix.c: Likewise.
21769 * config/mn10300/mn10300.c: Likewise.
21770 * config/moxie/moxie.c: Likewise.
21771 * config/msp430/msp430.c: Likewise.
21772 * config/nds32/nds32-isr.c: Likewise.
21773 * config/nds32/nds32.c: Likewise.
21774 * config/nios2/nios2.c: Likewise.
21775 * config/nvptx/nvptx.c: Likewise.
21776 * config/pa/pa.c: Likewise.
21777 * config/pdp11/pdp11.c: Likewise.
21778 * config/powerpcspe/powerpcspe.c: Likewise.
21779 * config/riscv/riscv.c: Likewise.
21780 * config/rl78/rl78.c: Likewise.
21781 * config/rx/rx.c: Likewise.
21782 * config/s390/s390.c: Likewise.
21783 * config/sh/sh.c: Likewise.
21784 * config/sol2.c: Likewise.
21785 * config/sparc/sparc.c: Likewise.
21786 * config/spu/spu.c: Likewise.
21787 * config/stormy16/stormy16.c: Likewise.
21788 * config/tilegx/tilegx.c: Likewise.
21789 * config/tilepro/tilepro.c: Likewise.
21790 * config/v850/v850.c: Likewise.
21791 * config/vax/vax.c: Likewise.
21792 * config/visium/visium.c: Likewise.
21793 * config/xtensa/xtensa.c: Likewise.
21795 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21798 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21799 constraints since the -mupper-regs-* switches have been
21801 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21802 into a vector from a double word element that was extracted from
21803 another vector, and eliminate extra XXPERMDI instructions.
21804 (vsx_concat_<mode>_2): Likewise.
21805 (vsx_concat_<mode>_3): Likewise.
21806 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21807 concat to allow optimizing inserts from previous extracts.
21809 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21811 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21812 memory reference to a SSP offset in TLS address space.
21813 (ix86_print_operand) <case '@'>: Remove.
21814 (ix86_print_operand_punct_valid_p): Remove '@' code.
21815 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21816 UNSPEC_SP_TLS_TEST.
21817 (stack_tls_protect_set_<mode>): Remove.
21818 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21819 (stack_tls_protect_test_<mode>): Remove.
21820 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21822 2017-08-07 Olivier Hainque <hainque@adacore.com>
21825 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21827 2017-08-07 Douglas Rupp <rupp@adacore.com>
21829 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21830 variable was referenced as multidir in command.
21832 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21835 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21838 2017-08-07 Martin Liska <mliska@suse.cz>
21840 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21841 * config/rl78/rl78.c: Add include of attribs.h.
21842 * config/sh/sh.c: Likewise.
21843 * config/v850/v850.c: Likewise.
21845 2017-08-07 Tom de Vries <tom@codesourcery.com>
21847 PR middle-end/78266
21848 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21850 2017-08-07 Martin Liska <mliska@suse.cz>
21852 * config/mips/mips.c: Include attribs.h.
21854 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21857 * doc/invoke.texi: Document change in behvaior for -Ofast for
21860 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21862 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21864 (aarch64_pop_regs): Likewise.
21865 (aarch64_gen_load_pair): Likewise.
21866 (aarch64_save_callee_saves): Likewise.
21867 (aarch64_restore_callee_saves): Likewise.
21869 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21871 * config/i386/i386.c: Revert the last change.
21873 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21876 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21878 (ix86_finalize_stack_frame_flags): This. Also clear
21879 frame_pointer_needed if -fno-omit-frame-pointer is used without
21881 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21882 with ix86_finalize_stack_frame_flags.
21883 (ix86_expand_epilogue): Likewise.
21884 (ix86_expand_split_stack_prologue): Likewise.
21886 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21889 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21890 to P_AES for Westmere.
21892 2017-08-07 Jonathan Yong <10walls@gmail.com>
21894 * config/i386/mingw.opt (fset-stack-executable): Removed.
21895 * config/i386/cygming.opt (fset-stack-executable): Moved
21897 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21899 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21901 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21903 2017-08-07 Marek Polacek <polacek@redhat.com>
21905 PR middle-end/81737
21906 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21908 2017-08-07 Martin Liska <mliska@suse.cz>
21910 * attribs.h (canonicalize_attr_name): New function.
21911 (cmp_attribs): Move from c-format.c and adjusted.
21912 (is_attribute_p): Moved from tree.h.
21913 * tree-inline.c: Add new includes.
21914 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21915 (private_is_attribute_p): Remove.
21916 (private_lookup_attribute): Likewise.
21917 (private_lookup_attribute_by_prefix): Simplify.
21918 (remove_attribute): Use is_attribute_p.
21919 * tree.h: Remove removed declarations.
21921 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21923 PR middle-end/81698
21924 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21925 instead of computing it in the function. Formatting fix.
21926 (expand_case): Don't rely on default_edge being the first edge,
21927 clear it if removing it, pass default_edge to
21928 emit_case_dispatch_table.
21929 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21932 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21934 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21935 insn in the function, emit NOP after the insn.
21937 2017-08-06 Tom de Vries <tom@codesourcery.com>
21939 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21942 2017-08-06 Tom de Vries <tom@codesourcery.com>
21944 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21947 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21949 PR tree-optimization/57371
21950 * match.pd: New pattern.
21952 2017-08-04 Marek Polacek <polacek@redhat.com>
21954 PR middle-end/81695
21955 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21956 perform the computation in offset_int.
21958 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21960 PR tree-optimization/81136
21961 * tree-vectorizer.h: Include tree-hash-traits.h.
21962 (vec_base_alignments): New typedef.
21963 (vec_info): Add a base_alignments field.
21964 (vect_record_base_alignments): Declare.
21965 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21967 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21968 (create_data_ref): Add an is_conditional_in_stmt argument.
21969 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21970 the is_conditional_in_stmt field.
21971 (data_ref_loc): Add an is_conditional_in_stmt field.
21972 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21973 (find_data_references_in_stmt): Update call to create_data_ref.
21974 (graphite_find_data_references_in_stmt): Likewise.
21975 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21976 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21977 (vect_record_base_alignment): New function.
21978 (vect_record_base_alignments): Likewise.
21979 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21980 for nested statements even if we fail to compute a misalignment.
21981 Use pooled base alignments for unconditional references.
21982 (vect_find_same_alignment_drs): Compare base addresses instead
21984 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21985 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21987 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21989 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21990 Add an explicit name for the enum. Use auto_vec for slp_instances
21991 and grouped_stores.
21992 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21994 (_bb_vec_info): Add a constructor and destructor.
21995 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21996 (destroy_loop_vec_info): Delete.
21997 (vect_destroy_datarefs): Likewise.
21998 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21999 (vec_info::vec_info): New function.
22000 (vec_info::~vec_info): Likewise.
22001 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22002 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22003 destroy_loop_vec_info.
22004 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22005 (_loop_vec_info::_loop_vec_info): ...this.
22006 (destroy_loop_vec_info): Replace with...
22007 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22008 the stmt_vec_infos. Leave handling of vec_info information to its
22009 destructor. Remove explicit vector releases.
22010 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22011 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22012 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22013 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22014 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22015 (destroy_bb_vec_info): Replace with...
22016 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22017 information to its destructor.
22018 (vect_slp_analyze_bb_1): Use new and delete instead of
22019 new_bb_vec_info and destroy_bb_vec_info.
22020 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22023 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22025 * tree-data-ref.h (subscript): Add access_fn field.
22026 (data_dependence_relation): Add could_be_independent_p.
22027 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22028 (same_access_functions): Move to tree-data-ref.c.
22029 * tree-data-ref.c (ref_contains_union_access_p): New function.
22030 (access_fn_component_p): Likewise.
22031 (access_fn_components_comparable_p): Likewise.
22032 (dr_analyze_indices): Add a reference to access_fn_component_p.
22033 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22035 (constant_access_functions): Likewise.
22036 (add_other_self_distances): Likewise.
22037 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22038 (initialize_data_dependence_relation): Use XCNEW and remove
22039 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22040 of access functions that have the same type. Allow the
22041 subsequence to end with different bases in some circumstances.
22042 Record the chosen access functions in SUB_ACCESS_FN.
22043 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22044 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22045 (subscript_dependence_tester_1): Likewise dra and drb.
22046 (build_classic_dist_vector): Update calls accordingly.
22047 (subscript_dependence_tester): Likewise.
22048 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22049 DDR_COULD_BE_INDEPENDENT_P.
22050 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22051 comp_alias_ddrs instead of may_alias_ddrs.
22052 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22054 (vect_analyze_data_ref_dependence): Use it if
22055 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22056 distance vectors if that fails.
22057 (dependence_distance_ge_vf): New function.
22058 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22059 LOOP_VINFO_MAY_ALIAS_DDRS.
22061 2017-08-04 Richard Biener <rguenther@suse.de>
22063 PR middle-end/81705
22064 * fold-const.c (fold_binary_loc): Properly restrict
22065 minus_var0 && minus_var1 case when associating undefined overflow
22068 2017-08-04 Tom de Vries <tom@codesourcery.com>
22070 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22072 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22074 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22075 Don't start diagnostic messages with a capital letter.
22076 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22078 (rs6000_invalid_builtin): Likewise.
22079 (rs6000_trampoline_init): Likewise.
22081 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22084 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22085 after setting changeable df flags.
22087 2017-08-03 Richard Biener <rguenther@suse.de>
22089 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22090 up if the use is in USE - X.
22092 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22094 * toplev.c (dumpfile.h): New include.
22095 (internal_error_reentered): New static function. Use it...
22096 (internal_error_function): ...here to handle reentered internal_error.
22098 2017-08-03 Richard Biener <rguenther@suse.de>
22100 PR middle-end/81148
22101 * fold-const.c (split_tree): Add minus_var and minus_con
22102 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22103 here but always use minus_*.
22104 (associate_trees): Assert we never associate with MINUS_EXPR
22105 and NULL first operand. Do not recurse for PLUS_EXPR operands
22106 when associating as MINUS_EXPR either.
22107 (fold_binary_loc): Track minus_var and minus_con.
22109 2017-08-03 Tom de Vries <tom@codesourcery.com>
22112 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22113 ACCEL_COMPILER, apply finish_options on
22114 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22116 2017-08-03 Tom de Vries <tom@codesourcery.com>
22119 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22120 function_entry_patch_area_size > 0.
22122 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22125 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22126 instead of 10??LU, perform unit multiplication in wide_int,
22127 don't change alloc_object_size_limit if the limit is larger
22130 PR tree-optimization/81655
22131 PR tree-optimization/81588
22132 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22133 the case when ranges[i].low and high are 1 for unsigned type with
22136 PR middle-end/81052
22137 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22138 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22140 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22142 * tree-vrp.h: Add include guard.
22144 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22147 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22148 (ud2): New insn pattern.
22149 * config/i386/i386.c (ix86_expand_epilogue):
22150 For naked functions, generate ud2 instead of trap insn.
22152 2017-08-02 Marek Polacek <polacek@redhat.com>
22155 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22157 2017-08-02 Tom de Vries <tom@codesourcery.com>
22158 Cesar Philippidis <cesar@codesourcery.com>
22160 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22161 Add missing edge probabilities.
22163 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22165 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22166 Correct endianness.
22168 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22170 PR middle-end/79499
22171 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22172 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22173 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22174 prologue_seq sequences - if any.
22176 2017-08-02 Richard Biener <rguenther@suse.de>
22178 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22179 via vectors if supported, integer extracts via punning if supported
22180 or otherwise vector extracts.
22182 2017-08-02 Richard Biener <rguenther@suse.de>
22184 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22186 (bitmap_insert_into_set): ... this.
22188 2017-08-02 Richard Biener <rguenther@suse.de>
22190 PR tree-optimization/81633
22192 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22194 PR tree-optimization/71752
22195 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22197 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22199 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22200 (machine_function::call_ms2sysv_pad_out): Remove field.
22201 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22202 (ix86_compute_frame_layout): Likewise.
22204 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22207 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22208 attribute with interrupt attribute.
22210 2017-08-01 Andrew Pinski <apinski@cavium.com>
22212 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22213 BIT_INSERT_EXPR's operand 1
22214 to see if the types precision matches.
22216 2017-08-01 Martin Liska <mliska@suse.cz>
22218 PR middle-end/70140
22219 * builtins.c (expand_builtin_memcpy_args): Remove.
22220 (expand_builtin_memcpy): Call newly added function
22221 expand_builtin_memory_copy_args.
22222 (expand_builtin_memcpy_with_bounds): Likewise.
22223 (expand_builtin_mempcpy): Remove last argument.
22224 (expand_builtin_mempcpy_with_bounds): Likewise.
22225 (expand_builtin_memory_copy_args): New function created from
22226 expand_builtin_mempcpy_args with small modifications.
22227 (expand_builtin_mempcpy_args): Remove.
22228 (expand_builtin_stpcpy): Remove unused argument.
22229 (expand_builtin): Likewise.
22230 (expand_builtin_with_bounds): Likewise.
22232 2017-08-01 Martin Liska <mliska@suse.cz>
22235 Make mempcpy more optimal (PR middle-end/70140).
22237 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22240 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22241 __builtin_vec_cmpne verify both arguments are compatible vectors
22242 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22243 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22244 move computation of aligned to after checking the argument types.
22248 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22249 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22252 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22253 Doug Rupp <rupp@adacore.com>
22254 Olivier Hainque <hainque@adacore.com>
22256 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22257 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22259 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22260 for TARGET_OS_CPP_BUILTIN.
22261 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22262 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22264 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22265 passing required abi options to the assembler for EABI configurations.
22266 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22267 of .text.hot and .text.unlikely sections for kernel modules when
22268 using ARM style exceptions.
22269 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22270 options. Add EXTRA_CC1_SPEC.
22271 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22273 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22275 (ARM_TARGET2_DWARF_FORMAT): Define.
22276 * config/arm/t-vxworks: Adjust multilib control to removal of the
22277 Diab command line options.
22279 2017-08-01 Martin Liska <mliska@suse.cz>
22281 PR gcov-profile/81561
22282 * gcov.c (unblock): Make unblocking safe as we need to preserve
22283 index correspondence of blocks and block_lists.
22285 2017-08-01 Richard Biener <rguenther@suse.de>
22287 PR tree-optimization/81181
22288 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22289 (compute_antic): ... end of iteration here.
22291 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22293 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22294 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22295 (ftree-slp-vectorize): Likewise.
22296 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22297 can no longer be set independent of flag_tree_loop_vectorize.
22298 * omp-general.c (emp_max_vf): Likewise.
22299 * opts.c (enable_fdo_optimizations): Remove references to
22300 flag_tree_vectorize, these are now implicit.
22301 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22302 and leave it for the options machinery.
22304 2017-08-01 Martin Liska <mliska@suse.cz>
22306 PR middle-end/70140
22307 * builtins.c (expand_builtin_memcpy_args): Remove.
22308 (expand_builtin_memcpy): Call newly added function
22309 expand_builtin_memory_copy_args.
22310 (expand_builtin_memcpy_with_bounds): Likewise.
22311 (expand_builtin_mempcpy): Remove last argument.
22312 (expand_builtin_mempcpy_with_bounds): Likewise.
22313 (expand_builtin_memory_copy_args): New function created from
22314 expand_builtin_mempcpy_args with small modifications.
22315 (expand_builtin_mempcpy_args): Remove.
22316 (expand_builtin_stpcpy): Remove unused argument.
22317 (expand_builtin): Likewise.
22318 (expand_builtin_with_bounds): Likewise.
22320 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22323 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22324 print "ds:" only for immediates in generic address space.
22326 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22329 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22330 (ix86_function_ok_for_sibcall): Return false for naked functions.
22332 2017-08-01 Richard Biener <rguenther@suse.de>
22334 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22335 (compute_antic): Seed worklist with exit block predecessors.
22336 * cfganal.c (dfs_find_deadend): For a cycle return the source
22337 of the edge closing it.
22339 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22341 * config/aarch64/aarch64.c
22342 (aarch64_can_const_movi_rtx_p): Move 0 check.
22344 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22346 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22347 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22349 * match.pd: Ditto in address comparison pattern.
22351 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22353 PR tree-optimization/81627
22354 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22355 closed ssa form for store-store chain.
22357 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22359 PR tree-optimization/81620
22360 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22361 for store-store chain.
22363 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22365 PR tree-optimization/81588
22366 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22367 ranges[i].in_p, invert comparison code ccode. For >/>=,
22368 swap rhs1 and rhs2 and comparison code unconditionally,
22369 for </<= don't do that. Don't swap rhs1/rhs2 again if
22370 ranges[i].in_p, instead invert comparison code ccode if
22371 opcode or oe->rank is BIT_IOR_EXPR.
22374 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22375 a direct optab to conversion optab.
22376 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22377 with GET_MODE_INNER as last argument instead of optab_handler.
22378 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22379 vector extraction if possible and optab is available.
22380 * expr.c (store_constructor): Use convert_optab_handler instead
22381 of optab_handler. Use vector initialization from smaller
22382 vectors if possible and optab is available.
22383 * tree-vect-stmts.c (vectorizable_load): Likewise.
22384 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22385 now have two modes.
22386 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22387 of vec_init from half-sized vectors with the same element mode.
22388 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22389 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22390 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22391 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22392 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22393 after mode in gen_vec_extract* calls.
22394 (vec_extract<mode>): Renamed to ...
22395 (vec_extract<mode><ssescalarmodelower>): ... this.
22396 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22397 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22398 element mode after mode in gen_vec_init* calls.
22399 (VEC_INIT_HALF_MODE): New mode iterator.
22400 (vec_init<mode>): Renamed to ...
22401 (vec_init<mode><ssescalarmodelower>): ... this.
22402 (vec_init<mode><ssehalfvecmodelower>): New expander.
22403 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22404 (vec_extractv2sfsf): ... this.
22405 (vec_initv2sf): Renamed to ...
22406 (vec_initv2sfsf): ... this.
22407 (vec_extractv2si): Renamed to ...
22408 (vec_extractv2sisi): ... this.
22409 (vec_initv2si): Renamed to ...
22410 (vec_initv2sisi): ... this.
22411 (vec_extractv4hi): Renamed to ...
22412 (vec_extractv4hihi): ... this.
22413 (vec_initv4hi): Renamed to ...
22414 (vec_initv4hihi): ... this.
22415 (vec_extractv8qi): Renamed to ...
22416 (vec_extractv8qiqi): ... this.
22417 (vec_initv8qi): Renamed to ...
22418 (vec_initv8qiqi): ... this.
22419 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22420 (vec_init<mode>): Renamed to ...
22421 (vec_init<mode><VEC_base_l>): ... this.
22422 (vec_extract<mode>): Renamed to ...
22423 (vec_extract<mode><VEC_base_l>): ... this.
22424 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22425 (vec_initv2sfsf): ... this.
22426 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22427 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22428 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22429 element mode after mode in gen_vec_init* calls.
22430 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22431 (vec_init<mode><Vel>): ... this.
22432 (vec_extract<mode>): Renamed to ...
22433 (vec_extract<mode><Vel>): ... this.
22434 * config/aarch64/iterators.md (Vel): New mode attribute.
22435 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22436 Add element mode after mode in gen_vec_extract* calls.
22437 * config/s390/vector.md (non_vec_l): New mode attribute.
22438 (vec_extract<mode>): Renamed to ...
22439 (vec_extract<mode><non_vec_l>): ... this.
22440 (vec_init<mode>): Renamed to ...
22441 (vec_init<mode><non_vec_l>): ... this.
22442 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22443 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22445 * config/arm/iterators.md (V_elem_l): New mode attribute.
22446 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22447 (vec_extract<mode><V_elem_l>): ... this.
22448 (vec_extractv2di): Renamed to ...
22449 (vec_extractv2didi): ... this.
22450 (vec_init<mode>): Renamed to ...
22451 (vec_init<mode><V_elem_l>): ... this.
22452 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22453 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22454 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22455 Add element mode after gen_vec_extract* calls.
22456 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22457 (vec_init<mode><unitmode>): ... this.
22458 (vec_extract<mode>): Renamed to ...
22459 (vec_extract<mode><unitmode>): ... this.
22460 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22461 (vec_init<mode><unitmode>): ... this.
22462 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22463 (vec_initv2sfsf): ... this.
22464 (vec_extractv2sf): Renamed to ...
22465 (vec_extractv2sfsf): ... this.
22466 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22467 Add element mode after gen_vec_extract* calls.
22468 * config/mips/mips.md (unitmode): New mode iterator.
22469 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22470 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22471 * config/spu/spu.md (inner_l): New mode attribute.
22472 (vec_init<mode>): Renamed to ...
22473 (vec_init<mode><inner_l>): ... this.
22474 (vec_extract<mode>): Renamed to ...
22475 (vec_extract<mode><inner_l>): ... this.
22476 * config/sparc/sparc.md (veltmode): New mode iterator.
22477 (vec_init<VMALL:mode>): Renamed to ...
22478 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22479 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22480 (vec_initv2sisi): ... this.
22481 (vec_initv2sf): Renamed to ...
22482 (vec_initv2sfsf): ... this.
22483 (vec_extractv2sf): Renamed to ...
22484 (vec_extractv2sfsf): ... this.
22485 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22486 (vec_init<mode>): Renamed to ...
22487 (vec_init<mode><VEC_base_l>): ... this.
22488 (vec_extract<mode>): Renamed to ...
22489 (vec_extract<mode><VEC_base_l>): ... this.
22490 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22491 (vec_initv2sfsf): ... this.
22492 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22493 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22494 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22495 gen_vec_init* calls.
22497 2017-08-01 Richard Biener <rguenther@suse.de>
22499 PR tree-optimization/81297
22500 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22501 TREE_OVERFLOW from INTEGER_CSTs.
22503 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22505 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22507 2017-07-31 Carl Love <cel@us.ibm.com>
22509 * config/rs6000/rs6000-c: Add support for built-in functions
22510 vector signed char vec_xl_be (signed long long, signed char *);
22511 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22512 vector signed int vec_xl_be (signed long long, signed int *);
22513 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22514 vector signed long long vec_xl_be (signed long long, signed long long *);
22515 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22516 vector signed short vec_xl_be (signed long long, signed short *);
22517 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22518 vector double vec_xl_be (signed long long, double *);
22519 vector float vec_xl_be (signed long long, float *);
22520 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22521 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22522 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22524 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22525 (altivec_expand_builtin): Add switch statement to call
22526 altivec_expand_xl_be for each builtin.
22527 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22528 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22529 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22530 __builtin_vsx_le_be_v16qi.
22531 * doc/extend.texi: Update the built-in documentation file for the
22532 new built-in functions.
22534 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22537 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22539 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22541 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22543 * config.gcc: Add z14.
22544 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22545 CPU model numbers for z13s and z14.
22546 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22548 * config/s390/s390-opts.h (enum processor_type): Rename
22549 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22550 * config/s390/s390.c (processor_table): Add field for CPU name to
22551 be passed to Binutils.
22552 (s390_asm_output_machine_for_arch): Use the new field in
22553 processor_table for Binutils.
22554 (s390_expand_builtin): Replace arch12 with z14.
22555 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22556 (s390_get_sched_attrmask): Likewise.
22557 (s390_get_unit_mask): Likewise.
22558 * config/s390/s390.opt: Add z14 to processor_type enum.
22560 2017-07-31 Martin Jambor <mjambor@suse.cz>
22563 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22564 regardless of optimization level.
22566 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22567 Martin Liska <mliska@suse.cz>
22569 * predict.def: Remove old comment and adjust probability.
22570 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22571 PREDICT statements.
22573 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22576 * config/i386/i386.c (ix86_function_naked): New function.
22577 (ix86_can_use_return_insn_p): Return false for naked functions.
22578 (ix86_expand_prologue): Skip prologue for naked functions.
22579 (ix86_expand_epilogue): Skip epilogue for naked functions
22580 and emit trap instruction.
22581 (ix86_warn_func_return): New function.
22582 (ix86_attribute_table): Add "naked" attribute specification.
22583 (TARGET_WARN_FUNC_RETURN): Define.
22584 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22586 2017-07-31 Martin Liska <mliska@suse.cz>
22588 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22589 (dump_gimple_bb_header): Always dump BB info.
22590 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22592 2017-07-31 Martin Liska <mliska@suse.cz>
22595 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22596 also with current_function_decl non-null equality.
22598 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22601 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22602 change type to the element type, instead add eltype variable and
22603 use it where we are interested in the element type.
22605 PR tree-optimization/81603
22606 * ipa-polymorphic-call.c
22607 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22608 offset arithmetic in offset_int, bail out if the resulting bit offset
22609 doesn't fit into shwi.
22611 2017-07-31 Martin Liska <mliska@suse.cz>
22613 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22614 (gimplify_save_expr): Fix comment.
22616 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22619 * config/i386/i386.c (ix86_function_arg): Update arguments for
22621 (ix86_compute_frame_layout): Set the initial stack offset to
22622 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22623 INCOMING_FRAME_SP_OFFSET.
22624 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22625 stack before exception handler returns.
22626 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22627 the 'ERROR_CODE' for exception handler.
22629 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22631 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22632 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22633 (ASM_OUTPUT_REG_POP): Ditto.
22634 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22635 instead of asm_fprintf to output pure string.
22637 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22639 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22640 to imported_module_or_decl hook.
22641 (debug_nothing_tree_tree_tree_bool): Remove.
22642 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22643 * debug.c (do_nothing_debug_hooks): Use
22644 debug_nothing_tree_tree_tree_bool_bool instead of
22645 debug_nothing_tree_tree_tree_bool.
22646 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22647 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22648 * sdbout.c (sdb_debug_hooks): Likewise.
22649 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22650 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22652 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22653 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22654 attribute, don't add anything.
22656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22658 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22659 (fold_build2_stat_loc): Likewise.
22660 (fold_build3_stat_loc): Likewise.
22661 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22662 (fold_build1_loc): Remove macro.
22663 (fold_build2_loc): Likewise.
22664 (fold_build3_loc): Likewise.
22666 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22668 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22669 (gimple_build_debug_bind_source_stat): Likewise.
22670 * gimple.h (gimple_build_debug_bind): Remove macro.
22671 (gimple_build_debug_bind_source): Likewise.
22673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22675 * bitmap.c (bitmap_alloc): Adjust.
22676 (bitmap_gc_alloc): Likewise.
22677 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22681 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22682 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22683 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22684 (bitmap_gc_alloc_stat): Likewise.
22685 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22687 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22689 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22690 * rtl.h (shallow_copy_rtx): Remove macro.
22692 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22694 * emit-rtl.c (gen_raw_REG): Adjust.
22695 * gengenrtl.c (gendef): Likewise.
22696 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22697 * rtl.h (rtx_alloc): Remove macro.
22699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22701 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22702 (build_tree_list_stat): Likewise.
22703 * tree.h (build_tree_list): Remove macro.
22704 (build_tree_list_vec): Likewise.
22706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22708 * tree.c (make_vector_stat): Remove _stat from name.
22709 (build_vector_stat): Likewise.
22710 * tree.h (make_vector_stat): Remove macro.
22711 (build_vector_stat): Likewise.
22713 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22715 * tree.h (build_var_debug_value): Remove prototype.
22717 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22719 * tree.c (tree_cons_stat): Remove _stat from name.
22720 * tree.h (tree_cons): Remove macro.
22722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22724 * tree.c (build_vl_exp_stat): Remove _stat from name.
22725 * tree.h (build_vl_exp): Remove macro.
22727 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22729 * tree.c (build_decl_stat): Remove _stat from name.
22730 * tree.h (build_decl): Remove macro.
22732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22734 * gimple.c (gimple_build_with_ops_stat): Adjust.
22735 (gimple_alloc_stat): Remove _stat from name.
22736 * gimple.h (gimple_alloc): Remove macro.
22738 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22740 * tree.c (make_tree_vec_stat): Remove _stat from name.
22741 (grow_tree_vec_stat): Likewise.
22742 * tree.h (make_tree_vec_stat): Adjust prototype.
22743 (grow_tree_vec_stat): Likewise.
22744 (make_tree_vec): Remove macro.
22745 (grow_tree_vec): Likewise.
22747 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22749 * fold-const.c (fold_build1_stat_loc): Adjust.
22750 (fold_build2_stat_loc): Likewise.
22751 (fold_build3_stat_loc): Likewise.
22752 * tree.c (build0_stat): Remove _stat from name.
22753 (build1_stat): Likewise.
22754 (build2_stat): Likewise.
22755 (build3_stat): Likewise.
22756 (build4_stat): Likewise.
22757 (build5_stat): Likewise.
22758 * tree.h (build1_loc): Remove macro, and rename _stat function
22760 (build2_loc): Likewise.
22761 (build3_loc): Likewise.
22762 (build4_loc): Likewise.
22763 (build5_loc): Likewise.
22765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22767 * tree.c (make_int_cst_stat): Remove _stat from name.
22768 * tree.h (make_int_cst_stat): Adjust prototype.
22769 (make_int_cst): Remove macro.
22771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22773 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22774 * tree.h (make_tree_binfo_stat): Adjust prototype.
22775 (make_tree_binfo): Remove.
22777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22779 * tree.c (copy_node_stat): Rename to copy_node.
22780 (build_distinct_type_copy): Adjust.
22781 * tree.h (copy_node_stat): Adjust prototype.
22782 (copy_node): Remove macro.
22784 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22786 * tree.c (make_node_stat): rename to make_node.
22787 (build_tree_list_stat): Adjust.
22788 (build0_stat): Likewise.
22789 (build2_stat): Likewise.
22790 (build3_stat): Likewise.
22791 (build4_stat): Likewise.
22792 (build5_stat): Likewise.
22793 (build_decl_stat): Likewise.
22794 * tree.h (make_node_stat): Adjust prototype.
22795 (make_node): remove macro.
22797 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22799 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22800 (PPC_FEATURE2_SCV): Likewise.
22801 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22803 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22805 * config/aarch64/aarch64.c
22806 (aarch64_internal_mov_immediate): Add new special pattern.
22807 * config/aarch64/aarch64.md (*movdi_aarch64):
22808 Add reg/32bit const mov case.
22810 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22811 Richard Sandiford <richard.sandiford@linaro.org>
22813 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22814 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22815 Add integer and movi cases.
22816 (movi-split-hf-df-sf split, fp16): New.
22817 (enabled): Added TARGET_FP_F16INST.
22818 * config/aarch64/iterators.md (GPF_HF): New.
22819 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22821 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22823 * config/aarch64/aarch64.c
22824 (aarch64_simd_container_mode): Add prototype.
22825 (aarch64_expand_mov_immediate): Add HI support.
22826 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22827 (aarch64_can_const_movi_rtx_p): New.
22828 (aarch64_preferred_reload_class):
22829 Remove restrictions of using FP registers for certain SIMD operations.
22830 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22831 (aarch64_valid_floating_const): Add integer move validation.
22832 (aarch64_simd_imm_scalar_p): Remove.
22833 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22834 (aarch64_legitimate_constant_p): Expand list of supported cases.
22835 * config/aarch64/aarch64-protos.h
22836 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22837 (aarch64_reinterpret_float_as_int): New.
22838 (aarch64_simd_imm_scalar_p): Remove.
22839 * config/aarch64/constraints.md (Uvi): New.
22840 (Dd): Split into Ds and new Dd.
22841 * config/aarch64/aarch64.md (*movsi_aarch64):
22843 (*movdi_aarch64): Add SIMD mov case.
22845 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22847 * tree-predcom.c: (struct chain): Handle store-store chain in which
22848 stores for elimination only store loop invariant values.
22849 (execute_pred_commoning_chain): Ditto.
22850 (prepare_initializers_chain_store_elim): Ditto.
22851 (prepare_finalizers): Ditto.
22852 (is_inv_store_elimination_chain): New function.
22853 (initialize_root_vars_store_elim_1): New function.
22855 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22857 * tree-predcom.c: Revise general description of the pass.
22858 (enum chain_type): New enum type for store elimination.
22859 (struct chain): New field supporting store elimination.
22860 (struct component): Ditto.
22861 (dump_chain): Dump store-stores chain.
22862 (release_chain): Release resources.
22863 (split_data_refs_to_components): Compute and create component
22864 contains only stores for elimination.
22865 (get_chain_last_ref_at): New function.
22866 (make_invariant_chain): Initialization.
22867 (make_rooted_chain): Specify chain type in parameter and record it.
22868 (add_looparound_copies): Skip for store-stores chain.
22869 (determine_roots_comp): Compute type of chain and pass it to
22871 (initialize_root_vars_store_elim_2): New function.
22872 (finalize_eliminated_stores): New function.
22873 (remove_stmt): Handle store for elimination.
22874 (execute_pred_commoning_chain): Execute predictive commoning on
22875 store-store chains.
22876 (determine_unroll_factor): Skip unroll for store-stores chain.
22877 (prepare_initializers_chain_store_elim): New function.
22878 (prepare_initializers_chain): Hanlde store-store chain.
22879 (prepare_finalizers_chain, prepare_finalizers): New function.
22880 (tree_predictive_commoning_loop): Return integer value indicating
22881 if loop is unrolled or lcssa form is corrupted.
22882 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22884 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22886 * tree-predcom.c (initialize_root): Delete.
22887 (execute_pred_commoning_chain): Initialize root vars and replace
22888 reference of non-combined chain directly, rather than call above
22891 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22893 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22894 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22896 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22898 * tree-predcom.c (struct chain): New field init_seq.
22899 (release_chain): Release init_seq.
22900 (prepare_initializers_chain): Record intialization stmts in above
22902 (insert_init_seqs): New function.
22903 (tree_predictive_commoning_loop): Call insert_init_seqs.
22905 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22907 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22909 2017-07-28 Richard Biener <rguenther@suse.de>
22911 * match.pd: Remove superfluous :c.
22912 * genmatch.c (simplify::id): Add member.
22913 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22915 (current_id): New global.
22916 (dt_node::parent): Move from ...
22917 (dt_operand::parent): ... here. Add for_id member.
22918 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22919 (decision_tree::find_node): Relax order requirement when
22920 merging DT_TRUE nodes to ones inbetween the current simplify
22921 and the one we try to merge with. Add diagnostic whenever
22922 we need to enforce pattern order by not merging.
22923 (decision_tree::insert): Set current_id.
22924 (decision_tree::print_node): Dump parent node and for_id.
22925 (parser::last_id): Add member.
22926 (parser::push_simplify): Assign unique id.
22927 (parser::parser): Initialize last_id.
22929 2017-07-28 Martin Liska <mliska@suse.cz>
22932 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22933 gimple_build_debug_bind.
22935 2017-07-28 Richard Biener <rguenther@suse.de>
22937 PR tree-optimization/81502
22938 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22940 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22942 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22943 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22944 for BIT_FIELD_REF args.
22945 * fold-const.c (make_bit_field_ref): Likewise.
22946 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22948 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22951 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22952 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22953 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22954 Or it into SANITIZER_UNDEFINED.
22955 * ubsan.c: Include gimple-fold.h and varasm.h.
22956 (ubsan_expand_ptr_ifn): New function.
22957 (instrument_pointer_overflow): New function.
22958 (maybe_instrument_pointer_overflow): New function.
22959 (instrument_object_size): Formatting fix.
22960 (pass_ubsan::execute): Call instrument_pointer_overflow
22961 and maybe_instrument_pointer_overflow.
22962 * internal-fn.c (expand_UBSAN_PTR): New function.
22963 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22964 * sanitizer.def (__ubsan_handle_pointer_overflow,
22965 __ubsan_handle_pointer_overflow_abort): New builtins.
22966 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22967 * internal-fn.def (UBSAN_PTR): New internal function.
22968 * opts.c (sanitizer_opts): Add pointer-overflow.
22969 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22970 * fold-const.c (build_range_check): Compute pointer range check in
22971 integral type if pointer arithmetics would be needed. Formatting
22974 2017-07-28 Martin Liska <mliska@suse.cz>
22977 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22978 parameters that are of a variable-length.
22980 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22982 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22984 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22985 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22986 (CRT_CALL_STATIC_FUNCTION): Likewise.
22987 (ASM_DEFAULT_SPEC): New define.
22988 (ASM_SPEC32): Likewise.
22989 (ASM_SPEC64): Likewise.
22990 (ASM_SPEC_COMMON): Likewise.
22991 (ASM_SPEC): Likewise.
22992 (INVALID_64BIT): Likewise.
22993 (LINK_OS_DEFAULT_SPEC): Likewise.
22994 (LINK_OS_SPEC32): Likewise.
22995 (LINK_OS_SPEC64): Likewise.
22996 (POWERPC_LINUX): Likewise.
22997 (PTRDIFF_TYPE): Likewise.
22998 (RESTORE_FP_PREFIX): Likewise.
22999 (RESTORE_FP_SUFFIX): Likewise.
23000 (SAVE_FP_PREFIX): Likewise.
23001 (SAVE_FP_SUFFIX): Likewise.
23002 (SIZE_TYPE): Likewise.
23003 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23004 (TARGET_64BIT): Likewise.
23005 (TARGET_64BIT): Likewise.
23006 (TARGET_AIX): Likewise.
23007 (WCHAR_TYPE_SIZE): Likewise.
23008 (WCHAR_TYPE): Undefine.
23009 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23010 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23011 (CPP_OS_RTEMS_SPEC): Delete.
23012 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23013 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23015 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23017 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23019 PR tree-optimization/81578
23020 * tree-parloops.c (build_new_reduction): Bail out if
23021 reduction_code isn't one of the standard OpenMP reductions.
23022 Move the details printing after that decision.
23024 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23026 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23027 related to reload_in_progress.
23028 (splat_input_operand): Likewise.
23029 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23031 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23033 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23034 (TARGET_INSTANTIATE_DECLS): Likewise.
23035 (legitimate_indexed_address_p): Delete reload_in_progress code.
23036 (rs6000_debug_legitimate_address_p): Likewise.
23037 (rs6000_eliminate_indexed_memrefs): Likewise.
23038 (rs6000_emit_le_vsx_store): Likewise.
23039 (rs6000_emit_move_si_sf_subreg): Likewise.
23040 (rs6000_emit_move): Likewise.
23041 (register_to_reg_type): Likewise.
23042 (rs6000_pre_atomic_barrier): Likewise.
23043 (rs6000_machopic_legitimize_pic_address): Likewise.
23044 (rs6000_allocate_stack_temp): Likewise.
23045 (rs6000_address_for_fpconvert): Likewise.
23046 (rs6000_address_for_altivec): Likewise.
23047 (rs6000_secondary_memory_needed_rtx): Delete function.
23048 (rs6000_check_sdmode): Likewise.
23049 (rs6000_alloc_sdmode_stack_slot): Likewise.
23050 (rs6000_instantiate_decls): Likewise.
23051 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23052 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23053 Delete reload_in_progress.
23054 (*vec_reload_and_plus_<mptrsize>): Likewise.
23055 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23056 (vsx_div_v2di): Likewise.
23057 (vsx_udiv_v2di): Likewise.
23059 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23061 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23062 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23063 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23064 (rs6000_debug_reg_global): Delete print of LRA status.
23065 (rs6000_option_override_internal): Delete dead LRA related code.
23066 (rs6000_lra_p): Delete function.
23067 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23069 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23071 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23072 * config/riscv/rtems.h: New file.
23074 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23075 Sudakshina Das <sudi.das@arm.com>
23077 * config/aarch64/aarch64.md
23078 (define_split for and<mode>3nr_compare): Move
23079 non aarch64_logical_operand to a register.
23080 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23081 register immediate operand to a register.
23082 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23084 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23086 PR middle-end/81564
23087 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23089 2017-07-27 Richard Biener <rguenther@suse.de>
23091 PR tree-optimization/81573
23092 PR tree-optimization/81494
23093 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23094 multi defuse cycle case.
23096 2017-07-27 Richard Biener <rguenther@suse.de>
23098 PR tree-optimization/81571
23099 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23102 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23104 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23105 earlier and only if MASK_FPU is set. Adjust formatting.
23107 2017-07-27 Martin Liska <mliska@suse.cz>
23109 * opt-functions.awk: Add validation of value of Init.
23110 * optc-gen.awk: Pass new argument.
23112 2017-07-27 Martin Liska <mliska@suse.cz>
23114 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23115 Fix wrong condition.
23117 2017-07-27 Martin Liska <mliska@suse.cz>
23119 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23120 BBs and edges seen by autoFDO.
23122 2017-07-27 Richard Biener <rguenther@suse.de>
23124 PR tree-optimization/81502
23125 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23126 with incompatible but same sized type.
23127 (execute_update_addresses_taken): Likewise.
23129 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23131 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23132 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23134 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23137 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23138 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23139 Change s_operand to memory_operand.
23141 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23143 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23144 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23145 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23146 Emit instructions rather than returning an expression. Handle TFmode
23147 and KFmode by casting to TImode.
23148 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23149 (rs6000_emit_le_vsx_store): Likewise.
23150 * config/rs6000/vsx.md (VSX_TI): New iterator.
23151 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23152 (*vsx_le_undo_permute_<mode>): Likewise.
23153 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23154 emit the split sequence.
23155 (*vsx_le_perm_store_<mode>): Likewise.
23157 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23159 PR tree-optimization/81555
23160 PR tree-optimization/81556
23161 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23162 if true, force CHANGED for the recursive invocation.
23163 (reassociate_bb): Remember original length of ops array, pass
23164 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23166 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23167 attributes for noipa attribute. For naked attribute use
23168 lookup_attribute first before lookup_attribute_spec.
23169 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23171 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23172 for functions with noipa attribute.
23173 (cgraph_externally_visible_p): Return true for functions with noipa
23175 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23176 for functions with noipa attribute.
23177 * doc/extend.texi: Document noipa function attribute.
23178 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23179 also for functions with noipa attribute.
23180 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23182 2017-07-26 Andrew Pinski <apinski@cavium.com>
23184 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23185 vec_unalign_load_cost and vec_unalign_store_cost.
23187 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23189 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23190 -mvsx-small-integer option.
23191 (ISA_3_0_MASKS_IEEE): Likewise.
23192 (OTHER_VSX_VECTOR_MASKS): Likewise.
23193 (POWERPC_MASKS): Likewise.
23194 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23195 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23196 code, only testing for DImode being allowed in non-VSX floating
23198 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23199 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23201 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23202 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23203 TARGET_P8_VECTOR test.
23204 (rs6000_secondary_reload_simple_move): Likewise.
23205 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23206 since TARGET_P9_VECTOR was already tested.
23207 (rs6000_opt_masks): Remove -mvsx-small-integer.
23208 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23209 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23211 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23212 test for TARGET_VEXTRACTUB was used, and that uses
23214 (p9 extract splitter): Likewise.
23215 (vsx_extract_<mode>_di_p9): Likewise.
23216 (vsx_extract_<mode>_store_p9): Likewise.
23217 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23218 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23219 the elimination of TARGET_VSX_SMALL_INTEGER.
23220 (vsx_extract_<mode>_p8): Likewise.
23221 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23222 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23223 (vsx_set_<mode>_p9): Likewise.
23224 (vsx_set_v4sf_p9): Likewise.
23225 (vsx_set_v4sf_p9_zero): Likewise.
23226 (vsx_insert_extract_v4sf_p9): Likewise.
23227 (vsx_insert_extract_v4sf_p9_2): Likewise.
23228 * config/rs6000/rs6000.md (sign extend splitter): Change
23229 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23230 (floatsi<mode>2_lfiwax_mem): Likewise.
23231 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23232 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23233 since a test for TARGET_P9_VECTOR was used.
23234 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23235 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23236 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23237 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23238 TARGET_P8_VECTOR test.
23239 (fix_trunc<mode>si2_stfiwx): Likewise.
23240 (fix_trunc<mode>si2_internal): Likewise.
23241 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23242 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23244 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23245 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23246 TARGET_P8_VECTOR test.
23247 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23248 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23249 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23251 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23252 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23253 since a test for TARGET_P9_VECTOR was used.
23254 (splitter for loading small constants): Likewise.
23256 2017-07-26 Andrew Pinski <apinski@cavium.com>
23258 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23261 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23264 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23265 (fp_valid_at): Likewise.
23267 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23269 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23270 (qdf24xx_addrcost_table): Likewise.
23271 (cortexa57_tunings): Update to use generic_branch_cost.
23272 (cortexa72_tunings): Likewise.
23273 (cortexa73_tunings): Likewise.
23274 (qdf24xx_tunings): Likewise.
23276 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23278 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23279 (thunderx2t99_branch_cost): Likewise.
23280 (cortexa35_tunings): Update to use generic_branch_cost.
23281 (cortexa53_tunings): Likewise.
23282 (cortexa57_tunings): Likewise.
23283 (cortexa72_tunings): Likewise.
23284 (cortexa73_tunings): Likewise.
23285 (thunderx2t99_tunings): Likewise.
23287 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23289 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23290 (sparc_option_override): Honour MASK_FSMULD.
23291 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23292 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23293 * config/sparc/sparc.opt (mfsmuld): New option.
23294 * doc/invoke.texi (mfsmuld): Document option.
23296 2017-07-26 Marek Polacek <polacek@redhat.com>
23298 PR middle-end/70992
23299 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23301 2017-07-26 Richard Biener <rguenther@suse.de>
23303 * gimple-match-head.c (do_valueize): Return OP if valueize
23305 (get_def): New helper to get at the def stmt of a SSA name
23306 if valueize allows.
23307 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23308 do_valueize to get at the def stmt.
23309 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23311 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23313 PR middle-end/46932
23314 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23316 2017-07-26 Martin Liska <mliska@suse.cz>
23319 * function.c (expand_function_start): Make expansion of
23320 nonlocal_goto_save_area after parm_birth_insn.
23322 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23324 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23325 from all CPU target flags enable members.
23327 2017-07-26 Richard Biener <rguenther@suse.de>
23329 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23330 (decision_tree::gen): Make 'type' const.
23331 (write_predicate): Likewise.
23333 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23335 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23336 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23337 (rs6000_option_override_internal): Likewise.
23338 (rs6000_expand_vector_set): Likewise.
23339 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23340 (TARGET_UPPER_REGS_SF): Likewise.
23341 (TARGET_UPPER_REGS_DI): Likewise.
23342 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23343 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23344 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23345 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23346 (Splitters for DI constants in Altivec registers): Likewise.
23347 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23348 (vsx_set_v4sf_p9): Likewise.
23349 (vsx_set_v4sf_p9_zero): Likewise.
23350 (vsx_insert_extract_v4sf_p9): Likewise.
23351 (vsx_insert_extract_v4sf_p9_2): Likewise.
23353 2017-07-25 Carl Love <cel@us.ibm.com>
23355 * doc/extend.texi: Update the built-in documentation file for the
23356 existing built-in functions
23357 vector signed char vec_cnttz (vector signed char);
23358 vector unsigned char vec_cnttz (vector unsigned char);
23359 vector signed short vec_cnttz (vector signed short);
23360 vector unsigned short vec_cnttz (vector unsigned short);
23361 vector signed int vec_cnttz (vector signed int);
23362 vector unsigned int vec_cnttz (vector unsigned int);
23363 vector signed long long vec_cnttz (vector signed long long);
23364 vector unsigned long long vec_cnttz (vector unsigned long long);
23366 2017-07-25 Andrew Pinski <apinski@cavium.com>
23368 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23369 accesses where the use is for the first operand of a BIT_INSERT.
23371 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23374 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23375 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23377 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23379 * config/i386/gstabs.h: Delete.
23380 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23382 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23384 * config/i386/i386.c (ix86_decompose_address): Do not check for
23385 register RTX when looking at index_reg or base_reg.
23386 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23388 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23390 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23391 to update EH info here.
23393 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23395 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23397 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23399 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23401 2017-07-25 Torsten Duwe <duwe@suse.de>
23403 * common.opt: Introduce -fpatchable-function-entry
23404 command line option, and its variables function_entry_patch_area_size
23405 and function_entry_patch_area_start.
23406 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23407 including a two-value parser.
23408 * target.def (print_patchable_function_entry): New target hook.
23409 * targhooks.h (default_print_patchable_function_entry): New function.
23410 * targhooks.c (default_print_patchable_function_entry): Likewise.
23411 * toplev.c (process_options): Switch off IPA-RA if
23412 patchable function entries are being generated.
23413 * varasm.c (assemble_start_function): Look at the
23414 patchable-function-entry command line switch and current
23415 function attributes and maybe generate NOP instructions by
23416 calling the print_patchable_function_entry hook.
23417 * doc/extend.texi: Document patchable_function_entry attribute.
23418 * doc/invoke.texi: Document -fpatchable_function_entry
23419 command line option.
23420 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23422 * doc/tm.texi: Re-generate.
23424 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23427 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23428 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23430 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23432 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23434 2017-07-25 Richard Biener <rguenther@suse.de>
23436 PR tree-optimization/81455
23437 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23438 not walk in cycles when looking for guards.
23440 2017-07-25 Richard Biener <rguenther@suse.de>
23442 PR tree-optimization/81529
23443 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23444 when optimizing backedge uses.
23446 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23448 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23450 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23451 to dl_section_ref. On AIX, append an expression to subtract
23452 the size of the section length to dl_section_ref.
23454 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23456 * configure.ac: If any of the config.* scripts fail, exit 1.
23457 * configure: Regenerate.
23459 2017-07-25 Richard Biener <rguenther@suse.de>
23461 PR middle-end/81546
23462 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23463 of immediate uses, be more verbose on errors.
23465 2017-07-25 Richard Biener <rguenther@suse.de>
23467 PR tree-optimization/81510
23468 * tree-vect-loop.c (vect_is_simple_reduction): When the
23469 reduction stmt is not inside the loop bail out.
23471 2017-07-25 Richard Biener <rguenther@suse.de>
23473 PR tree-optimization/81303
23474 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23475 profitability check against LOOP_VINFO_NITERSM1.
23477 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23479 * domwalk.c (cmp_bb_postorder): Simplify.
23480 (sort_bbs_postorder): New function. Use it...
23481 (dom_walker::walk): ...here to optimize common cases.
23483 2017-07-25 Martin Liska <mliska@suse.cz>
23486 * ipa-visibility.c (function_and_variable_visibility): Make the
23487 redirection just on target that supports aliasing.
23488 Fix GNU coding style.
23490 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23493 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23495 * config/arm/rtems.h: Likewise.
23496 * config/bfin/rtems.h: Likewise.
23497 * config/i386/rtemself.h: Likewise.
23498 * config/lm32/rtems.h: Likewise.
23499 * config/m32c/rtems.h: Likewise.
23500 * config/m68k/rtemself.h: Likewise.
23501 * config/microblaze/rtems.h: Likewise.
23502 * config/mips/rtems.h: Likewise.
23503 * config/moxie/rtems.h: Likewise.
23504 * config/nios2/rtems.h: Likewise.
23505 * config/powerpcspe/rtems.h: Likewise.
23506 * config/rs6000/rtems.h: Likewise.
23507 * config/rtems.h: Likewise.
23508 * config/sh/rtems.h: Likewise.
23509 * config/sh/rtemself.h: Likewise.
23510 * config/sparc/rtemself.h: Likewise.
23512 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23515 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23516 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23517 * tree-ssa-structalias.c (alias_get_name): Same.
23519 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23522 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23523 instructions if no du chain is found.
23525 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23527 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23529 2017-07-25 Richard Biener <rguenther@suse.de>
23531 PR middle-end/81505
23532 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23535 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23537 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23538 upper-regs options.
23539 (ISA_2_7_MASKS_SERVER): Likewise.
23540 (ISA_3_0_MASKS_IEEE): Likewise.
23541 (OTHER_P8_VECTOR_MASKS): Likewise.
23542 (OTHER_VSX_VECTOR_MASKS): Likewise.
23543 (POWERPC_MASKS): Likewise.
23544 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23545 duplicate list of options.
23546 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23547 explicit -mupper-regs options.
23548 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23549 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23550 alias for -mupper-regs-df.
23551 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23552 (rs6000_init_hard_regno_mode_ok): Likewise.
23553 (rs6000_option_override_internal): Likewise.
23554 (rs6000_opt_masks): Likewise.
23555 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23556 options in terms of whether -mvsx or -mpower8-vector was used.
23557 (TARGET_UPPER_REGS_DI): Likewise.
23558 (TARGET_UPPER_REGS_SF): Likewise.
23559 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23560 -mupper-regs-* options.
23562 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23564 * passes.c (emergency_dump_function): Print some empty lines and a
23565 header before the RTL dump.
23567 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23569 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23571 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23574 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23575 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23577 2017-07-24 Carl Love <cel@us.ibm.com>
23579 * config/rs6000/rs6000-c.c: Add support for built-in functions
23580 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23581 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23582 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23583 vec_extract_fp_from_shortl): Add defines for the two builtins.
23584 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23585 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23587 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23588 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23589 * doc/extend.texi: Update the built-in documentation file for the
23590 new built-in function.
23592 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23595 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23597 * doc/generic.texi: Likewise.
23598 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23599 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23601 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23603 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23604 (aarch64_mls_elt_merge<mode>): Likewise.
23606 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23608 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23609 having __cxa_atexit.
23611 2017-07-23 Michael Collison <michael.collison@arm.com>
23613 * config/arm/arm.c (arm_option_override): Deprecate
23614 use of -mstructure-size-boundary.
23615 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23616 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23618 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23621 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23622 Reduce cost estimate for direct moves.
23624 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23627 * config/i386/i386.c (ix86_option_override_internal): Disable
23628 BMI, BMI2 and TBM instructions for -m16.
23630 2017-07-21 Carl Love <cel@us.ibm.com>
23632 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23633 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23634 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23635 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23636 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23637 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23638 VMULOSW): New enum "unspec" values.
23639 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23640 altivec_vmulosw): New patterns.
23641 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23642 VMULOSW): Add definitions.
23644 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23646 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23647 (qdf24xx): Likewise.
23648 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23649 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23650 (AARCH64_FL_V8_1): Renumber.
23651 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23652 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23653 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23654 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23655 rdma to feature modifiers list.
23657 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23659 PR middle-end/56727
23660 * ipa-visibility (function_and_variable_visibility): Convert
23661 recursive PLT call to direct call if appropriate.
23663 2017-07-21 Andrew Pinski <apinski@cavium.com>
23665 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23666 operand 1 to see if the types precision matches.
23667 * fold-const.c (operand_equal_p): Likewise.
23669 2017-07-21 Richard Biener <rguenther@suse.de>
23671 PR tree-optimization/81303
23672 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23673 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23674 (vect_peeling_hash_get_lowest_cost): Adjust.
23675 (vect_enhance_data_refs_alignment): Likewise. Use
23676 vect_get_peeling_costs_all_drs to compute the penalty for no
23677 peeling to match up costs.
23679 2017-07-21 Richard Biener <rguenther@suse.de>
23681 PR tree-optimization/81500
23682 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23683 we didn't identify a reduction path.
23685 2017-07-21 Tom de Vries <tom@codesourcery.com>
23686 Cesar Philippidis <cesar@codesourcery.com>
23688 PR gcov-profile/81442
23689 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23692 2017-07-21 Tom de Vries <tom@codesourcery.com>
23695 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23697 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23698 nvptx_override_options_after_change.
23700 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23702 * dwarf2out.c (output_file_names): Avoid double testing for
23703 dwarf_version >= 5.
23705 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23707 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23709 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23711 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23713 (try_crossjump_to_edge): Do not punt on partitioned functions.
23715 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23717 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23718 Put all BBs reachable only via paths crossing cold region to cold
23720 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23722 2016-07-21 Richard Biener <rguenther@suse.de>
23724 PR tree-optimization/81303
23725 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23726 into account prologue and epilogue iterations when raising
23727 min_profitable_iters to sth at least covering one vector iteration.
23729 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23731 * config/arm/arm.c (arm_test_cpu_arch_dat):
23734 2017-07-20 Nathan Sidwell <nathan@acm.org>
23736 Remove TYPE_METHODS.
23737 * tree.h (TYPE_METHODS): Delete.
23738 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23739 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23740 (dbxout_type_methods): Scan TYPE_FIELDS.
23741 (dbxout_type): Don't check TYPE_METHODS here.
23742 * function.c (use_register_for_decl): Always ignore register for
23743 class types when not optimizing.
23744 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23745 * tree.c (free_lang_data_in_type): Stitch out member functions and
23746 templates from TYPE_FIELDS.
23747 (build_distinct_type_copy, verify_type_variant,
23748 verify_type): Member fns are on TYPE_FIELDS.
23749 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23750 * tree-pretty-print.c (dump_generic_node): Likewise.
23752 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23755 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23756 V2TImode and V4TImode.
23757 (ix86_expand_vector_extract): Likewise.
23758 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23759 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23760 (ssescalarmode): Handle V4TImode and V2TImode.
23761 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23762 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23763 (VEXTRACTI128_MODE): New mode iterator.
23764 (splitter for *vec_extractv?ti first element): New.
23765 (VEC_INIT_MODE): New mode iterator.
23766 (vec_init<mode>): Consolidate 3 expanders into one using
23767 VEC_INIT_MODE mode iterator.
23769 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23771 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23772 non_spilled_static_chain_regno_p.
23774 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23776 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23778 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23780 * bb-reorder.c (connect_traces): Allow copying of blocks within
23783 2017-07-20 Richard Biener <rguenther@suse.de>
23785 * gimple.h (gimple_phi_result): Add gphi * overload.
23786 (gimple_phi_result_ptr): Likewise.
23787 (gimple_phi_arg): Likewise. Adjust index assert to only
23788 allow actual argument accesses rather than all slots available
23790 (gimple_phi_arg_def): Add gphi * overload.
23791 * tree-phinodes.c (make_phi_node): Initialize only actual
23793 (resize_phi_node): Clear memory not covered by old node,
23794 do not initialize excess argument slots.
23795 (reserve_phi_args_for_new_edge): Initialize new argument slot
23798 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23800 PR tree-optimization/81388
23802 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23804 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23805 by removing computation of may_be_zero.
23807 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23808 Tom de Vries <tom@codesourcery.com>
23810 PR middle-end/81030
23811 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23812 when gimple level profile disagrees with what RTL expander did.
23814 2017-07-20 Richard Biener <rguenther@suse.de>
23816 PR tree-optimization/61171
23817 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23818 (vect_analyze_stmt): Add slp instance parameter.
23819 (vectorizable_reduction): Likewise.
23820 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23821 (vect_is_simple_reduction): Deal with chains not detected
23822 as SLP reduction chain, specifically not properly associated
23823 chains containing a mix of plus/minus.
23824 (get_reduction_op): Remove.
23825 (get_initial_defs_for_reduction): Simplify, pass in whether
23826 this is a reduction chain, pass in the SLP node for the PHIs.
23827 (vect_create_epilog_for_reduction): Get the SLP instance as
23829 (vectorizable_reduction): Get the SLP instance as arg.
23830 During analysis remember the SLP node with the PHIs in the
23831 instance. Simplify getting at the vectorized reduction PHIs.
23832 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23833 through SLP instance.
23834 (vect_slp_analyze_operations): Likewise.
23835 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23836 (vect_transform_stmt): Likewise.
23838 2017-07-20 Tom de Vries <tom@codesourcery.com>
23840 PR tree-optimization/81489
23841 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23842 read of phi arg location to before loop that modifies phi.
23844 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23846 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23849 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23851 PR middle-end/81331
23852 * except.c (execute): Fix ordering issue.
23854 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23856 PR rtl-optimization/81423
23857 * combine.c (make_compound_operation_int): Don't try to optimize
23858 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23860 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23862 PR rtl-optimization/81423
23863 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23864 with a constant that is -1 in the truncated to mode.
23866 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23868 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23869 (determine_unlikely_bbs): ... here.
23870 * predict.h (propagate_unlikely_bbs_forward): Declare.
23871 * cfgexpand.c (pass_expand::execute): Use it.
23872 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23874 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23875 propagate_unlikely_bbs_forward.
23877 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23879 PR middle-end/81331
23880 * except.c (maybe_add_nop_after_section_switch): New function.
23883 2017-07-19 Tom de Vries <tom@codesourcery.com>
23885 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23887 2017-07-19 Tom de Vries <tom@codesourcery.com>
23889 * gimple.h (gimple_phi_arg): Make assert more strict.
23891 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23893 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23894 * config/rs6000/mmintrin.h: New file.
23895 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23897 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23899 PR tree-optimization/81346
23900 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23902 2017-07-19 Tom de Vries <tom@codesourcery.com>
23904 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23906 2017-07-19 Tom de Vries <tom@codesourcery.com>
23908 * config/nvptx/nvptx-modes.def: Add V2DImode.
23909 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23910 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23911 (nvptx_output_mov_insn): Handle lack of mov.b128.
23912 (nvptx_print_operand): Handle 'H' and 'L' codes.
23913 (nvptx_vector_mode_supported): Allow V2DImode.
23914 (nvptx_preferred_simd_mode): New function.
23915 (nvptx_data_alignment): New function.
23916 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23917 nvptx_preferred_simd_mode.
23918 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23920 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23922 2017-07-19 Tom de Vries <tom@codesourcery.com>
23924 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23925 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23926 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23927 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23928 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23929 (mov<VECIM>_insn): New define_insn.
23930 (define_expand "mov<VECIM>): New define_expand.
23932 2017-07-19 Tom de Vries <tom@codesourcery.com>
23934 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23936 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23938 PR tree-optimization/81346
23939 * fold-const.h (fold_div_compare, range_check_type): Declare.
23940 * fold-const.c (range_check_type): New function.
23941 (build_range_check): Use range_check_type.
23942 (fold_div_compare): No longer static, rewritten into
23943 a match.pd helper function.
23944 (fold_comparison): Don't call fold_div_compare here.
23945 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23946 as helper function.
23948 2017-07-19 Nathan Sidwell <nathan@acm.org>
23950 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23951 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23952 * tree.c (find_decls_types_r, verify_type): Use
23953 TYPE_{MIN,MAX}_VALUE_RAW.
23954 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23955 (hash_tree): Likewise.
23956 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23958 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23961 2017-07-18 Tom de Vries <tom@codesourcery.com>
23963 PR middle-end/81464
23964 * omp-expand.c (expand_omp_for_static_chunk): Handle
23965 equal-argument loop exit phi.
23967 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23970 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23971 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23972 operand 2 predicate.
23973 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23974 operand 2 predicate.
23975 (ror,rol -> rorx splitters): Use const_int_operand as
23976 operand 2 predicate.
23978 2017-06-18 Richard Biener <rguenther@suse.de>
23980 PR tree-optimization/81410
23981 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23982 the gap in the ! slp_perm SLP case after each group.
23984 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23986 PR middle-end/81463
23987 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23990 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23992 PR middle-end/81462
23993 * predict.c (set_even_probabilities): Cleanup; do not affect
23994 probabilities that are already known.
23995 (combine_predictions_for_bb): Call even when count is set.
23997 2017-07-18 Nathan Sidwell <nathan@acm.org>
23999 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24002 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24005 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24006 optimization for loop niter analysis.
24008 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24011 * config/avr/avr.c (avr_optimize_casesi): Don't use
24012 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24014 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24016 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24017 body_cost_vec from _vect_peel_extended_info.
24018 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24019 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24022 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24024 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24026 2017-07-18 Richard Biener <rguenther@suse.de>
24028 PR tree-optimization/80620
24029 PR tree-optimization/81403
24030 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24031 info when re-using a VN table entry.
24033 2017-07-18 Richard Biener <rguenther@suse.de>
24035 PR tree-optimization/81418
24036 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24037 vectype_in. Verify that with lane-reducing reduction operations
24038 we have a single def-use cycle.
24040 2017-07-17 Carl Love <cel@us.ibm.com>
24042 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24044 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24045 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24046 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24047 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24048 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24049 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24050 VMULOSW): New enum "unspec" values.
24051 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24052 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24053 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24054 altivec_vmulosw): New patterns.
24055 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24056 VMULOSW): Add definitions.
24058 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24060 * config/alpha/alpha.c: Include predict.h.
24062 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24064 * tree-vrp.c (compare_assert_loc): Fix comparison function
24065 to return predictable results.
24067 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24069 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24071 (subdi3): Likewise.
24072 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24073 * doc/invoke.texi (mexpand-adddi): Update text.
24075 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24077 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24078 that also clobbers the CC register. The old expand code is moved
24080 (*arc_clzsi2): ... here.
24081 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24082 the CC register. The old expand code is moved to ...
24083 (arc_ctzsi2): ... here.
24085 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24087 * config/arc/arc.opt (mindexed-loads): Use initial value
24088 TARGET_INDEXED_LOADS_DEFAULT.
24089 (mauto-modify-reg): Use initial value
24090 TARGET_AUTO_MODIFY_REG_DEFAULT.
24091 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24092 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24093 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24094 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24096 2017-07-17 Martin Liska <mliska@suse.cz>
24099 * opts.c (finish_options): Do not allow -fgnu-tm
24100 w/ -fsanitize={kernel-,}address. Say sorry.
24102 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24105 * tree-loop-distribution.c (classify_partition): Only assert on
24106 numer of iterations.
24107 (merge_dep_scc_partitions): Delete prameter. Update function call.
24108 (distribute_loop): Remove code handling loop with unknown niters.
24109 (pass_loop_distribution::execute): Skip loop with unknown niters.
24111 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24114 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24115 function sort_partitions_by_post_order.
24117 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24119 PR tree-optimization/81374
24120 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24121 the max index of basic blocks, rather than number of basic blocks.
24123 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24125 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24127 (arc_legitimate_pic_operand_p): Likewise.
24128 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24130 (arc_needs_pcl_p): Likewise.
24131 (arc_legitimate_pc_offset_p): Likewise.
24132 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24133 function is also used in constrains.md.
24134 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24135 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24136 PLUS. Only return true/false in known cases, otherwise assert.
24137 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24138 is already called in arc_legitimate_constant_p.
24139 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24141 (LEGITIMATE_PIC_OPERAND_P): Use
24142 arc_raw_symbolic_reference_mentioned_p function.
24143 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24148 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24149 Andrew Burgess <andrew.burgess@embecosm.com>
24151 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24152 (arc_return_address_register): New function.
24153 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24154 (arc_handle_fndecl_attribute): Add naked attribute.
24155 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24156 (TARGET_WARN_FUNC_RETURN): Likewise.
24157 (arc_allocate_stack_slots_for_args): New function.
24158 (arc_warn_func_return): Likewise.
24159 (machine_function): Change type fn_type.
24160 (arc_compute_function_type): Consider new naked function type,
24161 change function return type.
24162 (arc_must_save_register): Adapt to handle new
24163 arc_compute_function_type's return type.
24164 (arc_expand_prologue): Likewise.
24165 (arc_expand_epilogue): Likewise.
24166 (arc_return_address_regs): Delete.
24167 (arc_return_address_register): New function.
24168 (arc_epilogue_uses): Use above function.
24169 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24170 (arc_function_type): Change encoding, add naked type.
24171 (ARC_INTERRUPT_P): Change to handle the new encoding.
24172 (ARC_FAST_INTERRUPT_P): Likewise.
24173 (ARC_NORMAL_P): Define.
24174 (ARC_NAKED_P): Likewise.
24175 (arc_compute_function_type): Delete prototype.
24176 * config/arc/arc.md (in_ret_delay_slot): Use
24177 arc_return_address_register function.
24178 (simple_return): Likewise.
24179 (p_return_i): Likewise.
24181 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24183 PR tree-optimization/81428
24184 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24185 can't be built for those types.
24187 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24189 Remove stuff dead since r239246.
24191 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24192 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24193 (avr_inform_devices): Remove dead stuff.
24195 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24197 * config/arm/arm_neon.h: Fix softp typo.
24199 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24201 PR tree-optimization/81365
24202 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24203 aggregate moves onto bb predecessor edges, make sure there are no
24204 loads that could alias the lhs in between the start of bb and the
24207 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24210 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24211 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24212 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24214 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24216 PR tree-optimization/81396
24217 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24218 (init_symbolic_number): Initialize it to 1.
24219 (perform_symbolic_merge): Add n_ops from both operands into the new
24221 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24222 without base_addr as useless if they need more than one operation.
24223 (bswap_replace): Handle !bswap case for NULL base_addr.
24225 2017-07-17 Tom de Vries <tom@codesourcery.com>
24228 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24231 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24233 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24234 conditional builtin define __FIX_LEON3FT_B2BST.
24236 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24238 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24239 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24242 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24244 PR rtl-optimization/81424
24245 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24246 to remove potential trapping from operands if -fnon-call-exceptions.
24248 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24250 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24251 profile_proability for scalling.
24252 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24254 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24256 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24258 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24260 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24261 fixpoint arithmetics.
24263 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24265 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24266 fixpoint arithmetics.
24268 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24270 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24271 fixpoint arithmetics.
24273 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24275 * profile-count.h (profile_probability::from_reg_br_prob_note,
24276 profile_probability::to_reg_br_prob_note): New functions.
24277 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24278 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24279 * predict.c (probability_reliable_p): Update.
24280 (edge_probability_reliable_p): Update.
24281 (br_prob_note_reliable_p): Update.
24282 (invert_br_probabilities): Update.
24283 (add_reg_br_prob_note): New function.
24284 (combine_predictions_for_insn): Update.
24285 * asan.c (asan_clear_shadow): Update.
24286 * cfgbuild.c (compute_outgoing_frequencies): Update.
24287 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24288 (update_br_prob_note): Update.
24289 (rtl_verify_edges): Update.
24290 (purge_dead_edges): Update.
24291 (fixup_reorder_chain): Update.
24292 * emit-rtl.c (try_split): Update.
24293 * ifcvt.c (cond_exec_process_insns): Update.
24294 (cond_exec_process_if_block): Update.
24295 (dead_or_predicable): Update.
24296 * internal-fn.c (expand_addsub_overflow): Update.
24297 (expand_neg_overflow): Update.
24298 (expand_mul_overflow): Update.
24299 * loop-doloop.c (doloop_modify): Update.
24300 * loop-unroll.c (compare_and_jump_seq): Update.
24301 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24302 * predict.h: Update.
24303 * reorg.c (mostly_true_jump): Update.
24305 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24306 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24307 * config/arc/arc.c: (emit_unlikely_jump): Update.
24308 * config/arm/arm.c: (emit_unlikely_jump): Update.
24309 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24310 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24311 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24312 (ix86_print_operand): Update.
24313 (ix86_split_fp_branch): Update.
24314 (predict_jump): Update.
24315 * config/ia64/ia64.c (ia64_print_operand): Update.
24316 * config/mmix/mmix.c (mmix_print_operand): Update.
24317 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24318 (rs6000_expand_split_stack_prologue): Update.
24319 * config/rs6000/rs6000.c: Update.
24320 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24321 (s390_expand_vec_movstr): Update.
24322 (s390_expand_cs_tdsi): Update.
24323 (s390_expand_split_stack_prologue): Update.
24324 * config/sh/sh.c (sh_print_operand): Update.
24325 (expand_cbranchsi4): Update.
24326 (expand_cbranchdi4): Update.
24327 * config/sparc/sparc.c (output_v9branch): Update.
24328 * config/spu/spu.c (get_branch_target): Update.
24329 (ea_load_store_inline): Update.
24330 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24331 * config/tilepro/tilepro.c: Update.
24333 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24335 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24336 (gimplify_save_expr): Likewise.
24338 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24340 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24342 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24344 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24346 * timevar.def (TV_IPA_FNSUMMARY): Define.
24348 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24350 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24351 to back store errata sensitive sequence from being generated.
24352 (sqrtdf2_fix): Likewise.
24354 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24356 * tree-ssa-threadupdate.c (compute_path_counts,
24357 update_joiner_offpath_counts): Use profile_probability.
24359 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24362 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24364 * config/arm/arm-c.c (arm_cpu_builtins): Define
24365 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24367 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24369 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24370 array entries to represent __ieee128 versions of the
24371 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24372 scalar_extract_sig, and scalar_insert_exp built-in functions.
24373 (altivec_resolve_overloaded_builtin): Add special case handling
24374 for the __builtin_scalar_insert_exp function, as represented by
24375 the P9V_BUILTIN_VEC_VSIEDP constant.
24376 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24377 exponent support for __ieee128 argument.
24378 (VSESQP): Add scalar extract signature support for __ieee128
24380 (VSTDCNQP): Add scalar test negative support for __ieee128
24382 (VSIEQP): Add scalar insert exponent support for __int128 argument
24383 with __ieee128 result.
24384 (VSIEQPF): Add scalar insert exponent support for __ieee128
24385 argument with __ieee128 result.
24386 (VSTDCQP): Add scalar test data class support for __ieee128
24388 (VSTDCNQP): Add overload support for scalar test negative with
24389 __ieee128 argument.
24390 (VSTDCQP): Add overload support for scalar test data class
24391 __ieee128 argument.
24392 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24393 UNSPEC_VSX_SXSIGDP.
24394 (UNSPEC_VSX_SIEXPQP): New constant.
24395 (xsxexpqp): New insn for VSX scalar extract exponent quad
24397 (xsxsigqp): New insn for VSX scalar extract significand quad
24399 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24400 precision with floating point argument.
24401 (xststdcqp): New expand for VSX scalar test data class quad
24403 (xststdcnegqp): New expand for VSX scalar test negative quad
24405 (xststdcqp): New insn to match expansions for VSX scalar test data
24406 class quad precision and VSX scalar test negative quad precision.
24407 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24408 special case operand checking to enforce that second operand of
24409 VSX scalar test data class with quad precision argument is a 7-bit
24411 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24412 prototypes and descriptions of __ieee128 versions of
24413 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24414 scalar_test_data_class, and scalar_test_neg built-in functions.
24416 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24418 PR tree-optimization/81162
24419 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24420 replace a negate with an add.
24422 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24424 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24426 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24428 * config/arm/arm-c.c (arm_cpu_builtins): Define
24429 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24431 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24433 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24434 (armv8-r): Set ARM Cortex-R52 as default CPU.
24435 * config/arm/arm-tables.opt: Regenerate.
24436 * config/arm/arm-tune.md: Regenerate.
24437 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24439 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24440 extension for -mcpu=cortex-r52.
24442 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24444 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24445 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24446 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24447 (fp-armv8): Define it as FP_ARMv8 only.
24448 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24449 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24451 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24452 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24453 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24454 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24455 than TARGET_FPU_ARMV8.
24456 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24457 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24458 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24460 * config/arm/neon.md (neon_vrint): Likewise.
24461 (neon_vcvt): Likewise.
24462 (neon_<fmaxmin_op><mode>): Likewise.
24463 (<fmaxmin><mode>3): Likewise.
24464 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24465 * config/arm/predicates.md (arm_cond_move_operator): Check against
24466 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24468 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24470 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24471 to top of function.
24473 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24475 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24476 loop in comment with memset.
24478 2017-07-14 Martin Liska <mliska@suse.cz>
24480 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24481 * dwarf2out.c (is_java): Remove the function.
24482 (output_pubname): Remove usage of the function.
24483 (lower_bound_default): Remove usage of DW_LANG_Java.
24484 (gen_compile_unit_die): Likewise.
24485 * gcc.c: Remove compiler defaults for .java and .zip files.
24486 * gimple-expr.c (remove_suffix): Change as there's no longer
24487 extension than 4-letter one.
24488 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24489 (gimplify_save_expr): Likewise.
24490 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24491 as it's possible even for other languages than Java.
24492 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24493 * lto-opts.c (lto_write_options): Remove reference to Java.
24494 * opts.c (strip_off_ending): Update file extension handling.
24495 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24496 * tree-eh.c (lower_resx): Likewise.
24497 * tree.c (free_lang_data_in_type): Remove dead code.
24498 (find_decls_types_r): Likewise.
24499 (build_common_builtin_nodes): Remove Java from a comment.
24500 (verify_type): Remove dead code.
24501 * varasm.c (assemble_external): Remove Java from a comment.
24503 2017-07-14 Martin Liska <mliska@suse.cz>
24505 * opts.c (finish_options): Add quotes.
24506 (common_handle_option): Likewise.
24508 2017-07-14 Martin Liska <mliska@suse.cz>
24510 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24511 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24512 Remove N_SO_PASCAL.
24513 * dwarf2out.c (lower_bound_default): Do not handle
24515 (gen_compile_unit_die): Likewise.
24516 * gcc.c: Remove default extension binding for GNU Pascal.
24517 * stmt.c: Remove Pascal language from a comment.
24518 * xcoffout.c: Likewise.
24520 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24523 * diagnostic-show-locus.c (fixit_cmp): New function.
24524 (layout::layout): Sort m_fixit_hints.
24525 (column_range::column_range): Assert that the values are valid.
24526 (struct char_span): New struct.
24527 (correction::overwrite): New method.
24528 (struct source_line): New struct.
24529 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24530 calls in terms of classes source_line and char_span, and
24531 correction::overwrite.
24532 (selftest::test_overlapped_fixit_printing_2): New function.
24533 (selftest::diagnostic_show_locus_c_tests): Call it.
24535 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24537 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24538 early if there is no lhs.
24540 2017-07-13 Martin Liska <mliska@suse.cz>
24542 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24543 (gen_reference_type_die): Likewise.
24544 * stor-layout.c: Remove Pascal-related comment.
24546 2017-07-13 Martin Liska <mliska@suse.cz>
24548 * opts.c (finish_options): Add quotes to error messages.
24549 (parse_sanitizer_options): Likewise.
24551 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24553 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24555 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24557 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24559 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24561 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24563 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24565 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24568 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24569 provides the hardware capability bits, define the macro
24570 __BUILTIN_CPU_SUPPORTS__.
24571 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24572 if GLIBC does not provide the hardware capability bits. Add a
24573 gcc_unreachable call if the built-in cpu function is neither
24574 __builtin_cpu_is nor __builtin_cpu_supports.
24575 (rs6000_get_function_versions_dispatcher): Change the warning
24576 that an old GLIBC is used which does not export the capability
24577 bits to be an error.
24578 * doc/extend.texi (target_clones attribute): Document the
24579 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24580 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24581 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24582 the macros defined by GCC if the newer GLIBC is available.
24584 2017-07-12 Jeff Law <law@redhat.com>
24586 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24587 remaining includes slightly.
24588 * config/riscv/riscv-builtins.c: Include profile-count.h.
24590 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24593 * config/avr/avr.c (avr_set_current_function): In diagnostic
24594 messages: Quote keywords and (parts of) identifiers.
24595 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24598 2017-07-12 Carl Love <cel@us.ibm.com>
24600 * config/rs6000/rs6000-c.c: Add support for built-in functions
24601 vector bool char vec_revb (vector bool char);
24602 vector bool short vec_revb (vector short char);
24603 vector bool int vec_revb (vector bool int);
24604 vector bool long long vec_revb (vector bool long long);
24605 * doc/extend.texi: Update the built-in documentation file for the
24606 new built-in functions.
24608 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24610 * config/s390/s390.md: Remove movcc splitter.
24612 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24614 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24615 load/store on condition.
24617 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24620 * config/avr/avr.c (avr_encode_section_info)
24621 [progmem && !TREE_READONLY]: Error if progmem object needs
24624 2017-07-11 Michael Collison <michael.collison@arm.com>
24626 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24629 2017-07-11 Carl Love <cel@us.ibm.com>
24631 * config/rs6000/rs6000-c.c: Add support for builtins
24632 vector unsigned int vec_parity_lsbb (vector signed int);
24633 vector unsigned int vec_parity_lsbb (vector unsigned int);
24634 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24635 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24636 vector unsigned long long vec_parity_lsbb (vector signed long long);
24637 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24638 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24639 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24640 * doc/extend.texi: Update the built-in documentation file for the
24641 new built-in functions.
24643 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24645 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24646 (layout::m_primary_loc): New field.
24647 (layout::layout): Initialize new field. Move location filtering
24648 logic from here to...
24649 (layout::maybe_add_location_range): ...this new method. Add
24650 support for filtering to just the lines already specified by other
24652 (layout::will_show_line_p): New method.
24653 (gcc_rich_location::add_location_if_nearby): New method.
24654 (selftest::test_add_location_if_nearby): New test function.
24655 (selftest::diagnostic_show_locus_c_tests): Call it.
24656 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24659 2017-07-11 Tom de Vries <tom@codesourcery.com>
24661 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24662 (bb_first_real_insn): New function.
24663 (nvptx_single): Add extra initialization of broadcasted condition
24666 2017-07-11 Nathan Sidwell <nathan@acm.org>
24668 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24670 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24672 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24673 Binutils doc as TEXI will mess them up.
24674 * doc/invoke.texi (AVR Options): Same here.
24676 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24678 * config/sparc/sparc.opt (mfix-ut700): New option.
24679 (mfix-gr712rc): Likewise.
24680 (sparc_fix_b2bst): New variable.
24681 * doc/invoke.texi (SPARC options): Document them.
24682 (ARM options): Fix warnings.
24683 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24684 instructions to prevent sequences that can trigger the store-store
24685 errata for certain LEON3FT processors.
24686 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24687 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24688 * config/sparc/sparc.md (fix_b2bst): New attribute.
24689 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24691 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24694 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24696 (*rsqrtsf2_sse): Ditto.
24698 (div<mode>3): Macroize insn from divdf3 and divsf3
24699 using MODEF mode iterator.
24701 2017-07-10 Martin Sebor <msebor@redhat.com>
24703 PR tree-optimization/80397
24704 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24705 instead of testing for equality to INTEGER_TYPE.
24707 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24709 * config.gcc: Remove uclibc from arc target spec.
24711 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24713 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24715 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24718 * lto-wrapper.c (remove_option): New function.
24719 (merge_and_complain): Merge PIC/PIE options more realistically.
24721 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24723 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24727 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24728 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24729 * config.in: Regenerate.
24730 * configure: Regenerate.
24731 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24732 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24733 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24734 (TARGET_GASISR_PROLOGUES): ...target mask.
24735 * common/config/avr/avr-common.c
24736 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24737 Set -mgas-isr-prologues.
24738 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24739 INSERT_PASS_BEFORE for it.
24740 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24741 * config/avr/avr.c (avr_option_override)
24742 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24743 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24744 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24745 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24746 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24748 (make_avr_pass_pre_proep): New function.
24749 (emit_push_sfr) <treg>: Add argument to function and use it
24750 instead of TMP_REG.
24751 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24752 and set machine->gasisr.yes.
24753 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24754 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24755 __gcc_isr.n_pushed to .L__stack_usage.
24756 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24757 (avr_asm_final_postscan_insn): ...this new static function.
24758 * config/avr/avr.h (machine_function)
24759 <is_no_gccisr, use_L__stack_usage>: New fields.
24760 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24761 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24762 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24763 (gasisr, *gasisr): New expander and insn.
24764 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24765 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24766 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24768 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24770 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24773 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24775 Move jump-tables out of .text again.
24778 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24779 (ASM_OUTPUT_ADDR_VEC): New function.
24780 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24781 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24782 INSN_ADDRESSes as asm comment.
24783 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24784 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24785 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24786 * config/avr/avr.md (*tablejump): Adjust comment.
24787 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24788 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24790 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24791 (avr_output_addr_vec): New proto.
24792 (avr_log_t) <insn_addresses>: New field.
24794 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24797 * config/i386/i386.c (ix86_function_arg_advance): Set
24798 outgoing_args_on_stack to true if there are outgoing arguments
24800 (ix86_function_arg): Likewise.
24801 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24802 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24803 * config/i386/i386.h (machine_function): Add
24804 outgoing_args_on_stack.
24806 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24808 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24809 supporting pthreds.
24810 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24812 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24814 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24815 (REAL_H): Remove $(MACHMODE_H).
24816 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24818 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24819 $(MACHMODE_H) and double-int.h.
24820 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24822 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24825 2017-07-07 Andrew Pinski <apinski@cavium.com>
24827 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24828 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24830 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24832 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24833 Add warning if GCC was not configured to link against a GLIBC that
24834 exports the hardware capability bits.
24835 (make_resolver_func): Make resolver function private and not a
24836 COMDAT function. Create the name with clone_function_name instead
24837 of make_unique_name.
24840 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24841 correct operand in doing the split.
24843 2017-07-07 Carl Love <cel@us.ibm.com>
24845 * config/rs6000/rs6000-c: Add support for built-in function
24846 vector unsigned short vec_pack_to_short_fp32 (vector float,
24848 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24849 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24850 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24851 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24852 (convert_4f32_8i16): Add define_expand.
24853 * doc/extend.texi: Update the built-in documentation file for the
24854 new built-in function.
24856 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24858 * config/sparc/m8.md: New file.
24859 * config/sparc/sparc.md: Include m8.md.
24861 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24863 * config/sparc/sparc.opt: New option -mvis4b.
24864 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24865 (sparc_option_override): Handle VIS4B.
24866 (enum sparc_builtins): Define
24867 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24868 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24869 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24870 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24871 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24872 (check_constant_argument): New function.
24873 (sparc_vis_init_builtins): Define builtins
24874 __builtin_vis_dictunpack{8,16,32},
24875 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24876 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24877 __builtin_vis_fpcmpde{8,16,32}shl and
24878 __builtin_vis_fpcmpur{8,16,32}shl.
24879 (sparc_expand_builtin): Check that the constant operands to
24880 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24881 constant and in range.
24882 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24884 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24885 (SPARC_IMM5_P): Likewise.
24886 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24887 (enabled): Handle vis4b.
24888 (UNSPEC_DICTUNPACK): New unspec.
24889 (UNSPEC_FPCMPSHL): Likewise.
24890 (UNSPEC_FPUCMPSHL): Likewise.
24891 (UNSPEC_FPCMPDESHL): Likewise.
24892 (UNSPEC_FPCMPURSHL): Likewise.
24893 (cpu_feature): New CPU feature `vis4b'.
24894 (dictunpack{8,16,32}): New insns.
24895 (FPCSMODE): New mode iterator.
24896 (fpcscond): New code iterator.
24897 (fpcsucond): Likewise.
24898 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24899 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24900 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24901 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24902 * config/sparc/constraints.md: Define constraints `q' for unsigned
24903 2-bit integer constants and `t' for unsigned 5-bit integer
24905 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24907 (imm5_operand_dictunpack16): Likewise.
24908 (imm5_operand_dictunpack32): Likewise.
24909 (imm2_operand): Likewise.
24910 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24911 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24912 ditunpack* and fpcmp*shl builtins.
24914 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24916 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24917 * config.in: Add HAVE_AS_SPARC6 define.
24918 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24920 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24922 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24923 (CPP_CPU_SPEC): Handle m8.
24924 (ASM_CPU_SPEC): Likewise.
24925 * config/sparc/sparc-opts.h (enum processor_type): Add
24927 * config/sparc/sparc.c (m8_costs): New struct.
24928 (sparc_option_override): Handle TARGET_CPU_m8.
24929 (sparc32_initialize_trampoline): Likewise.
24930 (sparc64_initialize_trampoline): Likewise.
24931 (sparc_issue_rate): Likewise.
24932 (sparc_register_move_cost): Likewise.
24933 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24934 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24935 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24936 (CPP_CPU_SPEC): Handle M8.
24937 (ASM_CPU_SPEC): Likewise.
24938 (AS_M8_FLAG): Define.
24939 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24940 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24941 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24943 * configure: Regenerate.
24944 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24949 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24951 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24952 ("*movdi_insn_sp32"): Do not set v3pipe.
24953 ("*movsi_insn"): Likewise.
24954 ("*movdi_insn_sp64"): Likewise.
24955 ("*movsf_insn"): Likewise.
24956 ("*movdf_insn_sp32"): Likewise.
24957 ("*movdf_insn_sp64"): Likewise.
24958 ("*zero_extendsidi2_insn_sp64"): Likewise.
24959 ("*sign_extendsidi2_insn"): Likewise.
24960 ("*mov<VM32:mode>_insn"): Likewise.
24961 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24962 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24963 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24964 ("<vlop:code><VL:mode>3"): Likewise.
24965 ("*not_<vlop:code><VL:mode>3"): Likewise.
24966 ("*nand<VL:mode>_vis"): Likewise.
24967 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24968 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24969 ("one_cmpl<VL:mode>2"): Likewise.
24970 ("faligndata<VM64:mode>_vis"): Likewise.
24971 ("alignaddrsi_vis"): Likewise.
24972 ("alignaddrdi_vis"): Likweise.
24973 ("alignaddrlsi_vis"): Likewise.
24974 ("alignaddrldi_vis"): Likewise.
24975 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24976 ("bmaskdi_vis"): Likewise.
24977 ("bmasksi_vis"): Likewise.
24978 ("bshuffle<VM64:mode>_vis"): Likewise.
24979 ("cmask8<P:mode>_vis"): Likewise.
24980 ("cmask16<P:mode>_vis"): Likewise.
24981 ("cmask32<P:mode>_vis"): Likewise.
24982 ("pdistn<P:mode>_vis"): Likewise.
24983 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24985 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24987 * config/sparc/sparc.md ("subtype"): New insn attribute.
24988 ("*wrgsr_sp64"): Set insn subtype.
24989 ("*rdgsr_sp64"): Likewise.
24990 ("alignaddrsi_vis"): Likewise.
24991 ("alignaddrdi_vis"): Likewise.
24992 ("alignaddrlsi_vis"): Likewise.
24993 ("alignaddrldi_vis"): Likewise.
24994 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24995 ("fexpand_vis"): Likewise.
24996 ("fpmerge_vis"): Likewise.
24997 ("faligndata<VM64:mode>_vis"): Likewise.
24998 ("bshuffle<VM64:mode>_vis"): Likewise.
24999 ("cmask8<P:mode>_vis"): Likewise.
25000 ("cmask16<P:mode>_vis"): Likewise.
25001 ("cmask32<P:mode>_vis"): Likewise.
25002 ("fchksm16_vis"): Likewise.
25003 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25004 ("fmean16_vis"): Likewise.
25005 ("fp<plusminus_insn>64_vis"): Likewise.
25006 ("<plusminus_insn>v8qi3"): Likewise.
25007 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25008 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25009 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25010 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25011 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25012 ("*movqi_insn"): Likewise.
25013 ("*movhi_insn"): Likewise.
25014 ("*movsi_insn"): Likewise.
25015 ("movsi_pic_gotdata_op"): Likewise.
25016 ("*movdi_insn_sp32"): Likewise.
25017 ("*movdi_insn_sp64"): Likewise.
25018 ("movdi_pic_gotdata_op"): Likewise.
25019 ("*movsf_insn"): Likewise.
25020 ("*movdf_insn_sp32"): Likewise.
25021 ("*movdf_insn_sp64"): Likewise.
25022 ("*zero_extendhisi2_insn"): Likewise.
25023 ("*zero_extendqihi2_insn"): Likewise.
25024 ("*zero_extendqisi2_insn"): Likewise.
25025 ("*zero_extendqidi2_insn"): Likewise.
25026 ("*zero_extendhidi2_insn"): Likewise.
25027 ("*zero_extendsidi2_insn_sp64"): Likewise.
25028 ("ldfsr"): Likewise.
25029 ("prefetch_64"): Likewise.
25030 ("prefetch_32"): Likewise.
25031 ("tie_ld32"): Likewise.
25032 ("tie_ld64"): Likewise.
25033 ("*tldo_ldub_sp32"): Likewise.
25034 ("*tldo_ldub1_sp32"): Likewise.
25035 ("*tldo_ldub2_sp32"): Likewise.
25036 ("*tldo_ldub_sp64"): Likewise.
25037 ("*tldo_ldub1_sp64"): Likewise.
25038 ("*tldo_ldub2_sp64"): Likewise.
25039 ("*tldo_ldub3_sp64"): Likewise.
25040 ("*tldo_lduh_sp32"): Likewise.
25041 ("*tldo_lduh1_sp32"): Likewise.
25042 ("*tldo_lduh_sp64"): Likewise.
25043 ("*tldo_lduh1_sp64"): Likewise.
25044 ("*tldo_lduh2_sp64"): Likewise.
25045 ("*tldo_lduw_sp32"): Likewise.
25046 ("*tldo_lduw_sp64"): Likewise.
25047 ("*tldo_lduw1_sp64"): Likewise.
25048 ("*tldo_ldx_sp64"): Likewise.
25049 ("*mov<VM32:mode>_insn"): Likewise.
25050 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25051 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25053 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25055 * config/sparc/sparc.md ("type"): New insn type viscmp.
25056 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25058 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25059 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25060 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25061 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25063 ("n7_vis_logical_11cycle"): Likewise.
25064 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25065 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25066 * config/sparc/niagara.md ("niag_vis"): Likewise.
25067 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25068 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25070 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25072 * config/sparc/sparc.md: New instruction type `bmask'.
25073 (bmaskdi_vis): Use the `bmask' type.
25074 (bmasksi_vis): Likewise.
25075 * config/sparc/ultra3.md (us3_array): Likewise.
25076 * config/sparc/niagara7.md (n7_array): Likewise.
25077 * config/sparc/niagara4.md (n4_array): Likewise.
25078 * config/sparc/niagara2.md (niag2_vis): Likewise.
25079 (niag3_vis): Likewise.
25080 * config/sparc/niagara.md (niag_vis): Likewise.
25082 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25084 * ipa-comdats.c: Remove optimize check from gate.
25085 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25086 for functions not optimized.
25087 (ipa_fn_summary_read): Skip optimize check.
25088 (ipa_fn_summary_write): Likewise.
25089 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25091 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25093 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25094 (check_callers): Give up on caller which is not optimized.
25095 (inline_small_functions): Likewise.
25096 (ipa_inline): Do not give up when not optimizing.
25097 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25098 away unoptimizes cdtors.
25099 (whole_program_function_and_variable_visibility): Do
25100 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25101 * ipa.c (process_references): Do not check optimize.
25102 (symbol_table::remove_unreachable_nodes): Update optimize check.
25103 (set_writeonly_bit): Update optimize check.
25104 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25105 (pass_ipa_single_use::gate): Remove.
25107 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25109 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25110 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25111 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25112 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25113 permute_load, permute_store, adjust_extract, adjust_splat,
25114 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25115 replace_swap_with_copy, dump_swap_insn_table,
25116 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25117 recombine_lvx_pattern, recombine_stvx_pattern,
25118 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25119 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25120 to file rs6000-p8swap.c.
25121 * config/rs6000/rs6000-p8swap.c: New file.
25122 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25123 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25124 and rs6000*-*-* targets.
25126 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25128 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25130 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25132 * lto-wrapper.c (merge_and_complain): Do not merge
25133 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25134 fsigned_zeros, ftrapping_math, fwrapv.
25135 (append_compiler_options): Do not track these options.
25136 (append_linker_options): Likewie
25138 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25140 * cgraphunit.c (cgraph_node::finalize_function): When
25141 !flag_toplevel_reorde set no_reorder flag.
25142 (varpool_node::finalize_decl): Likewise.
25143 (symbol_table::compile): Drop no toplevel reorder path.
25145 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25147 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25148 edges; zero probability is not better than uninitialized.
25150 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25152 * asan.h (asan_sanitize_allocas_p): Declare.
25153 * asan.c (asan_sanitize_allocas_p): New function.
25154 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25155 (handle_builtin_alloca): Likewise.
25156 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25157 if !asan_sanitize_allocas_p.
25158 * params.def (asan-instrument-allocas): Add new option.
25159 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25160 * opts.c (common_handle_option): Disable allocas sanitization for
25163 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25165 * asan.c: Include gimple-fold.h.
25166 (get_last_alloca_addr): New function.
25167 (handle_builtin_stackrestore): Likewise.
25168 (handle_builtin_alloca): Likewise.
25169 (asan_emit_allocas_unpoison): Likewise.
25170 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25171 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25172 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25173 (instrument_builtin_call): Pass gimple iterator to
25174 get_mem_refs_of_builtin_call.
25175 (last_alloca_addr): New global.
25176 * asan.h (asan_emit_allocas_unpoison): Declare.
25177 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25178 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25179 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25180 if function calls alloca.
25181 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25182 * gimple-fold.h (replace_call_with_value): Declare.
25183 * internal-fn.c: Include asan.h.
25184 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25185 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25187 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25189 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25190 (C_SELFTEST_FLAGS): New.
25191 (CPP_SELFTEST_FLAGS): New.
25192 (SELFTEST_DEPS): New, from deps of s-selftest.
25193 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25194 (CPP_SELFTEST_DEPS): New.
25195 (selftest): Add dependency on s-selftest-c++.
25196 (s-selftest): Rename to...
25197 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25198 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25199 than SELFTEST_FLAGS.
25200 (selftest-gdb): Rename to...
25201 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25203 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25204 (selftest-valgrind): Rename to...
25205 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25207 (selftest-valgrind): Reintroduce as an alias for
25208 selftest-c-valgrind.
25209 (s-selftest-c++): New.
25210 (selftest-c++-gdb): New.
25211 (selftest-c++-valgrind): New.
25213 2017-07-06 Olivier Hainque <hainque@adacore.com>
25215 * gcc.c (process_command): When deciding if undefined variables
25216 should be ignored when processing specs, accept "gcc -v" as well.
25218 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25220 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25221 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25223 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25225 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25226 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25227 * config/arm/arm-tables.opt: Regenerate.
25228 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25230 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25232 2017-07-06 Carl Love <cel@us.ibm.com>
25234 * ChangeLog: Clean up from mid air collision
25236 2017-07-06 Carl Love <cel@us.ibm.com>
25238 * config/rs6000/rs6000-c.c: Add support for built-in functions
25239 vector signed int vec_subc (vector signed int, vector signed int);
25240 vector signed __int128 vec_subc (vector signed __int128,
25241 vector signed __int128);
25242 vector unsigned __int128 vec_subc (vector unsigned __int128,
25243 vector unsigned __int128);
25244 vector signed int vec_sube (vector signed int, vector signed int,
25245 vector signed int);
25246 vector unsigned int vec_sube (vector unsigned int,
25247 vector unsigned int,
25248 vector unsigned int);
25249 vector signed __int128 vec_sube (vector signed __int128,
25250 vector signed __int128,
25251 vector signed__int128);
25252 vector unsigned __int128 vec_sube (vector unsigned __int128,
25253 vector unsigned __int128,
25254 vector unsigned __int128);
25255 vector signed int vec_subec (vector signed int, vector signed int,
25256 vector signed int);
25257 vector unsigned int vec_subec (vector unsigned int,
25258 vector unsigned int,
25259 vector unsigned int);
25260 vector signed __int128 vec_subec (vector signed __int128,
25261 vector signed __int128,
25262 vector signed__int128);
25263 vector unsigned __int128 vec_subec (vector unsigned __int128,
25264 vector unsigned __int128,
25265 vector unsigned __int128);
25266 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25267 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25268 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25269 BU_ALTIVEC_OVERLOAD_X definitions.
25270 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25271 * doc/extend.texi: Update the built-in documentation file for the new
25272 built-in functions.
25274 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25277 * diagnostic-show-locus.c (layout::layout): Use start and finish
25278 spelling location for the start and finish of each range.
25279 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25280 Add unused aspect param.
25281 * input.c (expand_location_1): Add "aspect" param, and use it
25282 to access the correct part of the location.
25283 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25285 (expand_location_to_spelling_point): Likewise.
25286 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25287 param, and pass it to expand_location_1.
25289 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25291 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25292 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25293 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25294 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25295 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25296 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25297 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25298 _mm_maskz_getmant_ss): New intrinsics.
25299 (__builtin_ia32_getexpss128_mask): Changed to ...
25300 __builtin_ia32_getexpss128_round ... this.
25301 (__builtin_ia32_getexpsd128_mask): Changed to ...
25302 __builtin_ia32_getexpsd128_round ... this.
25303 * config/i386/i386-builtin-types.def
25304 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25305 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25306 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25307 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25308 __builtin_ia32_getmantss_mask_round): New builtins.
25309 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25310 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25311 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25312 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25313 * config/i386/sse.md
25314 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25315 avx512f_sgetexp<mode><mask_scalar_name>
25316 <round_saeonly_scalar_name> ... this.
25317 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25318 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25319 vgetexp<ssescalarmodesuffix>
25320 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25321 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25322 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25323 avx512f_vgetmant<mode><mask_scalar_name>
25324 <round_saeonly_scalar_name> ... this.
25325 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25326 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25327 vgetmant<ssescalarmodesuffix>
25328 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25329 %0<mask_scalar_operand4>, %1, %2
25330 <round_saeonly_scalar_mask_op4>, %3} ... this.
25331 * config/i386/subst.md (mask_scalar_operand4,
25332 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25333 round_saeonly_scalar_nimm_predicate): New subst attributes.
25335 2017-07-06 Julia Koval <julia.koval@intel.com>
25337 * config/i386/i386.c (ix86_erase_embedded_rounding):
25338 Remove code for old rounding pattern.
25340 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25342 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25344 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25346 * doc/sourcebuild.texi (Test Directives, Variants of
25347 dg-require-support): Add documentation for dg-require-stack-check.
25349 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25351 * config/i386/subst.md (mask_scalar, round_scalar,
25352 round_saeonly_scalar): New meta-templates.
25353 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25354 round_scalar_mask_operand3, round_scalar_mask_op3,
25355 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25356 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25357 round_saeonly_scalar_constraint,
25358 round_saeonly_scalar_prefix): New subst attribute.
25359 * config/i386/sse.md
25360 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25361 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25362 <round_scalar_name> ... this.
25363 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25364 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25365 <round_scalar_name> ... this.
25366 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25367 <sse>_vm<code><mode>3<mask_scalar_name>
25368 <round_saeonly_scalar_name> ... this.
25369 (v<plusminus_mnemonic><ssescalarmodesuffix>
25370 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25371 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25372 v<plusminus_mnemonic><ssescalarmodesuffix>
25373 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25374 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25375 (v<multdiv_mnemonic><ssescalarmodesuffix>
25376 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25377 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25378 v<multdiv_mnemonic><ssescalarmodesuffix>
25379 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25380 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25381 (v<maxmin_float><ssescalarmodesuffix>
25382 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25383 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25384 v<maxmin_float><ssescalarmodesuffix>
25385 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25386 %0<mask_scalar_operand3>, %1, %<iptr>2
25387 <round_saeonly_scalar_mask_op3>} ... this.
25389 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25391 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25392 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25394 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25395 Alan Hayward <alan.hayward@arm.com>
25396 David Sherwood <david.sherwood@arm.com>
25398 * combine.c (simplify_if_then_else): Remove "enum" before
25400 * compare-elim.c (can_eliminate_compare): Likewise.
25401 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25403 (aarch64_lookup_simd_builtin_type): Likewise.
25404 (aarch64_simd_builtin_type): Likewise.
25405 (aarch64_init_simd_builtin_types): Likewise.
25406 (aarch64_simd_expand_args): Likewise.
25407 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25409 (aarch64_reverse_mask): Likewise.
25410 (aarch64_simd_emit_reg_reg_move): Likewise.
25411 (aarch64_gen_adjusted_ldpstp): Likewise.
25412 (aarch64_ccmp_mode_to_code): Likewise.
25413 (aarch64_operands_ok_for_ldpstp): Likewise.
25414 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25415 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25417 (aarch64_min_divisions_for_recip_mul): Likewise.
25418 (aarch64_reassociation_width): Likewise.
25419 (aarch64_get_condition_code_1): Likewise.
25420 (aarch64_simd_emit_reg_reg_move): Likewise.
25421 (aarch64_simd_attr_length_rglist): Likewise.
25422 (aarch64_reverse_mask): Likewise.
25423 (aarch64_operands_ok_for_ldpstp): Likewise.
25424 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25425 (aarch64_gen_adjusted_ldpstp): Likewise.
25426 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25428 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25429 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25430 (arm_lookup_simd_builtin_type): Likewise.
25431 (arm_simd_builtin_type): Likewise.
25432 (arm_init_simd_builtin_types): Likewise.
25433 (arm_expand_builtin_args): Likewise.
25434 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25435 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25436 (ft32_setup_incoming_varargs): Likewise.
25437 (ft32_function_arg): Likewise.
25438 (ft32_function_arg_advance): Likewise.
25439 (ft32_pass_by_reference): Likewise.
25440 (ft32_arg_partial_bytes): Likewise.
25441 (ft32_valid_pointer_mode): Likewise.
25442 (ft32_addr_space_pointer_mode): Likewise.
25443 (ft32_addr_space_legitimate_address_p): Likewise.
25444 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25446 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25447 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25448 (iamcu_alignment): Likewise.
25449 (canonicalize_vector_int_perm): Likewise.
25450 (ix86_noce_conversion_profitable_p): Likewise.
25451 (ix86_mpx_bound_mode): Likewise.
25452 (ix86_operands_ok_for_move_multiple): Likewise.
25453 * config/microblaze/microblaze-protos.h
25454 (microblaze_expand_conditional_branch_reg): Likewise.
25455 * config/microblaze/microblaze.c
25456 (microblaze_expand_conditional_branch_reg): Likewise.
25457 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25459 (rs6000_reassociation_width): Likewise.
25460 (rs6000_invalid_binary_op): Likewise.
25461 (fusion_p9_p): Likewise.
25462 (emit_fusion_p9_load): Likewise.
25463 (emit_fusion_p9_store): Likewise.
25464 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25466 (riscv_hard_regno_mode_ok_p): Likewise.
25467 (riscv_address_insns): Likewise.
25468 (riscv_split_symbol): Likewise.
25469 (riscv_legitimize_move): Likewise.
25470 (riscv_function_value): Likewise.
25471 (riscv_hard_regno_nregs): Likewise.
25472 (riscv_expand_builtin): Likewise.
25473 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25474 (riscv_build_integer): Likewise.
25475 (riscv_split_integer): Likewise.
25476 (riscv_legitimate_constant_p): Likewise.
25477 (riscv_cannot_force_const_mem): Likewise.
25478 (riscv_regno_mode_ok_for_base_p): Likewise.
25479 (riscv_valid_base_register_p): Likewise.
25480 (riscv_valid_offset_p): Likewise.
25481 (riscv_valid_lo_sum_p): Likewise.
25482 (riscv_classify_address): Likewise.
25483 (riscv_legitimate_address_p): Likewise.
25484 (riscv_address_insns): Likewise.
25485 (riscv_load_store_insns): Likewise.
25486 (riscv_force_binary): Likewise.
25487 (riscv_split_symbol): Likewise.
25488 (riscv_force_address): Likewise.
25489 (riscv_legitimize_address): Likewise.
25490 (riscv_move_integer): Likewise.
25491 (riscv_legitimize_const_move): Likewise.
25492 (riscv_legitimize_move): Likewise.
25493 (riscv_address_cost): Likewise.
25494 (riscv_subword): Likewise.
25495 (riscv_output_move): Likewise.
25496 (riscv_canonicalize_int_order_test): Likewise.
25497 (riscv_emit_int_order_test): Likewise.
25498 (riscv_function_arg_boundary): Likewise.
25499 (riscv_pass_mode_in_fpr_p): Likewise.
25500 (riscv_pass_fpr_single): Likewise.
25501 (riscv_pass_fpr_pair): Likewise.
25502 (riscv_get_arg_info): Likewise.
25503 (riscv_function_arg): Likewise.
25504 (riscv_function_arg_advance): Likewise.
25505 (riscv_arg_partial_bytes): Likewise.
25506 (riscv_function_value): Likewise.
25507 (riscv_pass_by_reference): Likewise.
25508 (riscv_setup_incoming_varargs): Likewise.
25509 (riscv_print_operand): Likewise.
25510 (riscv_elf_select_rtx_section): Likewise.
25511 (riscv_save_restore_reg): Likewise.
25512 (riscv_for_each_saved_reg): Likewise.
25513 (riscv_register_move_cost): Likewise.
25514 (riscv_hard_regno_mode_ok_p): Likewise.
25515 (riscv_hard_regno_nregs): Likewise.
25516 (riscv_class_max_nregs): Likewise.
25517 (riscv_memory_move_cost): Likewise.
25518 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25519 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25520 (rl78_addr_space_address_mode): Likewise.
25521 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25523 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25524 (rs6000_reassociation_width): Likewise.
25525 (rs6000_invalid_binary_op): Likewise.
25526 (fusion_p9_p): Likewise.
25527 (emit_fusion_p9_load): Likewise.
25528 (emit_fusion_p9_store): Likewise.
25529 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25530 (ok_for_simple_move_operands): Likewise.
25531 (ok_for_simple_move_strict_operands): Likewise.
25532 (ok_for_simple_arith_logic_operands): Likewise.
25533 (visium_legitimize_reload_address): Likewise.
25534 (visium_select_cc_mode): Likewise.
25535 (output_cbranch): Likewise.
25536 (visium_split_double_move): Likewise.
25537 (visium_expand_copysign): Likewise.
25538 (visium_expand_int_cstore): Likewise.
25539 (visium_expand_fp_cstore): Likewise.
25540 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25541 (visium_function_arg): Likewise.
25542 (visium_function_arg_advance): Likewise.
25543 (visium_libcall_value): Likewise.
25544 (visium_setup_incoming_varargs): Likewise.
25545 (visium_legitimate_constant_p): Likewise.
25546 (visium_legitimate_address_p): Likewise.
25547 (visium_legitimize_address): Likewise.
25548 (visium_secondary_reload): Likewise.
25549 (visium_register_move_cost): Likewise.
25550 (visium_memory_move_cost): Likewise.
25551 (prepare_move_operands): Likewise.
25552 (ok_for_simple_move_operands): Likewise.
25553 (ok_for_simple_move_strict_operands): Likewise.
25554 (ok_for_simple_arith_logic_operands): Likewise.
25555 (visium_function_value_1): Likewise.
25556 (rtx_ok_for_offset_p): Likewise.
25557 (visium_legitimize_reload_address): Likewise.
25558 (visium_split_double_move): Likewise.
25559 (visium_expand_copysign): Likewise.
25560 (visium_expand_int_cstore): Likewise.
25561 (visium_expand_fp_cstore): Likewise.
25562 (visium_split_cstore): Likewise.
25563 (visium_select_cc_mode): Likewise.
25564 (visium_split_cbranch): Likewise.
25565 (output_cbranch): Likewise.
25566 (visium_print_operand_address): Likewise.
25567 * expmed.c (flip_storage_order): Likewise.
25568 * expmed.h (emit_cstore): Likewise.
25569 (flip_storage_order): Likewise.
25570 * genrecog.c (validate_pattern): Likewise.
25571 * hsa-gen.c (gen_hsa_addr): Likewise.
25572 * internal-fn.c (expand_arith_overflow): Likewise.
25573 * ira-color.c (allocno_copy_cost_saving): Likewise.
25574 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25575 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25576 (process_invariant_for_inheritance): Likewise.
25577 * lra-eliminations.c (move_plus_up): Likewise.
25578 * omp-low.c (lower_oacc_reductions): Likewise.
25579 * simplify-rtx.c (simplify_subreg): Likewise.
25580 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25581 (TARGET_CHKP_BOUND_MODE): Likewise..
25582 * targhooks.c (default_chkp_bound_mode): Likewise.
25583 (default_setup_incoming_vararg_bounds): Likewise.
25584 * targhooks.h (default_chkp_bound_mode): Likewise.
25585 (default_setup_incoming_vararg_bounds): Likewise.
25586 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25587 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25588 (have_whole_vector_shift): Likewise.
25589 * tree-vect-stmts.c (vectorizable_load): Likewise.
25590 * doc/tm.texi: Regenerate.
25592 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25594 Graceful degrade if Binutils PR21472 is not available.
25597 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25598 .rodata in flash test fails.
25599 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25600 * confgure: Regenerate.
25601 * config.in: Regenerate.
25602 * config/avr/avr.c (avr_asm_named_section)
25603 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25604 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25605 (avr_asm_init_sections): Same.
25607 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25609 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25610 (fma<VH:mode>4_intrinsic): Likewise.
25611 (*fmsub<VCVTF:mode>4): Likewise.
25612 (*fmsub<VH:mode>4_intrinsic): Likewise.
25614 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25617 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25618 Don't depend on "optimize > 0".
25619 (out_movhi_r_mr, out_movqi_mr_r): Same.
25620 (out_movhi_mr_r, out_movqi_r_mr): Same.
25621 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25622 io_address_operand on "optimize > 0".
25624 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25626 * tree-loop-distribution.c: Add general explanantion on the pass.
25627 (generate_loops_for_partition): Mark distributed loop.
25628 (pg_add_dependence_edges): New parameter. Handle alias data
25629 dependence specially and record it in the parameter if asked.
25630 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25631 (init_partition_graph_vertices, add_partition_graph_edge): New.
25632 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25633 (free_partition_graph_vdata, build_partition_graph): New.
25634 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25635 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25636 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25637 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25638 (version_for_distribution_p, finalize_partitions): New.
25639 (distribute_loop): Handle alias data dependence specially. Factor
25640 out loop fusion code as functions and call these functions.
25642 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25644 * tree-loop-distribution.c (classify_partition): New parameter and
25645 better handle reduction statement.
25646 (rdg_build_partitions): Revise comment.
25647 (distribute_loop): Compute statements in all partitions and pass it
25648 to classify_partition.
25650 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25652 * tree-loop-distribution.c (enum partition_type): New.
25653 (struct partition): New field type.
25654 (partition_merge_into): Add parameter. Update partition type.
25655 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25656 (build_rdg_partition_for_vertex): Compute partition type.
25657 (rdg_build_partitions): Dump partition type.
25658 (distribute_loop): Update calls to partition_merge_into.
25660 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25662 * tree-loop-distribution.c (struct ddr_hasher): New.
25663 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25665 (classify_partition): Call get_data_dependence.
25666 (pg_add_dependence_edges): Ditto.
25667 (distribute_loop): Release data dependence hash table.
25669 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25671 * tree-loop-distribution.c (ref_base_address): Delete.
25672 (similar_memory_accesses): Rename ...
25673 (share_memory_accesses): ... to this. Check if partitions access
25674 the same memory reference.
25675 (distribute_loop): Call share_memory_accesses.
25677 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25679 * tree-loop-distribution.c (struct partition): New field recording
25680 its data reference.
25681 (partition_alloc, partition_free): Init and release data refs.
25682 (partition_merge_into): Merge data refs.
25683 (build_rdg_partition_for_vertex): Collect data refs for partition.
25684 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25686 (distribute_loop): Remve data refs from vertice data of partition
25689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25691 * tree-loop-distribution.c (params.h): Include header file.
25692 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25693 (datarefs_vec): New global var.
25694 (create_rdg_vertices): Use datarefs_vec directly.
25695 (free_rdg): Don't free data references.
25696 (build_rdg): Update use. Don't free data references.
25697 (distribute_loop): Compute global variable for data references.
25698 Bail out if there are too many data references.
25700 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25702 * tree-loop-distribution.c (loop_nest): New global var.
25703 (build_rdg): Use loop directly, rather than loop nest.
25704 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25706 (distribute_loop): Compute global variable loop nest. Update use.
25708 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25710 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25711 (partition_merge_into): New parameter. Dump reason for fusion.
25712 (distribute_loop): Update use of partition_merge_into.
25714 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25716 * tree-loop-distribution.c (bb_top_order_index): New.
25717 (bb_top_order_index_size, bb_top_order_cmp): New.
25718 (stmts_from_loop): Use topological order.
25719 (pass_loop_distribution::execute): Compute and release topological
25720 order for basic blocks.
25722 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25724 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25729 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25730 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25731 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25732 * internal-fn.def (LOOP_DIST_ALIAS): New.
25733 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25734 (fold_loop_internal_call): ... this.
25735 (vect_loop_dist_alias_call): New function.
25736 (set_uid_loop_bbs): Call fold_loop_internal_call.
25737 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25740 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25743 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25744 Require dead FLAGS_REG at the beginning of a peephole.
25746 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25749 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25750 arguments in the call to __builtin_ia32_sbb_u32.
25751 (_subborrow_u64): Swap _X and _Y arguments in the call to
25752 __builtin_ia32_sbb_u64.
25754 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25757 * tree-vrp.c (compare_assert_loc): Turn into a function template
25758 with stable template parameter. Only test if a->e is NULL,
25759 !a->e == !b->e has been verified already. Use e == NULL or
25760 e != NULL instead of e or ! e tests. If stable is true, don't use
25761 iterative_hash_expr, on the other side allow a or b or both NULL
25762 and sort the NULLs last.
25763 (process_assert_insertions): Sort using compare_assert_loc<false>
25764 instead of compare_assert_loc, later sort using
25765 compare_assert_loc<true> before calling process_assert_insertions_for
25766 in a loop. Use break instead of continue once seen NULL pointer.
25768 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25770 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25771 Cortex-R7 and Cortex-R8 processors.
25773 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25775 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25776 uninitialized while src is not.
25778 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25780 * common/config/arm/arm-common.c: Adjust include path for
25782 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25783 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25784 (arm-cpu-data.h): Likewise.
25785 (arm-cpu-cdata.h): Likewise.
25786 * config/arm/arm-cpu.h: Delete.
25787 * config/arm/arm-cpu-cdata.h: Delete.
25788 * config/arm/arm-cpu-data.h: Delete.
25790 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25792 * config/arm/arm-cpus.in (cortex-a55): New.
25793 (cortex-a75): Likewise.
25794 (cortex-a75.cortex-a55): Likewise.
25795 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25797 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25798 * config/arm/arm-cpu-cdata.h: Regenerate.
25799 * config/arm/arm-cpu-data.h: Regenerate.
25800 * config/arm/arm-cpu.h: Regenerate.
25801 * config/arm/arm-tables.opt: Regenerate.
25802 * config/arm/arm-tune.md: Regenerate.
25804 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25806 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25808 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25810 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25813 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25815 PR tree-optimization/81292
25816 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25817 full_string_p, also call adjust_related_strinfos if the adjustment
25818 is simple, otherwise invalidate related strinfos.
25820 2017-07-04 Martin Liska <mliska@suse.cz>
25823 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25824 newly created variable as DECL_IGNORED_P.
25826 2017-07-04 Martin Liska <mliska@suse.cz>
25829 * ipa-inline.c (inline_small_functions):
25830 Use xstrdup_for_dump.
25832 2017-07-04 Tom de Vries <tom@codesourcery.com>
25834 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25836 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25839 * config/darwin.c (darwin_function_switched_text_sections):
25842 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25844 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25846 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25848 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25850 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25852 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25853 min_profitable_iters, and th as inclusive lower bounds.
25854 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25855 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25856 for min_profitable_iters and min_profitable_estimate.
25857 (vect_transform_loop): Treat th as an inclusive lower bound.
25858 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25860 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25863 * config/darwin.c (darwin_function_switched_text_sections):
25864 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25865 in two pieces, and suppress the use of buf.
25867 2017-07-03 Nathan Sidwell <nathan@acm.org>
25869 * hash-table.h (hash_table_mod1): Fix indentation.
25871 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25873 PR middle-end/81290
25874 * predict.c (force_edge_cold): Be more careful about propagation
25876 * profile-count.h (profile_probability::guessed,
25877 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25879 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25881 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25883 * doc/invoke.texi (rcpc architecture extension): Document it.
25885 2017-07-03 Richard Biener <rguenther@suse.de>
25887 PR tree-optimization/60510
25888 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25889 the scalar reduction PHI and use it.
25890 (vectorizable_reduction): Properly guard the single_defuse_cycle
25891 path for non-SLP reduction chains where we cannot use it.
25892 Rework reduc_def/index and vector type deduction. Rework
25893 vector operand gathering during reduction op code-gen.
25894 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25895 chains dissolve the chain and leave it to non-SLP reduction
25898 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25900 * tree-data-ref.h (dr_alignment): Declare.
25901 * tree-data-ref.c (dr_alignment): New function.
25902 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25903 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25905 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25907 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25909 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25910 and base_misalignment fields.
25911 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25912 * tree-data-ref.c: Include builtins.h.
25913 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25914 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25915 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25916 * tree-vect-data-refs.c: Include tree-cfg.h.
25917 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25918 fields instead of calculating an alignment here.
25919 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25920 innermost_loop_behavior fields.
25922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25924 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25926 (DR_STEP_ALIGNMENT): New macro.
25927 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25928 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25929 (create_data_ref): Print it.
25930 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25931 to tell whether the step preserves vector (mis)alignment.
25932 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25933 Move the check for an integer step and generalise to all INTEGER_CST.
25934 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25935 Print the outer step alignment.
25937 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25939 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25940 with offset_alignment.
25941 (DR_ALIGNED_TO): Delete.
25942 (DR_OFFSET_ALIGNMENT): New macro.
25943 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25944 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25945 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25946 (create_data_ref): Likewise.
25947 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25948 (vect_analyze_data_refs): Likewise.
25949 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25950 creating dummy innermost behavior.
25952 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25954 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25955 with a "innermost_loop_behavior *" and refeence tree.
25956 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25957 (create_data_ref): Update call accordingly.
25958 * tree-predcom.c (find_looparound_phi): Likewise.
25960 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25962 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25963 fields with dr_wrt_vec_loop.
25964 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25965 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25966 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25967 (vect_dr_behavior): New function.
25968 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25969 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25970 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25971 track whether the step preserves the misalignment.
25972 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25973 Use vect_dr_behavior.
25974 (vect_setup_realignment): Update call accordingly.
25975 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25976 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25977 call to vect_create_addr_base_for_vector_ref.
25978 (vect_create_cond_for_align_checks): Likewise.
25979 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25980 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25981 (vect_recog_mask_conversion_pattern): Likewise.
25982 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25983 (new_stmt_vec_info): Remove redundant zeroing.
25985 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25987 * common/config/arm/arm-common.c (arm_be8_option): New function.
25988 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25989 (ISA_ARMv6): Add isa_bit_be8.
25990 * config/arm/arm.h (arm_be8_option): Add prototype.
25991 (BE8_SPEC_FUNCTION): New define.
25992 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25993 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25994 (mlittle-endian): Similarly.
25995 (mbe8, mbe32): New options.
25996 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25997 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25999 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26001 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26003 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26005 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26006 (cleanup_tree_cfg_bb): Use it.
26007 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26009 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26011 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26014 * loop-doloop.c (add_test): Update profile.
26016 2017-07-03 Martin Liska <mliska@suse.cz>
26019 * sanopt.c (rewrite_usage_of_param): New function.
26020 (sanitize_rewrite_addressable_params): Likewise.
26021 (pass_sanopt::execute): Call rewrite_usage_of_param.
26023 2017-07-03 Richard Biener <rguenther@suse.de>
26025 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26026 back to using VIEW_CONVERT_EXPR.
26028 2017-07-03 Martin Liska <mliska@suse.cz>
26031 * doc/extend.texi: Document when a resolver function is
26032 generated for target_clones.
26034 2017-07-03 Martin Liska <mliska@suse.cz>
26036 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26037 and shadow memory of auto variables which are subject of
26038 use-after-scope sanitization.
26039 (asan_expand_mark_ifn): Add do set only when is_poison.
26041 2016-07-03 Richard Biener <rguenther@suse.de>
26043 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26045 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26046 (vectorizable_reduction): Perform reduction PHI creation when
26047 visiting a reduction PHI and adjust and simplify code generation
26048 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26049 (vect_transform_loop): Visit reduction PHIs.
26050 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26051 defs into the SLP tree.
26052 (vect_build_slp_tree): Reduction defs terminate the recursion.
26053 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26055 (vect_get_vec_defs_for_stmt_copy): Export.
26056 (vect_get_vec_defs): Likewise.
26057 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26059 (vect_get_vec_defs_for_stmt_copy): Declare.
26060 (vect_get_vec_defs): Likewise.
26062 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26064 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26065 parameter with a "loop" parameter and use it instead of the
26066 loop containing DR_STMT. Don't check simple_iv when doing
26067 BB analysis. Describe the two analysis modes in the comment.
26069 2017-07-03 Tom de Vries <tom@codesourcery.com>
26071 PR tree-optimization/69468
26072 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26073 (find_same_succ_bb): Handle ignore_edge_flags.
26075 2017-07-03 Tom de Vries <tom@codesourcery.com>
26077 PR tree-optimization/81192
26078 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26080 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26082 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26084 2017-07-03 Tom de Vries <tom@codesourcery.com>
26086 PR tree-optimization/81192
26087 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26088 BB_SAME_SUCC (bb) == NULL.
26090 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26092 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26095 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26097 * dumpfile.c: Include profile-count.h
26098 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26100 (insert_cond_bb): Update profile.
26101 * tree-cfg.h (insert_cond_bb): Update prototype.
26102 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26103 * tree-dump.c: Do not include tree-cfg.
26105 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26107 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26109 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26111 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26114 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26116 * tree-complex.c (expand_complex_div_wide): update profile.
26118 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26119 Alan Hayward <alan.hayward@arm.com>
26120 David Sherwood <david.sherwood@arm.com>
26122 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26123 (CORETYPES_H): New define.
26124 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26125 (insn-modes-inline.h, s-modes-inline-h): New rules.
26126 (generated_files): Add insn-modes-inline.h.
26127 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26128 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26129 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26130 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26131 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26132 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26133 (build/gencodes.o, build/genconditions.o): Likewise.
26134 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26135 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26136 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26137 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26138 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26139 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26140 * coretypes.h: Include everything up to real.h for generators.
26141 Include insn-modes.h first. Include wide-int-print.h after
26142 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26143 * machmode.h: Don't include insn-modes.h here.
26144 * function-tests.c: Remove includes of signop.h, machmode.h,
26145 double-int.h and wide-int.h.
26147 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26149 * optc-save-gen.awk: Likewise.
26150 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26151 * godump.c: Remove include of wide-int-print.h.
26152 * pretty-print.h: Likewise.
26153 * wide-int-print.cc: Likewise.
26154 * wide-int.cc: Likewise.
26155 * hash-map-tests.c: Remove include of signop.h.
26156 * hash-set-tests.c: Likewise.
26157 * rtl-tests.c: Likewise.
26158 * mkconfig.sh: Remove include of machmode.h.
26159 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26161 (emit_insn_modes_inline_h): ...this new function. Emit the code
26162 into an insn-modes-inline.h header file, adding appropriate
26163 include guards and end comments.
26164 (emit_insn_modes_c_header): Remove include of machmode.h.
26165 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26167 (main): Handle -i flag and call emit_insn_modes_inline_h when
26170 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26172 * tree-ssa-strlen.c (strinfo): Rename the length field to
26173 nonzero_chars. Add a full_string_p field.
26174 (compare_nonzero_chars, zero_length_string_p): New functions.
26175 (get_addr_stridx): Add an offset_out parameter.
26176 Use compare_nonzero_chars.
26177 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26178 (new_strinfo): Update after above changes to strinfo.
26179 (set_endptr_and_length): Set full_string_p.
26180 (get_string_length): Update after above changes to strinfo.
26181 (unshare_strinfo): Update call to new_strinfo.
26182 (maybe_invalidate): Likewise.
26183 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26184 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26185 as a uhwi instead of an shwi. Update after above changes to
26186 strinfo and new_strinfo.
26187 (zero_length_string): Assert that chainsi contains full strings.
26188 Use zero_length_string_p. Update call to new_strinfo.
26189 (adjust_related_strinfos): Update after above changes to strinfo.
26190 Copy full_string_p from origsi.
26191 (adjust_last_stmt): Use zero_length_string_p.
26192 (handle_builtin_strlen): Update after above changes to strinfo and
26193 new_strinfo. Install the lhs as the string length if the previous
26194 entry didn't describe a full string.
26195 (handle_builtin_strchr): Update after above changes to strinfo
26197 (handle_builtin_strcpy): Likewise.
26198 (handle_builtin_strcat): Likewise.
26199 (handle_builtin_malloc): Likewise.
26200 (handle_pointer_plus): Likewise.
26201 (handle_builtin_memcpy): Likewise. Track nonzero characters
26202 that aren't necessarily followed by a nul terminator.
26203 (handle_char_store): Likewise.
26205 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26207 PR tree-optimization/80769
26208 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26209 for malloc and calloc. Document the new invariant that all related
26210 strinfos have delayed lengths or none do.
26211 (verify_related_strinfos): Move earlier in file.
26212 (set_endptr_and_length): New function, split out from...
26213 (get_string_length): ...here. Also set the lengths of related
26215 (zero_length_string): Assert that chainsi has known (rather than
26217 (adjust_related_strinfos): Likewise.
26219 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26221 PR tree-optimization/81136
26222 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26223 assert that two references with the same misalignment have the same
26224 compile-time misalignment if those compile-time misalignments
26227 2017-07-01 Andi Kleen <ak@linux.intel.com>
26229 * print-tree.c (print_node): Print all attributes.
26231 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26233 * cfg.c (scale_bbs_frequencies): New function.
26234 * cfg.h (scale_bbs_frequencies): Declare it.
26235 * cfgloopanal.c (single_likely_exit): Cleanup.
26236 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26238 (scale_loop_profile): Likewise.
26239 (loop_version): Likewise.
26240 (create_empty_loop_on_edge): Update.
26241 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26242 scale_loop_frequencies, scale_loop_profile, loopify,
26243 loop_version): Update prototypes.
26244 * modulo-sched.c (sms_schedule): Update.
26245 * predict.c (unlikely_executed_edge_p): Also check probability.
26246 (probably_never_executed_edge_p): Fix typo.
26247 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26248 * tree-parloops.c (gen_parallel_loop): Update.
26249 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26250 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26251 * tree-ssa-loop-split.c (split_loop): Update.
26252 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26253 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26254 (vect_loop_versioning): Update.
26255 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26257 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26259 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26261 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26263 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26264 to keep profile consistent.
26266 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26268 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26269 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26270 * profile-count.h (max_safe_multiplier): Make unsigned.
26271 (profile_count::guessed_zero): New.
26273 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26275 * bb-reorder.c (fix_up_crossing_landing_pad,
26276 fix_crossing_conditional_branches): Use make_single_succ_edge
26277 to keep profile consistent.
26279 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26281 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26284 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26287 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26288 the right scopes, make sure cond_jump isn't preserved between multiple
26289 iterations. Search for fallthru edge whenever there are 3+ edges and
26290 use find_fallthru_edge for it.
26292 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26294 Patch by Alexander Monakov <amonakov@ispras.ru>
26295 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26296 probabilities consistently.
26298 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26300 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26301 emit_cmp_and_jump_insns.
26303 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26306 * tree-inline.c (expand_call_inline): Combine profile statuses.
26308 2017-06-30 Andrew Pinski <apinski@cavium.com>
26310 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26311 fold_stmt returned true.
26313 2017-06-30 Nathan Sidwell <nathan@acm.org>
26315 * ggc.h (empty_string): Delete.
26316 * cfgexpand.c (expand_asm_stmt): Use plain "".
26317 * optabs.c (expand_asm_memory_barrier): Likewise.
26318 * stringpool.c (empty_string): Delete.
26319 (digit_vector, digit_string): Delete.
26320 (ggc_alloc_string): Use plain "", don't optimize single digit
26321 strings. Use ggc_alloc_atomic.
26323 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26325 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26326 comparison set and one other set, use the cost of the non-comparison
26329 2017-06-30 Nathan Sidwell <nathan@acm.org>
26331 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26334 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26336 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26337 loops. Remove now unneeded calls to gimple_switch_set_label() that
26338 just set removed labels to NULL_TREE.
26340 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26342 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26343 (set_range_info): ...here. Only call set_range_info_raw if domain
26345 (set_nonzero_bits): Call set_range_info_raw.
26346 * tree-ssanames.h (set_range_info_raw): New.
26348 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26351 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26352 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26353 of nonimmediate_operand and <store_mask_constraint> instead of m
26354 for the input operand. For V8FI iterator, always split if input
26355 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26356 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26357 <store_mask_predicate> instead of register_operand and
26358 <store_mask_constraint> instead of v for the input operand. Make
26359 sure both operands aren't MEMs for if not <mask_applied>.
26361 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26363 * lto-wrapper.c (copy_file) Close both file descriptors before
26366 2017-06-30 Martin Liska <mliska@suse.cz>
26369 * multiple_target.c (create_dispatcher_calls): Make ifunc
26370 also for function that don't have calls or are not referenced.
26372 2017-06-30 Richard Biener <rguenther@suse.de>
26374 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26375 analyze the first scalar stmt. Move vector type computation
26376 for the BB case here from ...
26377 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26378 live operation processing in the SLP case properly.
26380 2017-06-30 Richard Biener <rguenther@suse.de>
26382 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26384 2017-06-30 Martin Liska <mliska@suse.cz>
26387 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26388 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26390 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26392 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26395 2017-06-30 Richard Biener <rguenther@suse.de>
26397 PR tree-optimization/81249
26398 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26399 condition reduction result to original scalar type.
26401 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26403 * profile-count.h (enum profile_quality): Fix typos and whitespace
26406 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26408 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26409 type for branch probabilities.
26411 2017-06-29 Julian Brown <julian@codesourcery.com>
26412 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26414 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26415 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26416 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26417 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26419 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26421 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26422 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26423 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26424 CC usage from generic code to here.
26425 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26426 CC usage into the target macros.
26428 2017-06-29 Maya Rashish <coypu@sdf.org>
26430 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26433 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26435 * arm/arm-builtins.c: Include profile-count.h
26436 * except.c (sjlj_emit_function_enter): Use
26437 profile_probability::unlikely.
26439 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26441 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26442 and tocrel_offset be pointer args rather than implicitly using
26444 (legitimate_constant_pool_address_p, rs6000_emit_move,
26445 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26446 tocrel_offset and use in toc_relative_expr_p call.
26447 (print_operand, print_operand_address): Use static tocrel_base_oac
26448 and tocrel_offset_oac.
26449 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26452 2017-06-29 Maya Rashish <coypu@sdf.org>
26454 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26456 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26458 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26459 objects, take into account only the alignment of 'op0' and 'mode1' if
26462 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26464 * ccmp.c (ccmp_tree_comparison_p): New function.
26465 (ccmp_candidate_p): Update to use above function.
26466 (get_compare_parts): New function.
26467 (expand_ccmp_next): Update to use new functions.
26468 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26470 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26471 take mode as argument.
26472 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26473 * expr.c (expand_expr_real_1): Pass mode as argument.
26475 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26477 * combine.c (combine_instructions): Print insns to dump_file, together
26480 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26482 * asan.c (asan_emit_stack_protection): Update.
26483 (create_cond_insert_point): Update.
26484 * auto-profile.c (afdo_propagate_circuit): Update.
26485 * basic-block.h (struct edge_def): Turn probability to
26486 profile_probability.
26487 (EDGE_FREQUENCY): Update.
26488 * bb-reorder.c (find_traces_1_round): Update.
26489 (better_edge_p): Update.
26490 (sanitize_hot_paths): Update.
26491 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26492 (make_single_succ_edge): Update.
26493 (check_bb_profile): Update.
26494 (dump_edge_info): Update.
26495 (update_bb_profile_for_threading): Update.
26496 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26497 probabilitycount to 0.
26498 * cfgbuild.c (compute_outgoing_frequencies): Update.
26499 * cfgcleanup.c (try_forward_edges): Update.
26500 (outgoing_edges_match): Update.
26501 (try_crossjump_to_edge): Update.
26502 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26503 (expand_gimple_tailcall): Update.
26504 (construct_init_block): Use make_single_succ_edge.
26505 (construct_exit_block): Use make_single_succ_edge.
26506 * cfghooks.c (verify_flow_info): Update.
26507 (redirect_edge_succ_nodup): Update.
26508 (split_edge): Update.
26509 (account_profile_record): Update.
26510 * cfgloopanal.c (single_likely_exit): Update.
26511 * cfgloopmanip.c (scale_loop_profile): Update.
26512 (set_zero_probability): Remove.
26513 (duplicate_loop_to_header_edge): Update.
26514 * cfgloopmanip.h (loop_version): Update prototype.
26515 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26516 (force_nonfallthru_and_redirect): Update.
26517 (update_br_prob_note): Update.
26518 (rtl_verify_edges): Update.
26519 (purge_dead_edges): Update.
26520 (rtl_lv_add_condition_to_bb): Update.
26521 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26522 * cgraphunit.c (init_lowered_empty_function): Update.
26523 (cgraph_node::expand_thunk): Update.
26524 * cilk-common.c: Include profile-count.h
26525 * dojump.c (inv): Remove.
26526 (jumpifnot): Update.
26527 (jumpifnot_1): Update.
26528 (do_jump_1): Update.
26530 (do_jump_by_parts_greater_rtx): Update.
26531 (do_compare_rtx_and_jump): Update.
26532 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26533 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26534 * dwarf2cfi.c: Include profile-count.h
26535 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26536 (sjlj_emit_dispatch_table): Likewise.
26537 * explow.c: Include profile-count.h
26538 * expmed.c (emit_store_flag_force): Update.
26539 (do_cmp_and_jump): Update.
26540 * expr.c (compare_by_pieces_d::generate): Update.
26541 (compare_by_pieces_d::finish_mode): Update.
26542 (emit_block_move_via_loop): Update.
26543 (store_expr_with_bounds): Update.
26544 (store_constructor): Update.
26545 (expand_expr_real_2): Update.
26546 (expand_expr_real_1): Update.
26547 * expr.h (try_casesi, try_tablejump): Update prototypes.
26548 * gimple-pretty-print.c (dump_probability): Update.
26549 (dump_profile): New.
26550 (dump_gimple_label): Update.
26551 (dump_gimple_bb_header): Update.
26552 * graph.c (draw_cfg_node_succ_edges): Update.
26553 * hsa-gen.c (convert_switch_statements): Update.
26554 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26555 (find_if_case_1): Update.
26556 (find_if_case_2): Update.
26557 * internal-fn.c (expand_arith_overflow_result_store): Update.
26558 (expand_addsub_overflow): Update.
26559 (expand_neg_overflow): Update.
26560 (expand_mul_overflow): Update.
26561 (expand_vector_ubsan_overflow): Update.
26562 * ipa-cp.c (good_cloning_opportunity_p): Update.
26563 * ipa-split.c (split_function): Use make_single_succ_edge.
26564 * ipa-utils.c (ipa_merge_profiles): Update.
26565 * loop-doloop.c (add_test): Update.
26566 (doloop_modify): Update.
26567 * loop-unroll.c (compare_and_jump_seq): Update.
26568 (unroll_loop_runtime_iterations): Update.
26569 * lra-constraints.c (lra_inheritance): Update.
26570 * lto-streamer-in.c (input_cfg): Update.
26571 * lto-streamer-out.c (output_cfg): Update.
26572 * mcf.c (adjust_cfg_counts): Update.
26573 * modulo-sched.c (sms_schedule): Update.
26574 * omp-expand.c (expand_omp_for_init_counts): Update.
26575 (extract_omp_for_update_vars): Update.
26576 (expand_omp_ordered_sink): Update.
26577 (expand_omp_for_ordered_loops): Update.
26578 (expand_omp_for_generic): Update.
26579 (expand_omp_for_static_nochunk): Update.
26580 (expand_omp_for_static_chunk): Update.
26581 (expand_cilk_for): Update.
26582 (expand_omp_simd): Update.
26583 (expand_omp_taskloop_for_outer): Update.
26584 (expand_omp_taskloop_for_inner): Update.
26585 * omp-simd-clone.c (simd_clone_adjust): Update.
26586 * optabs.c (expand_doubleword_shift): Update.
26587 (expand_abs): Update.
26588 (emit_cmp_and_jump_insn_1): Update.
26589 (expand_compare_and_swap_loop): Update.
26590 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26591 * predict.c (predictable_edge_p): Update.
26592 (edge_probability_reliable_p): Update.
26593 (set_even_probabilities): Update.
26594 (combine_predictions_for_insn): Update.
26595 (combine_predictions_for_bb): Update.
26596 (propagate_freq): Update.
26597 (estimate_bb_frequencies): Update.
26598 (force_edge_cold): Update.
26599 * profile-count.c (profile_count::dump): Add missing space into dump.
26600 (profile_count::debug): Add newline.
26601 (profile_count::differs_from_p): Explicitly convert to unsigned.
26602 (profile_count::stream_in): Update.
26603 (profile_probability::dump): New member function.
26604 (profile_probability::debug): New member function.
26605 (profile_probability::differs_from_p): New member function.
26606 (profile_probability::differs_lot_from_p): New member function.
26607 (profile_probability::stream_in): New member function.
26608 (profile_probability::stream_out): New member function.
26609 * profile-count.h (profile_count_quality): Rename to ...
26610 (profile_quality): ... this one.
26611 (profile_probability): New.
26612 (profile_count): Update.
26613 * profile.c (compute_branch_probabilities): Update.
26614 * recog.c (peep2_attempt): Update.
26615 * sched-ebb.c (schedule_ebbs): Update.
26616 * sched-rgn.c (find_single_block_region): Update.
26617 (compute_dom_prob_ps): Update.
26618 (schedule_region): Update.
26619 * sel-sched-ir.c (compute_succs_info): Update.
26620 * stmt.c (struct case_node): Update.
26621 (do_jump_if_equal): Update.
26622 (get_outgoing_edge_probs): Update.
26623 (conditional_probability): Update.
26624 (emit_case_dispatch_table): Update.
26625 (expand_case): Update.
26626 (expand_sjlj_dispatch_table): Update.
26627 (emit_case_nodes): Update.
26628 * targhooks.c: Update.
26629 * tracer.c (better_p): Update.
26630 (find_best_successor): Update.
26631 * trans-mem.c (expand_transaction): Update.
26632 * tree-call-cdce.c: Update.
26633 * tree-cfg.c (gimple_split_edge): Upate.
26634 (move_sese_region_to_fn): Upate.
26635 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26636 * tree-eh.c (lower_resx): Upate.
26637 (cleanup_empty_eh_move_lp): Upate.
26638 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26639 * tree-inline.c (copy_edges_for_bb): Update.
26640 (copy_cfg_body): Update.
26641 * tree-parloops.c (gen_parallel_loop): Update.
26642 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26643 (gimple_gen_time_profiler): Update.
26644 * tree-ssa-dce.c (remove_dead_stmt): Update.
26645 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26646 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26647 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26648 (unloop_loops): Update.
26649 (try_peel_loop): Update.
26650 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26651 * tree-ssa-loop-split.c (connect_loops): Update.
26652 (split_loop): Update.
26653 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26654 (hoist_guard): Update.
26655 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26656 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26657 (value_replacement): Update.
26658 * tree-ssa-reassoc.c (branch_fixup): Update.
26659 * tree-ssa-tail-merge.c (replace_block_by): Update.
26660 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26661 (create_edge_and_update_destination_phis): Update.
26662 (compute_path_counts): Update.
26663 (recompute_probabilities): Update.
26664 (update_joiner_offpath_counts): Update.
26665 (freqs_to_counts_path): Update.
26666 (duplicate_thread_path): Update.
26667 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26668 (struct switch_conv_info): Update.
26669 (gen_inbound_check): Update.
26670 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26671 (vect_do_peeling): Update.
26672 (vect_loop_versioning): Update.
26673 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26674 (optimize_mask_stores): Update.
26675 * ubsan.c (ubsan_expand_null_ifn): Update.
26676 * value-prof.c (gimple_divmod_fixed_value): Update.
26677 (gimple_divmod_fixed_value_transform): Update.
26678 (gimple_mod_pow2): Update.
26679 (gimple_mod_pow2_value_transform): Update.
26680 (gimple_mod_subtract): Update.
26681 (gimple_mod_subtract_transform): Update.
26682 (gimple_ic): Update.
26683 (gimple_stringop_fixed_value): Update.
26684 (gimple_stringops_transform): Update.
26685 * value-prof.h: Update.
26687 2017-06-29 Carl Love <cel@us.ibm.com>
26689 * config/rs6000/rs6000-c.c: Add support for built-in functions
26690 vector signed int vec_signed (vector float);
26691 vector signed long long vec_signed (vector double);
26692 vector signed int vec_signed2 (vector double, vector double);
26693 vector signed int vec_signede (vector double);
26694 vector signed int vec_signedo (vector double);
26695 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26696 instruction generator.
26697 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26698 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26699 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26701 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26702 vunsignede_v2df): Add define_expands.
26703 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26704 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26705 VEC_UNSIGNEDO): Add definitions.
26706 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26707 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26708 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26709 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26710 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26711 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26712 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26713 vec_unsignede, vec_unsignedo): Add builtin defines.
26714 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26716 * doc/extend.texi: Update the built-in documentation file for the
26717 new built-in functions.
26719 2017-06-29 Richard Biener <rguenther@suse.de>
26721 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26722 reduction chains to LOOP_VINFO_REDUCTIONS.
26723 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26724 SLP reductions after processing reduction chains.
26726 2017-06-29 Nathan Sidwell <nathan@acm.org>
26728 * builtins.c (fold_builtin_FUNCTION): Use
26729 lang_hooks.decl_printable_name.
26731 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26733 PR middle-end/81194
26734 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26735 with only one label.
26736 * stmt.c (expand_case): Assert NCASES is greater than one.
26738 2017-06-29 Richard Biener <rguenther@suse.de>
26740 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26742 (group_case_labels): Likewise.
26743 (find_taken_edge): Push sanity checking on val to workers...
26744 (find_taken_edge_cond_expr): ... here
26745 (find_taken_edge_switch_expr): ... and here, handle cases
26746 with just a default label.
26747 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26748 (group_case_labels): Likewise.
26749 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26750 group_case_labels does anything cleanup the CFG again.
26752 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26754 PR tree-optimization/81196
26755 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26756 exit condition comparing two IVs.
26758 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26760 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26761 profile to the dummy entry at the end of the list of architectures.
26762 * config/arm/arm-cpu-cdata.h: Regenerated.
26764 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26765 Michael Collison <michael.collison@arm.com>
26768 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26770 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26771 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26772 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26773 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26774 of shift when the shift amount is masked with constant equal to
26775 the size of the mode.
26776 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26779 2017-06-29 Martin Liska <mliska@suse.cz>
26781 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26783 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26786 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26789 (edge_growth_cache_entry::edge_growth_cache_entry): New
26791 (reset_edge_growth_cache): Update to use constructor.
26793 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26795 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26796 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26797 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26799 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26801 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26802 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26804 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26806 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26807 (*-linux-uclibc*): Add t-uclibc tmake_file.
26808 * config/t-musl: New.
26809 * config/t-uclibc: New.
26811 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26813 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26815 (gen_comm_data): Emit architectural setting of arch_prof.
26816 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26818 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26819 (armv8-m.base, armv8-m.main): Likewise.
26820 * arm-protos.h (arm_build_target): Add profile field.
26821 (arch_option): Likewise.
26822 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26824 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26825 arm_active_target.profile.
26827 2017-06-28 Richard Biener <rguenther@suse.de>
26829 PR middle-end/81227
26830 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26831 TYPE_OVERFLOW_WRAPS.
26832 * match.pd (negate_expr_p): Likewise.
26833 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26834 fold_build2, not fold_binary.
26836 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26838 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26839 Convert memory address to Pmode.
26840 (aarch64_print_operand): Assert MEM operands are always Pmode.
26842 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26845 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26846 Remove redundant if.
26847 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26848 * config/arm/aarch-common-protos.h
26849 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26850 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26852 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26855 * multiple_target.c (create_dispatcher_calls): Set the default
26856 clone to be static, not public.
26858 2017-06-28 Richard Biener <rguenther@suse.de>
26860 * tree-vect-loop.c (vectorizable_reduction): Move special
26861 cond reduction IV var creation ...
26862 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26863 parameter. Use STMT_VINFO_VECTYPE.
26864 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26867 2017-06-28 Martin Liska <mliska@suse.cz>
26870 * ipa-visibility.c (non_local_p): Handle visibility.
26872 2017-06-28 Martin Liska <mliska@suse.cz>
26875 * common.opt: Add IntegerRange to various options.
26876 * opt-functions.awk (integer_range_info): New function.
26877 * optc-gen.awk: Add integer_range_info to cl_options struct.
26878 * opts-common.c (decode_cmdline_option): Handle
26879 CL_ERR_INT_RANGE_ARG.
26880 (cmdline_handle_error): Likewise.
26881 * opts.c (print_filtered_help): Show valid interval in
26882 when --help is provided.
26883 * opts.h (struct cl_option): Add range_min and range_max fields.
26884 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26886 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26888 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26889 (x * C EQ/NE y * C): New transformation.
26891 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26893 * genmultilib (combination_space): Accept '+' in option names.
26895 2017-06-28 Martin Liska <mliska@suse.cz>
26898 * asan.c (instrument_derefs): Bail out inner references
26899 that are hard register variables.
26901 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26904 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26905 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26907 2017-06-28 Richard Biener <rguenther@suse.de>
26909 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26910 (vect_get_slp_defs): Adjust.
26911 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26913 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26915 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26916 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26917 (vectorizable_reduction): Adjust.
26918 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26920 (vect_get_slp_defs): Likewise.
26921 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26922 (vectorizable_bswap): Adjust.
26923 (vectorizable_call): Likewise.
26924 (vectorizable_conversion): Likewise.
26925 (vectorizable_assignment): Likewise.
26926 (vectorizable_shift): Likewise.
26927 (vectorizable_operation): Likewise.
26928 (vectorizable_store): Likewise.
26929 (vectorizable_condition): Likewise.
26930 (vectorizable_comparison): Likewise.
26932 2017-06-28 Michael Collison <michael.collison@arm.com>
26935 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26937 (arm_gen_movmemqi): Removed unused variable 'i'.
26938 Convert 'for' loop into 'while' loop.
26939 (arm_expand_prologue): Remove last unnecessary set of insn.
26940 (thumb_pop): Remove unused variable 'pushed_words'.
26941 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26943 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26945 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26946 * config/s390/s390-protos.h: Add prototype of
26947 s390_rel_address_ok_p.
26948 * config/s390/s390.c (s390_got_symbol): New function.
26949 (s390_rel_address_ok_p): New function.
26950 (legitimize_pic_address): Use s390_rel_address_ok_p.
26951 (s390_load_got): Use s390_got_symbol.
26952 (s390_option_override): Issue error if
26953 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26954 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26956 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26958 2017-06-27 Andrew Pinski <apinski@cavium.com>
26960 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26961 (X * copysign (1.0, X)): New pattern.
26962 (X * copysign (1.0, -X)): New pattern.
26963 (copysign (-1.0, CST)): New pattern.
26965 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26967 * genmultilib (combination_space): Remove variable.
26968 Validate reuse rules against regular expression for any sequence
26969 of multilib options in any order.
26971 2017-06-27 Michael Collison <michael.collison@arm.com>
26973 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26974 call aarch64_split_simd_combine.
26975 * (aarch64_combine_internal<mode>): Delete pattern.
26976 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26977 Allow register and subreg operands.
26979 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26981 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26982 specific need, just fallback on defaults.
26983 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26985 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26986 Olivier Hainque <hainque@adacore.com>
26988 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26990 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26991 targets. Pick a default if no particular attempt applied.
26992 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26995 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26997 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26998 (x86_64-wrs-vxworks7): Likewise.
27000 2017-06-27 Marek Polacek <polacek@redhat.com>
27003 * ubsan.c (instrument_null): Check get_base_address's result for null.
27005 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27007 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27009 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27011 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27012 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27013 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27014 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27015 New function types.
27016 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27017 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27018 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27019 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27020 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27021 BUILT_IN_FEUPDATEENV): New builtins.
27022 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27023 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27024 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27025 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27027 (builtin_structptr_types): Adjust size.
27028 * tree.c (builtin_structptr_types): Add four entries.
27030 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27031 Olivier Hainque <hainque@adacore.com>
27033 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27034 (TLS_SYM): New local macro, forcing reference to __tls__ on
27035 link command lines for VxWorks 7 RTPs, triggering initialization
27037 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27038 OS features TLS support, true for RTPs on VxWorks 7.
27039 * config/vxworks.c (vxworks_override_options): Setup emutls
27042 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27044 * predict.c (test_prediction_value_range): Use -1U instead of -1
27045 to avoid narrowing conversion warning.
27046 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27047 to avoid narrowing conversion warning.
27048 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27050 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27052 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27054 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27055 64bit configurations.
27056 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27057 (SIZE_TYPE): Likewise.
27058 * config/vxworks.c (vxworks_emutls_var_fields): Use
27059 long_unsigned_type_node instead of unsigned_type_node as the offset
27060 field type, which is "pointer" mode in emutls.c.
27062 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27065 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27067 PR middle-end/81207
27068 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27069 gimple_vuse copying separately from gimple_vdef copying.
27071 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27073 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27076 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27077 Olivier Hainque <hainque@adacore.com>
27079 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27080 for all vxworks7 targets.
27081 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27082 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27083 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27084 variations for VX6/VX7 and 32/64bits later on in ...
27085 (VXWORKS_LIB_SPEC): Leverage new macros.
27086 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27087 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27089 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27091 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27092 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27094 2017-06-26 Carl Love <cel@us.ibm.com>
27096 * config/rs6000/rs6000-c.c: Add support for built-in functions
27097 vector bool char vec_reve (vector bool char);
27098 vector signed char vec_reve (vector signed char);
27099 vector unsigned char vec_reve (vector unsigned char);
27100 vector bool int vec_reve (vector bool int);
27101 vector signed int vec_reve (vector signed int);
27102 vector unsigned int vec_reve (vector unsigned int);
27103 vector bool long long vec_reve (vector bool long long);
27104 vector signed long long vec_reve (vector signed long long);
27105 vector unsigned long long vec_reve (vector unsigned long long);
27106 vector bool short vec_reve (vector bool short);
27107 vector signed short vec_reve (vector signed short);
27108 vector double vec_reve (vector double);
27109 vector float vec_reve (vector float);
27110 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27111 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27112 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27113 (altivec_vreve): New pattern.
27114 * config/rs6000/altivec.h (vec_reve): New define.
27115 * doc/extend.texi (vec_rev): Update the built-in documentation file
27116 for the new built-in functions.
27118 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27120 PR tree-optimization/71815
27121 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27123 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27125 (slsr_process_phi): Likewise.
27126 (replace_uncond_cands_and_profitable_phis): Don't replace a
27127 multiply candidate with a stride of 1 (copy or cast).
27128 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27130 (lowest_cost_path): Likewise.
27131 (total_savings): Likewise.
27133 2017-06-26 Richard Biener <rguenther@suse.de>
27136 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27137 Use def_builtin_pure for all gather builtins.
27139 2017-06-26 Richard Biener <rguenther@suse.de>
27141 PR tree-optimization/81203
27142 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27143 non-dominating BBs.
27145 2017-06-26 Marek Polacek <polacek@redhat.com>
27148 * doc/invoke.texi: Document -Wmultistatement-macros.
27150 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27152 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27153 arm_neon_ok_no_float_abi effective target.
27155 2017-06-26 Richard Biener <rguenther@suse.de>
27157 PR tree-optimization/80928
27158 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27159 (copy_bbs): Set BB_DUPLICATED flag early.
27160 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27162 (execute_on_shrinking_pred): Likewise.
27163 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27164 BB_DUPLICATED blocks.
27165 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27166 iterate over all PHIs considering removal of *gsi.
27168 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27170 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27173 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27175 * config/rs6000/rs6000-string.c: (expand_block_clear,
27176 do_load_for_compare, select_block_compare_mode,
27177 compute_current_alignment, expand_block_compare,
27178 expand_strncmp_align_check, expand_strn_compare,
27179 expand_block_move, rs6000_output_load_multiple)
27180 Move functions related to string/block move/compare
27181 to a separate file.
27182 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27183 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27184 for this function which is now used in two files.
27185 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27186 * config.gcc: Add rs6000-string.o to extra_objs for
27187 targets powerpc*-*-* and rs6000*-*-*.
27189 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27192 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27193 32-bit, since indexed is not valid for DImode.
27194 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27195 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27196 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27197 (define_peephole2 for Altivec d-form store): Likewise.
27200 * multiple_target.c (create_dispatcher_calls): Only create the
27201 dispatcher call if the function is the default clone of a
27202 versioned function.
27204 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27206 PR middle-end/80902
27207 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27208 a call, force the call to not be a tail call.
27210 2017-06-23 Jeff Law <law@redhat.com>
27212 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27215 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27217 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27218 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27219 handling for early expansion of vector shifts (sl,sr,sra,rl).
27220 (builtin_function_type): Add vector shift right instructions
27221 to the unsigned argument list.
27223 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27225 rtl-optimizatoin/79286
27226 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27227 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27228 trap. PIC register plus a const unspec without offset can never trap.
27230 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27232 * tree.h (builtin_structptr_type): New type.
27233 (builtin_structptr_types): Declare new array.
27234 * tree.c (builtin_structptr_types): New array.
27235 (free_lang_data, build_common_tree_nodes): Use it.
27237 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27240 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27243 2017-06-22 Matt Turner <mattst88@gmail.com>
27245 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27246 Lake models to skylake case. Assume skylake for unknown
27247 models with clflushopt.
27249 2017-06-22 Jeff Law <law@redhat.com>
27251 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27252 frame sizes that do not satisfy aarch64_uimm12_shift.
27254 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27256 * profile-count.h (apply_probability,
27257 apply_scale, probability_in): Fix checks for zero.
27259 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27261 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27262 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27264 2016-06-22 Richard Biener <rguenther@suse.de>
27266 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27267 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27268 REDUC_MAX_EXPR support.
27269 (vectorizable_reduction): Likewise.
27270 (vect_create_epilog_for_reduction): Likewise.
27272 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27274 * match.pd (A / (1 << B) -> A >> B): New.
27275 * generic-match-head.c: Include optabs-tree.h.
27276 * gimple-match-head.c: Likewise.
27277 * optabs-tree.h (target_supports_op_p): New.
27278 * optabs-tree.c (target_supports_op_p): New.
27280 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27282 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27283 $gcc_cv_ld --help output.
27284 (gcc_cv_ld_demangle): Likewise.
27285 (gcc_cv_ld_eh_frame_hdr): Likewise.
27286 (gcc_cv_ld_pie): Likewise.
27287 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27288 (gcc_cv_ld_buildid): Likewise.
27289 (gcc_cv_ld_sysroot): Likewise.
27290 (ld_bndplt_support): Likewise.
27291 (ld_pushpopstate_support): Likewise.
27292 * configure: Regenerate.
27293 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27295 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27298 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27299 operands indexes to avoid gap between operands and match_dups.
27301 2017-06-21 Andrew Pinski <apinski@cavium.com>
27303 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27304 Increment Arith_shift and Arith_shift_reg by 1.
27305 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27307 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27308 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27309 (aarch64_strip_extend): Add new argument and test for it.
27310 (aarch64_cheap_mult_shift_p): New function.
27311 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27312 add a cost if it is true.
27313 Update calls to aarch64_strip_extend.
27314 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27316 2017-06-21 Andrew Pinski <apinski@cavium.com>
27318 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27320 (thunderxt88): Likewise.
27321 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27322 (thunderx_prefetch_tune): New variable.
27323 (thunderx2t99_prefetch_tune): Update for the correct values.
27324 (thunderxt88_tunings): New variable.
27325 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27326 generic_prefetch_tune.
27327 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27329 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27331 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27332 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27333 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27334 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27335 (aarch64_atomic_cas<mode>, GPI): Likewise.
27337 2017-06-21 Martin Liska <mliska@suse.cz>
27339 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27340 statements on cold and hot labels.
27341 * predict.c (tree_estimate_probability_bb): Remove the
27342 prediction from this place.
27344 2017-06-21 Martin Liska <mliska@suse.cz>
27346 PR tree-optimization/79489
27347 * gimplify.c (maybe_add_early_return_predict_stmt): New
27349 (gimplify_return_expr): Call the function.
27350 * predict.c (tree_estimate_probability_bb): Remove handling
27352 * predict.def: Update comment about early return predictor.
27353 * gimple-predict.h (is_gimple_predict): New function.
27354 * predict.def: Change default value of early return to 66.
27355 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27357 * passes.def: Put pass_strip_predict_hints to the beginning of
27360 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27362 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27363 FUNCTION_DECL declarations.
27364 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27366 (dwaf2out_decl): Likewise.
27367 * godump.c (go_early_global_decl): Skip call to the real debug hook
27368 for FUNCTION_DECL declarations.
27369 * passes.c (rest_of_decl_compilation): Skip call to the
27370 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27371 -fdump-go-spec is passed.
27373 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27375 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27376 Reorder for compactness.
27377 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27378 (def_builtin_pure, def_builtin_pure2): New functions.
27379 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27381 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27383 * match.pd (nop_convert): New predicate.
27384 ((A +- CST1) +- CST2): Allow some NOP conversions.
27386 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27389 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27390 with ctors/dtors if GOVD_SHARED is set.
27392 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27394 * config/aarch64/aarch64.md (movti_aarch64):
27395 Emit mov rather than orr.
27396 (movtf_aarch64): Likewise.
27397 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27398 Emit mov rather than orr.
27400 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27402 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27403 Swap alternatives, make integer dup more expensive.
27405 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27407 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27408 Return true for non-tls symbols.
27410 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27412 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27413 (cortex-a75): Likewise.
27414 (cortex-a75.cortex-a55): Likewise.
27415 * config/aarch64/aarch64-tune.md: Regenerate.
27416 * doc/invoke.texi (-mtune): Document new values for -mtune.
27418 2017-06-21 Tom de Vries <tom@codesourcery.com>
27420 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27421 stack_size feature.
27422 (Effective-Target Keywords, Other attributes): Suggest using
27423 dg-add-options stack_size feature to get stack limit in stack_size
27424 effective target documentation.
27426 2017-06-21 Julian Brown <julian@codesourcery.com>
27427 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27429 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27430 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27431 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27433 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27434 attribute type list for neon_multiply.
27435 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27436 attribute type list for neon_multiply.
27437 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27438 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27439 attribute type list for neon_multiply.
27440 * config/arm/types.md (crypto_pmull): Add.
27441 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27442 attribute type list.
27444 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27446 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27449 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27451 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27452 to make sure not to dereference a NULL cost_classes_ptr pointer.
27454 2017-06-20 Carl Love <cel@us.ibm.com>
27456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27457 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27458 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27459 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27460 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27461 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27462 VMULOSW): New enum "unspec" values.
27463 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27464 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27465 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27466 altivec_vmulosw): New patterns.
27467 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27468 VMULOSW): Add definitions.
27470 2017-06-20 Julia Koval <julia.koval@intel.com>
27472 * config/i386/i386.c: Fix rounding expand for new pattern.
27473 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27475 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27477 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27478 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27480 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27482 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27485 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27487 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27488 architecture, then by implementer ID.
27489 * config/aarch64/aarch64-tune.md: Regenerate.
27491 2017-06-20 Richard Biener <rguenther@suse.de>
27493 PR middle-end/81097
27494 * fold-const.c (split_tree): Fold to type before negating.
27496 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27498 * diagnostic-show-locus.c
27499 (selftest::test_fixit_deletion_affecting_newline): New function.
27500 (selftest::diagnostic_show_locus_c_tests): Call it.
27502 2017-06-20 Andreas Schwab <schwab@suse.de>
27505 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27508 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27510 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27511 __ARM_FEATURE_COPROC according to support.
27513 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27515 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27516 Rewritten to avoid overflow for > 32-bit pointers.
27519 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27520 by removing enum keyword.
27521 (ubsan_type_descriptor): Likewise. Formatting fix.
27524 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27525 splitter): Require TARGET_SSE2 in the condition.
27527 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27530 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27531 for doing vector set of SFmode on ISA 3.0.
27532 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27533 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27535 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27536 SFmode value into a V4SF variable that was extracted from another
27537 V4SF variable without converting the element to double precision
27538 and back to single precision vector format.
27539 (vsx_insert_extract_v4sf_p9_2): Likewise.
27541 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27543 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27544 in UWHI to avoid undefined overflow.
27547 * ubsan.h (enum ubsan_encode_value_phase): New.
27548 (ubsan_encode_value): Change second argument to
27549 enum ubsan_encode_value_phase with default value of
27550 UBSAN_ENCODE_VALUE_GENERIC.
27551 * ubsan.c (ubsan_encode_value): Change second argument to
27552 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27553 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27554 create_tmp_var_raw instead of create_tmp_var and use a
27556 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27557 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27558 ubsan_encode_value callers.
27561 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27562 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27563 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27565 2017-06-19 Richard Biener <rguenther@suse.de>
27567 PR middle-end/81118
27568 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27569 estimates if we changed anything.
27571 2017-06-19 Richard Biener <rguenther@suse.de>
27573 PR tree-optimization/80887
27574 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27575 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27576 simplified lookups, then reset mprts_hook.
27577 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27579 (try_to_simplify): Likewise.
27581 2017-06-19 Martin Liska <mliska@suse.cz>
27584 * gimplify.c (gimplify_switch_expr):
27585 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27587 2017-06-19 Martin Liska <mliska@suse.cz>
27589 * doc/install.texi: Document that PGO runs in 4 stages.
27591 2017-06-19 Martin Liska <mliska@suse.cz>
27594 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27595 to dispatcher function name.
27596 * multiple_target.c (replace_function_decl): New function.
27597 (create_dispatcher_calls): Redirect both edges and references.
27599 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27601 * profile-count.c (profile_count::dump): Dump quality.
27602 (profile_count::differs_from_p): Update for unsigned val.
27603 * profile-count.h (profile_count_quality): New enum.
27604 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27606 2017-06-19 Richard Biener <rguenther@suse.de>
27608 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27609 struct function as arg.
27610 (estimate_numbers_of_iterations): Export overload with loop arg.
27611 (free_numbers_of_iterations_estimates_loop): Use an overload of
27612 free_numbers_of_iterations_estimates instead.
27613 * tree-cfg.c (remove_bb): Adjust.
27614 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27615 * tree-parloops.c (gen_parallel_loop): Likewise.
27616 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27618 (tree_unroll_loops_completely): Likewise.
27619 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27620 Use an overload instead and export.
27621 (estimated_loop_iterations): Adjust.
27622 (max_loop_iterations): Likewise.
27623 (likely_max_loop_iterations): Likewise.
27624 (estimate_numbers_of_iterations): Take struct function as arg
27626 (loop_exits_before_overflow): Adjust.
27627 (free_numbers_of_iterations_estimates_loop): Use an overload.
27628 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27629 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27631 2017-06-19 Richard Biener <rguenther@suse.de>
27634 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27635 RANGE_EXPR conservatively.
27637 2017-06-16 Carl Love <cel@us.ibm.com>
27639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27640 definitions for vec_float, vec_float2, vec_floato,
27641 vec_floate built-ins.
27642 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27643 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27645 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27646 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27647 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27648 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27649 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27650 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27651 vec_floato): Add builtin defines.
27652 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27653 Update the built-in documentation file for the new built-in
27656 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27658 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27659 (mthumb): Mark as the negative of -marm.
27661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27663 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27665 (ARM Options, -mtune): Document that this accepts the same
27666 extension options as -mcpu.
27667 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27671 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27672 permitted extensions.
27674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27676 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27677 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27678 (armv8-m.main): Add option +nodsp.
27679 * config/arm/arm-cpu-cdata.h: Regenerated.
27681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27683 * config/arm/t-fuchsia: New file.
27684 * config.gcc (arm*-*-fuchsia*): Use it.
27686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27688 * config/arm/t-symbian: Rewrite for new option infrastructure.
27690 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27692 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27693 (MULTILIB_REQUIRED): Likewise.
27695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27697 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27698 (MULTILIB_RESUE): Likewise.
27699 (MULTILIB_MATCHES): Likewise.
27700 (MULTLIB_REQUIRED): Likewise.
27702 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27704 * config/arm/t-rtems: Rewrite for new option framework.
27706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27708 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27709 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27710 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27711 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27712 * config/arm/t-multilib: ... here.
27713 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27714 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27715 armv7-a and armv8*-a when A-profile libraries have not been built.
27716 * config/arm/t-rmprofile: Rewrite.
27718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27720 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27721 with a backslash. Remove the backslash after substituting unescaped
27723 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27725 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27727 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27728 the driver to pass through the default setting of -mfloat-abi.
27729 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27731 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27732 (all_feat_combs): New rule.
27733 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27735 * config/arm/t-aprofile: Rewrite.
27737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27739 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27740 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27741 fpu is not specified by the user/command-line.
27742 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27743 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27744 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27745 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27746 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27747 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27751 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27752 * config/arm/t-arm-elf: Rewritten.
27754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27756 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27757 have some floating-point instructions.
27758 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27759 (TARGET_MAYBE_HARD_FLOAT): New macro.
27760 * config/arm/arm-builtins.c (arm_init_builtins): Use
27761 TARGET_MAYBE_HARD_FLOAT.
27762 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27766 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27767 (configargs.h): Include it.
27768 (arm_print_hint_for_fpu_option): New function.
27769 (arm_parse_fpu_option): New function.
27770 (candidate_extension): New class.
27771 (arm_canon_for_multilib): New function.
27772 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27773 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27774 (ARCH_CANONICAL_SPECS): New macro.
27775 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27779 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27780 are set after handling multilib fragments. Set target_cpu_default2
27783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27785 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27787 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27788 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27790 (check_arch): Likewise.
27791 * config/arm/arm.c (arm_configure_build_target): Handle
27792 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27793 options in the default.
27795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27797 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27798 when an option is an alias of another.
27799 * config/arm/parsecpu.awk (optalias): New parser token.
27800 (gen_comm_data): Mark non-alias options as such. Emit entries
27801 for extension aliases.
27802 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27803 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27804 (armv6kz, armv6zk, armv6t2): Likewise.
27805 (armv7): Make vfpv3-d16 an alias.
27806 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27808 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27809 Sort in canonical order.
27810 (armv8-a): Sort in canonical order.
27811 (armv8.1-a, armv8.2-a): Likewise.
27812 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27814 (cortex-a9): Sort in canonical order.
27815 * config/arm/arm.c (selftests.h): Include it.
27816 (arm_test_cpu_arch_data): New function.
27817 (arm_run_self_tests): New function.
27818 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27819 (targetm): Move declaration to the end of the file.
27820 * arm-cpu-cdata.h: Regenerated.
27822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27824 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27825 call to target_mode_check describing the type of option passed.
27826 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27827 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27828 arm_parse_cpu_option_name to match parameters against list of
27830 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27831 arm_arch_core_flags data structure.
27832 * config/arm/arm-cpu_cdata.h: Regenerated.
27834 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27836 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27838 (arm_print_hint_for_cpu_option): Likewise.
27839 (arm_print_hint_for_arch_option): Likewise.
27840 (arm_parse_cpu_option_name): Likewise.
27841 (arm_parse_arch_option_name): Likewise.
27842 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27843 of entries in the all_fpus list.
27844 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27845 (arm_parse_cpu_option_name): Declare.
27846 (arm_parse_arch_option_name): Declare.
27847 (arm_parse_option_features): Declare.
27848 (arm_intialize_isa): Declare.
27849 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27851 (gen_comm_data): ... here. Make definitions non-static.
27852 * config/arm/arm-cpu-data.h: Regenerated.
27853 * config/arm/arm-cpu-cdata.h: Regenerated.
27855 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27857 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27858 (cpu_arch_extension): New structure.
27859 (cpu_arch_option, arch_option, cpu_option): New structures.
27860 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27861 architecture types.
27862 (gen_data): Generate new format data tables.
27863 * config/arm/arm.c (cpu_tune): New structure.
27864 (cpu_option, processors): Delete.
27865 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27866 (arm_print_hint_for_cpu_option): ... this and ...
27867 (arm_print_hint_for_arch_option): ... this.
27868 (arm_parse_arch_cpu_name): Delete. Replace with ...
27869 (arm_parse_cpu_option_name): ... this and ...
27870 (arm_parse_arch_option_name): ... this.
27871 (arm_unrecognized_feature): Change type of target parameter to
27873 (arm_parse_arch_cpu_features): Delete. Replace with ...
27874 (arm_parse_option_features): ... this.
27875 (arm_configure_build_target): Rework to use new configuration data
27877 (arm_print_tune_info): Rework for new configuration data tables.
27878 * config/arm/arm-cpu-data.h: Regenerated.
27879 * config/arm/arm-cpu.h: Regenerated.
27881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27883 * Makefile.in (OBJS): Move sbitmap.o from here ...
27884 (OBJS-libcommon): ... to here.
27886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27888 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27889 (ISA_ALL_CRYPTO): New macro.
27890 (ISA_ALL_SIMD): New macro
27891 (ISA_ALL_FP): New macro.
27892 * config/arm/arm.c (fpu_bitlist): Update initializer.
27893 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27895 (arm9e): Add fpu. Add option for nofp
27896 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27897 (arm926ej-s, arm1026ej-s): Likewise.
27898 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27899 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27900 neon-fp16, neon-vfpv4, nofp and nosimd.
27901 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27902 (cortex-a8): Add fpu. Add option for nofp.
27903 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27904 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27905 (cortex-r4f): Add fpu.
27906 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27907 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27909 (cortex-r8): Likewise.
27910 (cortex-m4): Add fpu. Add option for nofp.
27911 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27912 (cortex-a17.cortex-a7): Likewise.
27913 (cortex-a32): Add fpu. Add options for crypto and nofp.
27914 (cortex-a35, cortex-a53): Likewise.
27915 (cortex-a57): Add fpu. Add option for crypto.
27916 (cortex-a72, cortex-a73): Likewise.
27917 (exynos-m1): Likewise.
27918 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27919 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27920 (cortex-m33): Add fpu. Add option for nofp.
27921 * config/arm/arm-cpu-cdata.h: Regenerated
27922 * config/arm/arm-cpu-data.h: Regenerated.
27924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27926 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27927 (armv5te, armv5tej): Likewise.
27928 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27929 (armv7): Add options fp and vfpv3-d16.
27930 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27931 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27933 (armv7ve): Likewise.
27934 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27935 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27936 (armv8-a): Add nocrypto option.
27937 (armv8.1-a, armv8.2-a): Likewise.
27938 (armv8-m.main): add options fp, fp.dp and nofp.
27940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27942 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27944 (armv8-a+crc): Delete.
27945 (armv8.1-a): Add options simd, crypto and nofp.
27946 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27947 (armv8.2-a+fp16): Delete.
27948 (armv8-m.main): Add option dsp.
27949 (armv8-m.main+dsp): Delete.
27950 (cortex-a8): Add fpu. Add nofp option.
27951 (cortex-a9): Add fpu. Add nofp and nosimd options.
27952 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27953 link to main cpu and architecture data structures.
27954 (gen_comm_data): Only put isa attributes from the main architecture
27956 (option): New statement for architecture and CPU entries.
27957 * arm.c (struct cpu_option): New structure.
27958 (struct processors): Add entry for options.
27959 (arm_unrecognized_feature): New function.
27960 (arm_parse_arch_cpu_name): Ignore any characters after the first
27962 (arm_parse_arch_cpu_feature): New function.
27963 (arm_configure_build_target): Separate out any CPU and architecture
27964 features and parse separately. Don't error out if -mfpu=auto is
27965 used with only an architecture string.
27966 (arm_print_asm_arch_directives): New function.
27967 (arm_file_start): Call it.
27968 * config/arm/arm-cpu-cdata.h: Regenerated.
27969 * config/arm/arm-cpu-data.h: Likewise.
27970 * config/arm/arm-tables.opt: Likewise.
27972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27974 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27975 assembler when it is not -mfpu=auto.
27977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27979 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27980 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27981 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27982 (ASM_CPU_SPEC): Rewrite.
27983 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27984 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27985 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27986 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27987 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27988 copied string is NUL-terminated. Also strip any characters prefixed
27990 (arm_rewrite_selected_arch): New function.
27991 (arm_rewrite_march): New function.
27993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27995 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27996 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27997 (march, mcpu, mtune): Convert to string-based options.
27998 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27999 (arm_parse_arch_cpu_name): New function.
28000 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28001 identify selected architecture or CPU.
28002 (arm_option_save): New function.
28003 (TARGET_OPTION_SAVE): Redefine.
28004 (arm_option_restore): Restore string options.
28005 (arm_option_print): Print string options.
28007 2017-06-16 Martin Sebor <msebor@redhat.com>
28009 PR tree-optimization/80933
28010 PR tree-optimization/80934
28011 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28012 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28013 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28014 (gimple_fold_builtin): Call them.
28016 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28018 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28019 as unlikely; update profile.
28021 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28023 * predict.c (force_edge_cold): Handle declaring edges impossible
28026 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28028 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28030 (try_unroll_loop_completely): Fix reporting.
28032 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28034 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28036 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28039 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28040 if given a non-constant argument for an intrinsic which requires a
28043 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28045 * profile.c (compare_freqs): New function.
28046 (branch_prob): Sort edge list.
28047 (find_spanning_tree): Assume that the list is priority sorted.
28049 2017-06-16 Richard Biener <rguenther@suse.de>
28051 PR tree-optimization/81090
28052 * passes.def (pass_record_bounds): Remove.
28053 * tree-pass.h (make_pass_record_bounds): Likewise.
28054 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28055 make_pass_record_bounds): Likewise.
28056 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28057 not free niter estimates at the beginning but at the end.
28058 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28060 2017-06-16 Richard Biener <rguenther@suse.de>
28062 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28063 initializer to workaround ICE in host GCC 4.8.
28065 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28067 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28069 (clone_inlined_nodes): Update.
28071 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28073 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28074 prefetch settings, and enable prefetching by default at -O3.
28076 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28078 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28079 Set flag_prefetch_loop_arrays according to tuning data.
28081 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28083 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28084 New tune structure.
28085 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28086 [Unrelated to main purpose of the patch] Place the pointer field last
28087 to enable type checking errors when tune structure are wrongly merged.
28088 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28089 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28090 (thunderx2t99_prefetch_tune): New tune constants.
28091 (tune_params *_tunings): Update all tunings (no functional change).
28092 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28093 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28094 from tunings structures.
28096 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28099 * ubsan.c (instrument_null): Add T argument, use it instead
28100 of computing it based on IS_LHS.
28101 (instrument_object_size): Likewise.
28102 (pass_ubsan::execute): Adjust instrument_null and
28103 instrument_object_size callers to pass gimple_get_lhs or
28104 gimple_assign_rhs1 result to it. Use instrument_null instead of
28105 calling get_base_address and instrument_mem_ref. Handle
28106 aggregate call arguments for object-size sanitization.
28108 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28110 PR tree-optimization/81089
28111 * tree-vrp.c (is_masked_range_test): Validate operands of
28114 2017-06-15 Martin Sebor <msebor@redhat.com>
28117 * dumpfile.c (dump_register): Avoid calling memset to initialize
28118 a class with a default ctor.
28119 * gcc.c (struct compiler): Remove const qualification.
28120 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28121 * hash-table.h: Ditto.
28122 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28124 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28125 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28127 * params.h (struct param_info): Make struct members non-const.
28128 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28129 with default initialization.
28130 * vec.h (vec_copy_construct, vec_default_construct): New helper
28132 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28133 with vec_copy_construct.
28134 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28135 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28136 * doc/invoke.texi (-Wclass-memaccess): Document.
28138 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28140 * emit-rtl.h (is_leaf): Update comment about local
28141 register allocator.
28143 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28146 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28147 for a variable to have a section before checking if the section has a
28149 Set section to.persistent if persistent attribute is set.
28150 Warn if .persistent attribute is used on an automatic variable.
28152 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28154 PR rtl-optimization/80474
28155 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28157 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28159 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28162 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28164 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28165 (casesi): Emit bounds checking as RTL.
28166 (casesi_internal_mips16_<mode>): Remove bounds checking.
28168 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28170 * config/xtensa/xtensa.c (xtensa_option_override): Append
28171 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28172 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28173 xtensa_doloop_hooks): Define unconditionally.
28174 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28176 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28177 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28178 for it in xtensa_option_override.
28179 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28180 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28182 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28184 * doc/cppopts.texi: Document '-' special value to -MF.
28186 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28188 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28189 (cortex_a53_fconst): Likewise.
28190 (cortex_a53_fpmul): Likewise.
28191 (cortex_a53_f_load_64): Likewise.
28192 (cortex_a53_f_load_many): Likewise.
28193 (cortex_a53_advsimd_alu): Likewise.
28194 (cortex_a53_advsimd_alu_q): Likewise.
28195 (cortex_a53_advsimd_mul): Likewise.
28196 (cortex_a53_advsimd_mul_q): Likewise.
28197 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28198 Add missing fmul, r2f_cvt and fconst cases.
28200 2017-06-14 Richard Biener <rguenther@suse.de>
28202 PR middle-end/81088
28203 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28205 (fold_binary_loc): When associating do not treat pre-existing
28206 TREE_OVERFLOW on literal constants as a reason to allow
28207 TREE_OVERFLOW on associated literal constants.
28209 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28211 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28212 (MASK_FEATURES): New macro.
28213 * config/sparc/sparc.c (sparc_option_override): Remove the special
28214 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28216 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28218 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28219 a division of 0 if non-call exceptions are enabled.
28221 2017-06-14 Andrew Pinski <apinski@cavium.com>
28222 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28225 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28226 Improve vector initialization code gen for only variable case.
28228 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28230 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28232 2017-06-14 Richard Biener <rguenther@suse.de>
28234 PR tree-optimization/81083
28235 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28238 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28240 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28241 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28242 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28243 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28244 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28245 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28247 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28249 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28250 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28252 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28254 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28256 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28258 * config/rs6000/t-rtems: Don't handle SPE.
28260 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28262 * config/rs6000/t-linux: Don't handle SPE.
28264 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28266 * config/rs6000/eabispe.h: Delete file.
28268 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28270 * config/rs6000/t-spe: Delete file.
28272 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28274 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28275 (rs6000_legitimate_offset_address_p): Return false for anything in
28276 V2SImode or V2SFmode.
28278 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28280 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28281 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28282 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28284 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28285 (rs6000_legitimate_offset_address_p): Ditto.
28286 (rs6000_emit_move): Ditto.
28287 (rs6000_init_builtins): Remove V4HI_type_node.
28289 2017-06-13 Martin Liska <mliska@suse.cz>
28292 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28293 (gate_asan): Likewise.
28294 * asan.h (asan_no_sanitize_address_p): Remove the function.
28295 (sanitize_flags_p): New function.
28296 * builtins.def: Fix coding style.
28297 * common.opt: Use renamed enum value.
28298 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28299 * doc/extend.texi: Document no_sanitize attribute.
28300 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28301 to SANITIZE_UNDEFINED_NONDEFAULT.
28302 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28303 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28304 Use sanitize_flags_p.
28305 * gimplify.c (gimplify_function_tree): Likewise.
28306 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28307 * opts.c (parse_no_sanitize_attribute): New function.
28308 (common_handle_option): Use renamed enum value.
28309 * opts.h (parse_no_sanitize_attribute): Declare.
28310 * tree.c (sanitize_flags_p): New function.
28311 * tree.h: Declared here.
28312 * tsan.c: Use sanitize_flags_p.
28313 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28314 (instrument_mem_ref): Likewise.
28315 (instrument_bool_enum_load): Likewise.
28316 (do_ubsan_in_current_function): Remove the function.
28317 (pass_ubsan::execute): Use sanitize_flags_p.
28318 * ubsan.h: Remove do_ubsan_in_current_function
28319 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28320 (dump_function_to_file): Use it here.
28322 2017-06-13 Martin Jambor <mjambor@suse.cz>
28324 PR tree-optimization/80803
28325 PR tree-optimization/81063
28326 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28327 (propagate_subaccesses_across_link): Enqueue subtree whenever
28328 necessary instead of relying on the caller.
28330 2017-06-13 Martin Jambor <mjambor@suse.cz>
28332 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28333 that have a first_link.
28334 (sort_and_splice_var_accesses): Do not check first_link before
28336 (subtree_mark_written_and_enqueue): Likewise.
28337 (propagate_all_subaccesses): Likewise and do not stop at first
28338 parent with a first_link.
28340 2017-06-13 Martin Jambor <mjambor@suse.cz>
28342 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28345 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28347 * match.pd: New pattern.
28349 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28351 * tree-vrp.c (is_masked_range_test): New function.
28352 (register_edge_assert_for): Determine ranges for
28355 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28357 PR tree-optimization/67328
28358 * fold-const.c (maskable_range_p): New function.
28359 (build_range_check): Generate bittests if possible.
28361 2017-06-13 Martin Liska <mliska@suse.cz>
28363 * gimple-pretty-print.c (dump_probability): Add new argument.
28364 (dump_edge_probability): Dump both probability and count.
28365 (dump_gimple_label): Likewise.
28366 (dump_gimple_bb_header): Likewise.
28368 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28371 * config/avr/avr-devices.c: Fix indentation.
28372 * config/avr/gen-avr-mmcu-specs.c: Dito.
28374 2017-06-13 Richard Biener <rguenther@suse.de>
28376 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28377 instead get vector type from stmt_info.
28378 (vectorizable_reduction): Adjust. Remove dead code.
28380 2017-06-13 Richard Biener <rguenther@suse.de>
28382 PR middle-end/81065
28383 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28384 case of C * (x * C2 + C3).
28385 (fold_addr_of_array_ref_difference): Properly fold index difference.
28387 2017-06-12 David S. Miller <davem@davemloft.net>
28390 * config/sparc/sparc.md (return expander): Emit frame blockage if
28391 function uses alloca.
28393 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28395 * combine.c (make_field_assignment): Check len rather than the mode
28396 precision when calling force_to_mode.
28398 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28400 Support multilibs and devices that see flash in RAM address range.
28403 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28404 (avr_mcu_t) <flash_pm_offset>: New field.
28405 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28406 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28407 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28408 (AVR_TINY_PM_OFFSET): Remove macro.
28409 * config/avr/avr.opt (-mshort-calls): New option.
28410 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28411 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28412 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28413 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28414 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28415 instead of avr_arch->have_jmp_call.
28416 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28417 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28418 avr_arch->flash_pm_offset to define.
28419 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28420 new field flash_pm_offset. Add entry for avrxmega3.
28421 (avr_texinfo): Add entry for avrxmega3.
28422 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28423 attiny212, attiny214,
28424 attiny412, attiny414, attiny416, attiny417,
28425 attiny814, attiny816, attiny817,
28426 attiny1614, attiny1616, attiny1617,
28427 attiny3214, attiny3216, attiny3217.
28428 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28429 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28430 (avr_print_operand_address) [AVR_TINY]: Same.
28431 (avr_asm_init_sections) <readonly_data_section>: Only patch
28432 callback if avr_arch->flash_pm_offset = 0.
28433 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28434 for rodata if avr_arch->flash_pm_offset != 0.
28435 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28436 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28437 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28438 (m_options): Append opt_rcall.
28439 (m_dirnames): Append dir_rcall.
28440 * config/avr/t-multilib: Regenerate.
28442 * configure.ac [target=avr]: Check whether avrxmega3 default
28443 linker description file works as needed.
28444 * configure: Regenerate.
28445 * doc/avr-mmcu.texi: Regenerate.
28446 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28447 <__AVR_ARCH__>: Document avrxmega3 and 103.
28448 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28449 <__AVR_SHORT_CALLS__>: Document it.
28450 <__AVR_PM_BASE_ADDRESS__>: Document it.
28451 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28452 (AVR Variable Attributes) <progmem>: Document this is
28453 not needed for avrxmega3.
28454 (AVR Named Address Spaces) <__flash>: Dito.
28456 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28458 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28460 2017-06-12 Doug Rupp <rupp@adacore.com>
28462 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28463 Append vxworks-stdint.h to the tm_file list.
28464 * config/vxworks-stdint.h: New file.
28466 2017-06-12 Martin Liska <mliska@suse.cz>
28468 PR tree-optimization/81041
28469 * tree-profile.c (gimple_gen_ic_func_profiler):
28470 Create an extra BB in profile-generate
28471 (gimple_gen_time_profiler): Likewise.
28473 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28475 PR tree-optimization/81003
28476 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28477 (update_range_test): Use it instead of force_gimple_operand_gsi.
28479 2017-06-12 Richard Biener <rguenther@suse.de>
28481 PR tree-optimization/81053
28482 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28483 with backedge value not defined in loop. Simplify def stmt
28486 2017-06-11 Tom de Vries <tom@codesourcery.com>
28489 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28491 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28492 nvptx_cannot_force_const_mem.
28494 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28496 * opts.c (finish_options): Move test for flag_split_stack after
28497 it has been initialized.
28499 2017-06-11 Jason Merrill <jason@redhat.com>
28501 * tree.h (id_equal): New.
28502 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28503 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28504 instead of strcmp of IDENTIFIER_POINTER.
28506 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28508 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28509 (mark_all_inlined_calls_cdtor): Fix formating.
28510 (inline_transform): Rescale profile before inlining.
28512 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28514 * cgraph.h (cgraph_edge::clone): Update prototype.
28515 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28516 (cgraph_node::create_clone): Update.
28517 (cgraph_node::create_version_clone): Update.
28518 * tree-inline.c (copy_bb): Update.
28519 (expand_call_inline): Update.
28521 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28523 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28524 factored out from ...
28525 (rs6000_emit_prologue): ... here.
28527 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28529 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28530 factored out from ...
28531 (rs6000_emit_prologue): ... here.
28533 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28535 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28537 (handle_missing_profiles): Fix computation of tp_first_run.
28538 (counts_to_freqs): Do not touch freqs when count is 0.
28540 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28542 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28545 2017-06-10 Tom de Vries <tom@codesourcery.com>
28547 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28548 attributes): Document signal effective target.
28550 2017-06-10 Tom de Vries <tom@codesourcery.com>
28552 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28553 Document effective target stack_size.
28555 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28557 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28558 to the edit_context if they can be auto-applied.
28560 2017-06-9 Ian Lance Taylor <iant@golang.org>
28562 * opts.c (finish_options): If -fsplit-stack, disable implicit
28563 -forder-blocks-and-partition.
28564 * doc/invoke.texi (Optimize Options): Document that when using
28565 -fsplit-stack -forder-blocks-and-partition is not implicitly
28568 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28570 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28571 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28572 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28573 * builtins.def (abort, trap, unreachable): Declare cold.
28574 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28575 * tree-core.h (ECF_COLD): New.
28576 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28577 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28579 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28581 * predict.c (unlikely_executed_stmt_p): Cleanup.
28583 2017-06-09 Richard Biener <rguenther@suse.de>
28585 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28586 model if the ref is always written to.
28588 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28590 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28592 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28594 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28597 2017-06-09 Tom de Vries <tom@codesourcery.com>
28600 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28601 "target cannot support label values" when encountering LABEL_REF.
28603 2017-06-09 Martin Liska <mliska@suse.cz>
28605 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28606 (gimple_gen_ic_func_profiler): Emit direct comparison
28607 of __gcov_indirect_call_callee with NULL.
28608 (gimple_gen_time_profiler): Change probability from
28609 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28611 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28613 * profile.c (edge_gcov_counts): Turn to pointer.
28614 (compute_branch_probabilities, compute_branch_probabilities): Update.
28615 (branch_prob): Do not clear edge_gcov_count.
28616 * profile.h (edge_gcov_counts): Turn to pointer.
28617 (edge_gcov_count): Update.
28619 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28621 * gimple.h (gimple_check_failed): Mark cold.
28623 2017-06-09 Richard Biener <rguenther@suse.de>
28625 PR tree-optimization/66623
28626 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28627 refactor check_reduction into two parts, properly computing
28628 whether we have to check reduction validity for outer loop
28631 2017-06-09 Richard Biener <rguenther@suse.de>
28633 PR tree-optimization/79483
28634 * graphite-scop-detection.c (order): New global.
28635 (get_order): Compute bb to order mapping that satisfies code
28636 generation constraints.
28637 (cmp_pbbs): New helper.
28638 (build_scops): Start domwalk at entry block, sort generated
28641 2017-06-09 Richard Biener <rguenther@suse.de>
28643 PR middle-end/81007
28644 * ipa-polymorphic-call.c
28645 (ipa_polymorphic_call_context::restrict_to_inner_class):
28646 Skip FIELD_DECLs with error_mark_node type.
28647 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28650 2017-06-09 Martin Liska <mliska@suse.cz>
28652 * predict.c (struct branch_predictor): New struct.
28653 (test_prediction_value_range): New test.
28654 (predict_c_tests): New function.
28655 * selftest-run-tests.c (selftest::run_tests): Run the function.
28656 * selftest.h: Declare new tests.
28658 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28661 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28662 gen_add3_insn did not fail.
28663 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28664 r0, construct that number in a temporary reg and add that reg to r0.
28665 If asked to put the result in r0 as well, fail.
28667 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28669 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28670 for early expansion of vec_eqv.
28672 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28674 PR middle-end/81005
28675 * ubsan.c (instrument_null): Avoid pointless code temporary.
28676 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28678 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28682 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28685 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28688 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28690 * predict.c (unlikely_executed_edge_p): Move ahead.
28691 (probably_never_executed_edge_p): Use it.
28693 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28695 PR middle-end/79988
28696 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28697 gimple_call_builtin_p call.
28699 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28701 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28702 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28703 rtl_check_failed_type2, rtl_check_failed_code1,
28704 rtl_check_failed_code2, rtl_check_failed_code_mode,
28705 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28706 rtvec_check_failed_bounds, rtl_check_failed_flag,
28707 _fatal_insn_not_found, _fatal_insn): Likewise.
28708 * tree.h (tree_contains_struct_check_failed,
28709 tree_check_failed, tree_not_check_failed,
28710 tree_class_check_failed, tree_range_check_failed,
28711 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28712 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28713 tree_operand_check_failed, omp_clause_check_failed,
28714 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28717 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28719 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28720 flag_branch_probabilities.
28721 * ipa-inline.c (edge_badness): Likewise.
28722 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28723 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28724 * predict.c (maybe_hot_frequency_p): Likewise.
28725 (probably_never_executed): Likewise.
28726 * sched-ebb.c (schedule_ebbs): Likewise.
28727 * sched-rgn.c (find_single_block_region): Likewise.
28728 * tracer.c (tail_duplicate): Likewise.
28730 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28732 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28733 longer requires x_flag_profile_use.
28735 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28737 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28738 instead of flag_reorder_blocks_and_partition.
28739 * dbxout.c (dbxout_function_end): Likewise.
28740 * dwarf2out.c (gen_subprogram_die): Likewise.
28741 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28742 * hw-doloop.c (reorg_loops): Likewise.
28743 * varasm.c (assemble_start_function,
28744 assemble_end_function): Likewise.
28745 (decide_function_section): Do not check for
28746 flag_reorder_blocks_and_partition.
28748 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28750 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28752 (chkp_get_hard_register_fake_addr_expr): Ditto.
28753 (chkp_build_addr_expr): Add check for hard reg case.
28754 (chkp_parse_array_and_component_ref): Ditto.
28755 (chkp_find_bounds_1): Ditto.
28756 (chkp_process_stmt): Don't generate bounds store for
28759 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28761 * predict.c (maybe_hot_bb_p): Do not check profile status.
28762 (maybe_hot_edge_p): Likewise.
28763 (probably_never_executed): Check for zero counts even if profile
28765 (unlikely_executed_edge_p): New function.
28766 (unlikely_executed_stmt_p): New function.
28767 (unlikely_executed_bb_p): New function.
28768 (set_even_probabilities): Use unlikely predicates.
28769 (combine_predictions_for_bb): Likewise.
28770 (predict_paths_for_bb): Likewise.
28771 (predict_paths_leading_to_edge): Likewise.
28772 (determine_unlikely_bbs): New function.
28773 (estimate_bb_frequencies): Use it.
28774 (compute_function_frequency): Use zero counts even if profile is
28776 * profile-count.h: Fix typo.
28778 2017-08-08 Julia Koval <julia.koval@intel.com>
28780 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28781 _mm512_mask_cvtsepi16_storeu_epi8,
28782 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28783 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28784 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28785 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28786 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28787 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28788 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28789 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28790 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28791 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28792 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28793 __builtin_ia32_pmovuswb256mem_mask,
28794 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28795 __builtin_ia32_pmovwb512mem_mask): New builtins.
28797 2017-08-08 Julia Koval <julia.koval@intel.com>
28799 PR target/73350,80862
28800 * config/i386/subst.md (round): Fix round pattern.
28801 * config/i386/i386.c (ix86_erase_embedded_rounding):
28802 Fix erasing rounding for the fixed pattern.
28804 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28806 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28808 2017-06-08 Martin Liska <mliska@suse.cz>
28810 PR gcov-profile/80911
28811 * gcov.c (block_info::block_info): New constructor.
28813 2017-06-07 Carl Love <cel@us.ibm.com>
28815 * config/rs6000/rs6000-c: The return type of the following
28816 built-in functions was implemented as int not long long. Fix sign
28817 of return value for the unsigned version of vec_mulo and vec_mule.
28818 vector unsigned long long vec_bperm (vector unsigned long long,
28819 vector unsigned char)
28820 vector signed long long vec_mule (vector signed int,
28822 vector unsigned long long vec_mule (vector unsigned int,
28823 vector unsigned int)
28824 vector signed long long vec_mulo (vector signed int,
28826 vector unsigned long long vec_mulo (vector unsigned int,
28827 vector unsigned int)
28828 * doc/extend.texi: Fix the documentation for the built-in
28831 2017-06-07 Carl Love <cel@us.ibm.com>
28834 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28837 2017-06-07 Carl Love <cel@us.ibm.com>
28839 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28840 support, Generate doublehv for signed int/float for BE case only.
28842 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28844 * doc/invoke.texi (mcx16): Rewrite.
28846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28848 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28849 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28850 *mov<mode>_softfloat, and an anonymous splitter): Use
28851 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28853 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28855 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28857 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28858 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28859 (rs6000_debug_reg_global): Adjust.
28860 (rs6000_init_hard_regno_mode_ok): Adjust.
28861 (rs6000_dbx_register_number): Adjust.
28862 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28863 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28864 Remove SPE_ACC and SPEFSCR.
28865 (REG_ALLOC_ORDER): Ditto.
28866 (FRAME_POINTER_REGNUM): Change to 111.
28867 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28868 (REG_CLASS_NAMES): Ditto.
28869 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28870 (REGISTER_NAMES): Ditto.
28871 (ADDITIONAL_REG_NAMES): Ditto.
28872 (rs6000_reg_names): Ditto.
28873 * config/rs6000/rs6000.md: Renumber some register number
28876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28878 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28880 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28881 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28883 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28884 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28885 Delete the SPE high registers.
28886 (REG_ALLOC_ORDER): Ditto.
28887 (enum reg_class): Remove SPE_HIGH_REGS.
28888 (REG_CLASS_NAMES): Ditto.
28889 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28890 (REGISTER_NAMES): Ditto.
28891 (rs6000_reg_names): Ditto.
28892 * doc/tm.texi.in: Remove SPE as example.
28893 * doc/tm.texi: Regenerate.
28895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28897 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28898 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28899 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28900 * config/rs6000/rs6000.md (type): Remove "brinc".
28902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28904 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28905 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28906 * config/rs6000/linuxspe.h: Delete file.
28907 * config/rs6000/rs6000.md: Don't include spe.md.
28908 * config/rs6000/spe.h: Delete file.
28909 * config/rs6000/spe.md: Delete file.
28910 * config/rs6000/t-rs6000: Remove spe.md.
28912 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28914 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28915 (reg_or_none500mem_operand): Delete.
28916 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28917 instead of reg_or_none500mem_operand.
28919 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28921 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28922 handling of SPE flags.
28923 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28925 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28927 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28929 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28930 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28931 paired_divv2sf3): Similar.
28932 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28933 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28934 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28936 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28937 Rename the paired_* instruction patterns.
28938 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28940 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28941 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28942 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28943 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28944 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28945 PAIRED_VECTOR_MODE.
28946 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28947 (spe_func_has_64bit_regs_p): Delete.
28948 (spe_expand_predicate_builtin): Delete.
28949 (spe_expand_evsel_builtin): Delete.
28950 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28951 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28952 (invalid_e500_subreg): Delete.
28953 (rs6000_legitimize_address): Always force_reg op2 as well, for
28954 paired single memory accesses.
28955 (rs6000_member_type_forces_blk): Delete.
28956 (rs6000_spe_function_arg): Delete.
28957 (rs6000_expand_unop_builtin): Delete SPE handling.
28958 (rs6000_expand_binop_builtin): Ditto.
28959 (spe_expand_stv_builtin): Delete.
28960 (bdesc_2arg_spe): Delete.
28961 (spe_expand_builtin): Delete.
28962 (spe_expand_predicate_builtin): Delete.
28963 (spe_expand_evsel_builtin): Delete.
28964 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28965 (spe_init_builtins): Delete.
28966 (spe_func_has_64bit_regs_p): Delete.
28967 (savres_routine_name): Delete "info" parameter. Adjust callers.
28968 (rs6000_emit_stack_reset): Ditto.
28969 (rs6000_dwarf_register_span): Delete.
28970 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28971 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28972 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28974 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28976 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28977 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28978 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28979 mulv2sf3, divv2sf3): Delete expanders.
28981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28983 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28985 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28987 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28988 * config/rs6000/rs6000.c: Ditto.
28990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28992 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28993 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28994 comparison_operator.
28996 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28998 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28999 * config/rs6000/rs6000.opt: Ditto.
29000 * config/rs6000/t-rtems: Ditto.
29002 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29004 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29005 TARGET_E500_SINGLE by 0, simplify.
29006 * config/rs6000/rs6000.c: Ditto.
29007 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29008 (spe_build_register_parallel): Delete.
29009 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29010 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29011 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29012 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29013 (E500_CONVERT): Delete.
29014 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29018 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29019 * config/rs6000/dfp.md: Ditto.
29020 (negdd2, *negdd2_fpr): Merge.
29021 (absdd2, *absdd2_fpr): Merge.
29022 (negtd2, *negtd2_fpr): Merge.
29023 (abstd2, *abstd2_fpr): Merge.
29024 * config/rs6000/e500.h: Delete file.
29025 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29026 TARGET_FPRS by 1 and simplify.
29027 * config/rs6000/rs6000-c.c: Ditto.
29028 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29029 TARGET_DF_SPE by 0.
29030 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29032 * config/rs6000/rs6000.md: Ditto.
29033 (floatdidf2, *floatdidf2_fpr): Merge.
29034 (move_from_CR_gt_bit): Delete.
29035 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29036 (E500_CR_IOR_COMPARE): Delete.
29037 (All patterns that require !TARGET_FPRS): Delete.
29038 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29040 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29042 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29044 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29046 * graphds.c (add_edge): Intitialize edge's attached data.
29047 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29048 pointer parameter. Call pointed function on each edge during
29049 graph traversing. Skip traversing the edge when the function
29051 (graphds_dfs, graphds_scc): Ditto.
29052 (for_each_edge): New parameter. Pass the new parameter to callback
29054 * graphds.h (skip_edge_callback): New function pointer type.
29055 (graphds_dfs, graphds_scc): New function pointer parameter.
29056 (graphds_edge_callback, for_each_edge): New parameter.
29058 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29060 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29061 out code checking if runtime alias check is possible to below ...
29062 Call the new function.
29063 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29064 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29066 2017-06-07 Marek Polacek <polacek@redhat.com>
29069 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29070 TYPE_OVERFLOW_WRAPS check.
29072 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29074 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29075 if versioning is required.
29076 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29077 peeling with the check for versioning.
29079 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29081 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29082 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29083 Set true to new parameter if new ssa variable is defined.
29084 (vect_gen_vector_loop_niters): Refactor. Set range information
29085 for the new vector loop bound variable.
29086 (vect_do_peeling): Ditto.
29088 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29090 * tree-affine.c (ssa.h): Include header file.
29091 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29092 has wrapping overflow behavior.
29094 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29096 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29098 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29100 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29101 (tree_to_aff_combination): ... here.
29103 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29105 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29106 reg_pressure model function.
29107 (ivopts_global_cost_for_size): Delete.
29108 (determine_set_costs, iv_ca_recount_cost): Call new model function
29109 ivopts_estimate_reg_pressure.
29111 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29113 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29114 expensive than udiv. Remove floating point cases from mod.
29116 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29118 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29119 Increase idiv cost.
29121 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29123 * config/aarch64/aarch64.md
29124 (copysignsf3): Fix mask generation.
29126 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29128 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29130 (class dump_manager): Add register_dumps method.
29131 * dumpfile.c: Include langhooks.h.
29132 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29133 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29134 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29135 (dump_manager::dump_register): Start with 512 entries instead of 32.
29136 (dump_manager::register_dumps): New method.
29137 * toplev.c (general_init): Instead of invoking register_dumps
29138 langhook, invoke register_dumps method on the dump manager.
29139 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29142 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29144 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29145 Say that # requires an associated define_split to exist, and that
29146 the define_split must be suitable for use after register allocation.
29148 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29150 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29151 (compute_outgoing_frequencies): Also initialize zero counts.
29152 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29153 around loops; preserve more of profile when nothing changes.
29155 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29157 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29159 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29160 * config/arm/arm-cpu-cdata.h: Regenerate.
29161 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29162 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29163 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29164 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29165 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29167 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29169 * config/arm/t-rmprofile: Likewise.
29170 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29172 2017-06-06 David S. Miller <davem@davemloft.net>
29175 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29176 blockage if function uses alloca.
29178 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29180 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29181 New "uid" fields to hold pretty-print IDs of group and ref.
29182 Memory references are now identified as <group_id>:<ref_id>
29183 instead of using [random] addresses.
29184 (dump_mem_details): Simplify, no functional change.
29185 (dump_mem_ref): Simplify and make output more concise.
29186 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29187 (find_or_create_group): Initialize group uid.
29188 (record_ref): Initialize ref uid. Improve debug output.
29189 (prune_group_by_reuse, should_issue_prefetch_p,)
29190 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29191 (mark_nontemporal_store, determine_loop_nest_reuse):
29192 Improve debug output.
29194 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29196 * dbgcnt.def (prefetch): New debug counter.
29197 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29198 (schedule_prefetches): Stop issueing prefetches if debug counter
29201 2017-06-06 Tom de Vries <tom@codesourcery.com>
29203 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29204 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29206 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29208 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29209 Use aarch64_reg_or_zero predicate for operand 4.
29210 (aarch64_compare_and_swap<mode> define_insn_and_split):
29211 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29212 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29214 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29216 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29217 (arm_compute_save_core_reg_mask): This.
29218 (thumb1_compute_save_reg_mask): Rename into ...
29219 (thumb1_compute_save_core_reg_mask): This.
29220 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29221 (arm_compute_frame_layout): Likewise.
29223 2017-06-06 Richard Biener <rguenther@suse.de>
29225 PR tree-optimization/80974
29226 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29227 keep or clear leaders SSA info.
29229 2017-06-06 Tom de Vries <tom@codesourcery.com>
29231 * config/nvptx/nvptx.c (split_mode_p): New function.
29232 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29234 2017-06-06 Tom de Vries <tom@codesourcery.com>
29236 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29238 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29241 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29244 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29246 * shrink-wrap.c (handle_simple_exit): Update profile.
29247 (try_shrink_wrapping): Upate profile.
29249 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29251 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29252 (tree_guess_outgoing_edge_probabilities): New.
29253 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29254 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29256 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29258 * ipa-split.c (split_function): Initialize return bb profile.
29260 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29262 * profile.c (compute_branch_probabilities): Also initialize
29263 EXIT_BLOCK profile.
29265 2017-06-06 Richard Biener <rguenther@suse.de>
29267 PR tree-optimization/80928
29268 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29269 (vect_analyze_loop_operations): Properly guard analysis for
29271 (vect_transform_loop): Likewise.
29272 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29273 (vect_model_induction_cost): Do not cost for pure SLP.
29274 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29275 of induction in inner loop vectorization.
29276 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29277 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29278 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29280 (vect_analyze_slp_cost_1): Cost induction.
29281 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29282 (vect_get_slp_vect_defs): Likewise.
29283 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29284 (vect_transform_stmt): Handle SLP reductions.
29285 * tree-vectorizer.h (vectorizable_induction): Adjust.
29287 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29289 * config/rs6000/rs6000.c (make_resolver_func): Update
29290 init_lowered_empty_function call.
29292 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29294 * doc/invoke.texi: Document the -fprofile-abs-path option.
29295 * common.opt (fprofile-abs-path): New option.
29296 * gcov-io.h (gcov_write_filename): Declare.
29297 * gcov-io.c (gcov_write_filename): New function.
29298 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29299 * profile.c (output_location): Likewise.
29301 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29303 * shring-wrap.c: Revert accidental commit.
29305 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29307 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29309 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29311 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29313 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29314 profile in callgraph edge.
29315 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29316 (apply_scale): Likewise.
29317 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29318 Also scale profile when inlining function with zero profile.
29319 (initialize_cfun): Update exit block profile even when it is zero.
29320 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29321 when profile is read.
29323 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29325 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29326 (CLONE_*): New constants to define the processors we can generate
29327 code for with the target_clone attribute.
29328 (rs6000_clone_map): New array to identify which clone processors
29329 the current program is running on.
29330 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29331 target_clone attribute.
29332 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29333 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29334 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29335 (cpu_expand_builtin): Add support for target_clone attribute.
29336 (rs6000_valid_attribute_p): Allow "default" attribute.
29337 (get_decl_name): New debug function to simplify printing the
29338 current function name in debugging statements.
29339 (rs6000_clone_priority): New functions to support the target_clone
29340 attribute, and be able to generate code to switch between ISA 2.05
29341 through ISA 3.0 (power6 through power9).
29342 (rs6000_compare_version_priority): Likewise.
29343 (rs6000_get_function_versions_dispatcher): Likewise.
29344 (make_resolver_func): Likewise.
29345 (add_condition_to_bb): Likewise.
29346 (dispatch_function_versions): Likewise.
29347 (rs6000_generate_version_dispatcher_body): Likewise.
29348 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29349 (fusion_gpr_load_p): Fix a spacing issue.
29350 * doc/extend.texi (Common Function Attributes): Document that the
29351 PowerPC supports the target_clone attribute.
29353 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29355 * config/arm/arm.h: explain F symbol found in description of ARM
29356 register allocation in its legend.
29358 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29360 * config/mips/frame-header-opt.c: Include profile-count.h.
29361 * config/riscv/riscv.c: Include profile-count.h
29363 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29365 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29367 (sm_set_flag_if_changed): Add bbs field.
29368 (execute_sm_if_changed_flag_set): Pass BBS.
29369 (execute_sm): Update.
29371 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29373 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29376 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29378 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29379 (peephole2): New peephole2 to emit the above.
29380 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29382 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29384 * config/aarch64/aarch64.c (define_peephole2 above
29385 *sub_<shift>_<mode>): New peephole.
29387 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29389 * config/i386/i386.c (make_resolver_func): Update.
29390 * Makefile.in: Add profile-count.h and profile-count.o
29391 * auto-profile.c (afdo_indirect_call): Update to new API.
29392 (afdo_set_bb_count): Update.
29393 (afdo_propagate_edge): Update.
29394 (afdo_propagate_circuit): Update.
29395 (afdo_calculate_branch_prob): Update.
29396 (afdo_annotate_cfg): Update.
29397 * basic-block.h: Include profile-count.h
29398 (struct edge_def): Turn count to profile_count.
29399 (struct basic_block_def): Likewie.
29400 (REG_BR_PROB_BASE): Move to profile-count.h
29401 (RDIV): Move to profile-count.h
29402 * bb-reorder.c (max_entry_count): Turn to profile_count.
29403 (find_traces): Update.
29404 (rotate_loop):Update.
29405 (connect_traces):Update.
29406 (sanitize_hot_paths):Update.
29407 * bt-load.c (migrate_btr_defs): Update.
29408 * cfg.c (RDIV): Remove.
29409 (init_flow): Use alloc_block.
29410 (alloc_block): Uninitialize count.
29411 (unchecked_make_edge): Uninitialize count.
29412 (check_bb_profile): Update.
29413 (dump_edge_info): Update.
29414 (dump_bb_info): Update.
29415 (update_bb_profile_for_threading): Update.
29416 (scale_bbs_frequencies_int): Update.
29417 (scale_bbs_frequencies_gcov_type): Update.
29418 (scale_bbs_frequencies_profile_count): New.
29419 * cfg.h (update_bb_profile_for_threading): Update.
29420 (scale_bbs_frequencies_profile_count): Declare.
29421 * cfgbuild.c (compute_outgoing_frequencies): Update.
29422 (find_many_sub_basic_blocks): Update.
29423 * cfgcleanup.c (try_forward_edges): Update.
29424 (try_crossjump_to_edge): Update.
29425 * cfgexpand.c (expand_gimple_tailcall): Update.
29426 (construct_exit_block): Update.
29427 * cfghooks.c (verify_flow_info): Update.
29428 (dump_bb_for_graph): Update.
29429 (split_edge): Update.
29430 (make_forwarder_block): Update.
29431 (duplicate_block): Update.
29432 (account_profile_record): Update.
29433 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29434 (get_estimated_loop_iterations): Update.
29435 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29436 (single_likely_exit): Update.
29437 * cfgloopmanip.c (scale_loop_profile): Update.
29439 (set_zero_probability): Update.
29440 (lv_adjust_loop_entry_edge): Update.
29441 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29442 (purge_dead_edges): Update.
29443 (rtl_account_profile_record): Update.
29444 * cgraph.c (cgraph_node::create): Uninitialize count.
29445 (symbol_table::create_edge): Uninitialize count.
29446 (cgraph_update_edges_for_call_stmt_node): Update.
29447 (cgraph_edge::dump_edge_flags): Update.
29448 (cgraph_node::dump): Update.
29449 (cgraph_edge::maybe_hot_p): Update.
29450 * cgraph.h: Include profile-count.h
29451 (create_clone), create_edge, create_indirect_edge): Update.
29452 (cgraph_node): Turn count to profile_count.
29453 (cgraph_edge0: Likewise.
29454 (make_speculative, clone): Update.
29455 (create_edge): Update.
29456 (init_lowered_empty_function): Update.
29457 * cgraphclones.c (cgraph_edge::clone): Update.
29458 (duplicate_thunk_for_node): Update.
29459 (cgraph_node::create_clone): Update.
29460 * cgraphunit.c (cgraph_node::analyze): Update.
29461 (cgraph_node::expand_thunk): Update.
29462 * final.c (dump_basic_block_info): Update.
29463 * gimple-streamer-in.c (input_bb): Update.
29464 * gimple-streamer-out.c (output_bb): Update.
29465 * graphite.c (print_global_statistics): Update.
29466 (print_graphite_scop_statistics): Update.
29467 * hsa-brig.c: Include basic-block.h.
29468 * hsa-dump.c: Include basic-block.h.
29469 * hsa-gen.c (T sum_slice): Update.
29470 (convert_switch_statements):Update.
29471 * hsa-regalloc.c: Include basic-block.h.
29472 * ipa-chkp.c (chkp_produce_thunks): Update.
29473 * ipa-cp.c (struct caller_statistics): Update.
29474 (init_caller_stats): Update.
29475 (gather_caller_stats): Update.
29476 (ipcp_cloning_candidate_p): Update.
29477 (good_cloning_opportunity_p): Update.
29478 (get_info_about_necessary_edges): Update.
29479 (dump_profile_updates): Update.
29480 (update_profiling_info): Update.
29481 (update_specialized_profile): Update.
29482 (perhaps_add_new_callers): Update.
29483 (decide_about_value): Update.
29484 (ipa_cp_c_finalize): Update.
29485 * ipa-devirt.c (struct odr_type_warn_count): Update.
29486 (struct decl_warn_count): Update.
29487 (struct final_warning_record): Update.
29488 (possible_polymorphic_call_targets): Update.
29489 (ipa_devirt): Update.
29490 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29491 * ipa-icf.c (sem_function::merge): Update.
29492 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29493 * ipa-inline.c (compute_uninlined_call_time): Update.
29494 (compute_inlined_call_time): Update.
29495 (want_inline_small_function_p): Update.
29496 (want_inline_self_recursive_call_p): Update.
29497 (edge_badness): Update.
29498 (lookup_recursive_calls): Update.
29499 (recursive_inlining): Update.
29500 (inline_small_functions): Update.
29501 (dump_overall_stats): Update.
29502 (dump_inline_stats): Update.
29503 * ipa-profile.c (ipa_profile_generate_summary): Update.
29504 (ipa_propagate_frequency): Update.
29505 (ipa_profile): Update.
29506 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29507 * ipa-utils.c (ipa_merge_profiles): Update.
29508 * loop-doloop.c (doloop_modify): Update.
29509 * loop-unroll.c (report_unroll): Update.
29510 (unroll_loop_runtime_iterations): Update.
29511 * lto-cgraph.c (lto_output_edge): Update.
29512 (lto_output_node): Update.
29513 (input_node): Update.
29514 (input_edge): Update.
29515 (merge_profile_summaries): Update.
29516 * lto-streamer-in.c (input_cfg): Update.
29517 * lto-streamer-out.c (output_cfg): Update.
29518 * mcf.c (create_fixup_graph): Update.
29519 (adjust_cfg_counts): Update.
29520 (sum_edge_counts): Update.
29521 * modulo-sched.c (sms_schedule): Update.
29522 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29523 * predict.c (maybe_hot_count_p): Update.
29524 (probably_never_executed): Update.
29525 (dump_prediction): Update.
29526 (combine_predictions_for_bb): Update.
29527 (propagate_freq): Update.
29528 (handle_missing_profiles): Update.
29529 (counts_to_freqs): Update.
29530 (rebuild_frequencies): Update.
29531 (force_edge_cold): Update.
29532 * predict.h: Include profile-count.h
29533 (maybe_hot_count_p, counts_to_freqs): UPdate.
29534 * print-rtl-function.c: Do not include cfg.h
29535 * print-rtl.c: Include basic-block.h
29536 * profile-count.c: New file.
29537 * profile-count.h: New file.
29538 * profile.c (is_edge_inconsistent): Update.
29539 (correct_negative_edge_counts): Update.
29540 (is_inconsistent): Update.
29541 (set_bb_counts): Update.
29542 (read_profile_edge_counts): Update.
29543 (compute_frequency_overlap): Update.
29544 (compute_branch_probabilities): Update; Initialize and deinitialize
29546 (branch_prob): Update.
29547 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29548 (edge_gcov_count): New.
29549 (bb_gcov_count): New.
29550 * shrink-wrap.c (try_shrink_wrapping): Update.
29551 * tracer.c (better_p): Update.
29552 * trans-mem.c (expand_transaction): Update.
29553 (ipa_tm_insert_irr_call): Update.
29554 (ipa_tm_insert_gettmclone_call): Update.
29555 * tree-call-cdce.c: Update.
29556 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29557 (gimple_duplicate_sese_tail): Update.
29558 (gimple_account_profile_record): Update.
29559 (execute_fixup_cfg): Update.
29560 * tree-inline.c (copy_bb): Update.
29561 (copy_edges_for_bb): Update.
29562 (initialize_cfun): Update.
29563 (freqs_to_counts): Update.
29564 (copy_cfg_body): Update.
29565 (expand_call_inline): Update.
29566 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29567 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29568 (try_unroll_loop_completely): Update.
29569 (try_peel_loop): Update.
29570 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29571 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29572 * tree-ssa-loop-split.c (connect_loops): Update.
29573 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29574 * tree-ssa-reassoc.c (branch_fixup): Update.
29575 * tree-ssa-tail-merge.c (replace_block_by): Update.
29576 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29577 (compute_path_counts): Update.
29578 (update_profile): Update.
29579 (recompute_probabilities): Update.
29580 (update_joiner_offpath_counts): Update.
29581 (estimated_freqs_path): Update.
29582 (freqs_to_counts_path): Update.
29583 (clear_counts_path): Update.
29584 (ssa_fix_duplicate_block_edges): Update.
29585 (duplicate_thread_path): Update.
29586 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29587 (struct switch_conv_info): Update.
29588 * tree-tailcall.c (decrease_profile): Update.
29589 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29590 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29591 * value-prof.c (check_counter): Update.
29592 (gimple_divmod_fixed_value): Update.
29593 (gimple_mod_pow2): Update.
29594 (gimple_mod_subtract): Update.
29595 (gimple_ic_transform): Update.
29596 (gimple_stringop_fixed_value): Update.
29597 * value-prof.h (gimple_ic): Update.
29599 2017-06-02 Carl Love <cel@us.ibm.com>
29601 * config/rs6000/rs6000-c: Add support for built-in functions
29602 vector double vec_doublee (vector signed int);
29603 vector double vec_doublee (vector unsigned int);
29604 vector double vec_doublee (vector float);
29605 vector double vec_doubleh (vector signed int);
29606 vector double vec_doubleh (vector unsigned int);
29607 vector double vec_doubleh (vector float);
29608 vector double vec_doublel (vector signed int);
29609 vector double vec_doublel (vector unsigned int);
29610 vector double vec_doublel (vector float);
29611 vector double vec_doubleo (vector signed int);
29612 vector double vec_doubleo (vector unsigned int);
29613 vector double vec_doubleo (vector float);.
29614 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29615 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29617 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29618 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29619 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29621 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29622 vec_doublel, vec_doubleh.
29623 * doc/extend.texi: Update the built-in documentation file for the
29624 new built-in functions.
29626 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29629 * ipa-inline-analysis.c (free_growth_caches): Set
29630 edge_removal_hook_holder to NULL after removing it.
29632 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29634 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29635 comparision with zero.
29637 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29638 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29639 for early expansion of vec_min and vec_max builtins.
29640 (builtin_function_type): Add min/max unsigned variants to those
29641 identified as having unsigned arguments.
29643 2017-06-02 Olivier Hainque <hainque@adacore.com>
29645 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29647 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29649 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29650 Use VALL_F16 iterator rather than VALL.
29652 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29654 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29655 Emit CBNZ inside loop when doing a strong exchange and comparing
29656 against zero. Generate the CC flags after the loop.
29658 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29660 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29661 (dl_section_ref): New.
29662 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29663 On AIX, append an expression to subtract the size of the
29664 section length to dl_section_ref.
29666 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29669 for early expansion of vector absolute builtins.
29671 2017-06-02 Richard Biener <rguenther@suse.de>
29673 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29674 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29676 2017-06-02 Richard Biener <rguenther@suse.de>
29678 PR tree-optimization/80948
29679 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29680 stmt order as well.
29682 2017-06-02 Richard Biener <rguenther@suse.de>
29684 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29686 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29687 for inductions as relevant.
29689 2017-06-02 Richard Biener <rguenther@suse.de>
29691 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29692 (vectorizable_induction): ... this. Remove dead code.
29694 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29696 * builtins. (expand_builtin_alloca): Remove second parameter and
29697 infer its value from the first parameter instead.
29698 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29700 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29702 PR rtl-optimization/80903
29703 * loop-doloop.c (add_test): Unshare sequence.
29705 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29707 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29709 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29711 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29713 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29714 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29715 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29716 (xlogue_layout::get_stub_rtx): Make static.
29717 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29718 (xlogue_layout::compute_stub_managed_regs): Rename to...
29719 (xlogue_layout::count_stub_managed_regs): ...this.
29720 (xlogue_layout::is_stub_managed_reg): New function.
29721 (xlogue_layout::m_stub_names): Rename to...
29722 (xlogue_layout::s_stub_names): ...this, make static.
29723 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29724 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29725 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29726 xlogue_layout::s_stub_names): Instantiate statics.
29727 (stub_managed_regs): Remove.
29728 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29729 (disable_call_ms2sysv_xlogues): Rename to...
29730 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29731 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29733 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29734 change after reload_completed.
29735 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29737 (ix86_expand_prologue): Likewise.
29738 (ix86_expand_epilogue): Likewise.
29739 (ix86_expand_split_stack_prologue): Likewise.
29740 (ix86_compute_frame_layout): Remove frame parameter ...
29741 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29742 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29744 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29745 (ix86_frame): Move from here ...
29746 * config/i386/i386.h (ix86_frame): ... to here.
29747 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29748 complete ix86_frame data structure instead. Remove some_ld_name.
29750 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29752 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29753 symbols that hold a DECL_VALUE_EXPR.
29755 2017-06-01 Martin Jambor <mjambor@suse.cz>
29757 PR tree-optimization/80898
29758 * tree-sra.c (process_subtree_disqualification): Removed.
29759 (disqualify_candidate): Do not acll
29760 process_subtree_disqualification.
29761 (subtree_mark_written_and_enqueue): New function.
29762 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29763 RHS has been disqualified and re-queue LHS if necessary. Apart
29764 from that, ignore disqualified RHS.
29766 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29768 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29769 address fetch for z10 or later.
29771 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29773 * config/arc/arc.md (tst_movb): Add guard when splitting.
29775 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29777 * config/arc/arc.c (arc_can_eliminate): Test against
29778 arc_frame_pointer_needed.
29780 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29782 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29783 to prevent store reordering.
29784 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29785 (type): Add block type.
29786 (stack_tie): Define special instruction to be used in
29789 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29791 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29792 constraint. It is not valid for the pattern.
29793 (noncommutative_binary_comparison): Likewise.
29795 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29797 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29800 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29802 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29803 be used by the reg-alloc.
29805 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29807 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29808 reg-alloc when having mul64 or mul32x16 instructions.
29809 (mulsidi3): Likewise.
29810 (umulsidi3): Likewise.
29811 (mulsi32x16): New pattern.
29812 (mulsi64): Likewise.
29813 (mulsidi64): Likewise.
29814 (umulsidi64): Likewise.
29815 (MUL32x16_REG): Define.
29816 (mul64_600): Use MUL32x16_REG.
29817 (mac64_600): Likewise.
29818 (umul64_600): Likewise.
29819 (umac64_600): Likewise.
29821 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29823 * config/arc/arc.md (mulsi3_700): Make it commutative.
29825 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29827 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29829 (*sign_extendsidi2_insn): Likewise for movstosw.
29831 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29833 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29834 the type of the input discriminant value. Convert the
29835 discriminant value of signedness vary.
29837 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29839 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29840 Add to -Wall section.
29842 2017-06-01 Richard Biener <rguenther@suse.de>
29844 PR middle-end/66313
29845 * fold-const.c (fold_plusminus_mult_expr): If the factored
29846 factor may be zero use a wrapping type for the inner operation.
29847 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29848 and handle moved defs.
29849 (process_assignment): Properly guard the unary op case. Return a
29850 tri-state indicating that moving the stmt before the call may allow
29851 to continue. Pass through to_move.
29852 (find_tail_calls): Handle moving unrelated defs before
29855 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29858 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29859 splitter result in the canonical way.
29861 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29863 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29864 also for 32bit target. Update insn attributes.
29865 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29867 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29869 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29870 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29871 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29872 (_mm_maskz_min_ss): New intrinsics.
29874 2017-05-31 Martin Liska <mliska@suse.cz>
29876 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29877 Change comment style to one we normally use.
29878 (vectorizable_reduction): Likewise.
29879 (vectorizable_induction): Likewise.
29880 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29881 (vectorizable_call): Likewise.
29882 (vectorizable_simd_clone_call): Likewise.
29883 (vectorizable_conversion): Likewise.
29884 (vectorizable_assignment): Likewise.
29885 (vectorizable_shift): Likewise.
29886 (vectorizable_operation): Likewise.
29887 (vectorizable_store): Likewise.
29888 (vectorizable_load): Likewise.
29889 * tree-vectorizer.h: Likewise.
29891 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29893 * passes.c (emergency_dump_function): New.
29894 * tree-pass.h (emergency_dump_function): Declare.
29895 * plugin.c (plugins_internal_error_function): Remove.
29896 * plugin.h (plugins_internal_error_function): Remove declaration.
29897 * toplev.c (internal_error_function): New static function. Use it...
29898 (general_init): ...here.
29900 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29902 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29903 (arc_rtx_costs): Add costs for new patterns.
29904 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29905 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29907 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29909 * tree-ssa-strlen.c (get_next_strinfo): New function.
29910 (get_stridx_plus_constant): Use it.
29911 (zero_length_string): Likewise.
29912 (adjust_related_strinfos): Likewise.
29913 (adjust_last_stmt): Likewise.
29915 2017-05-31 Richard Biener <rguenther@suse.de>
29918 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29919 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29921 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29923 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29924 loop_vinfo argument and use of dependence distance vectors.
29925 Check instead whether the two references differ only in their
29926 initial value and assume that they have the same alignment if the
29927 difference is a multiple of the vector alignment.
29928 (vect_analyze_data_refs_alignment): Update call accordingly.
29930 2017-05-31 Martin Liska <mliska@suse.cz>
29933 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29935 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29937 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29938 (create_intersect_range_checks): Move from ...
29939 * tree-data-ref.c (create_intersect_range_checks_index)
29940 (create_intersect_range_checks): ... to here.
29941 (create_runtime_alias_checks): New function factored from ...
29942 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29943 here. Call above function.
29944 * tree-data-ref.h (create_runtime_alias_checks): New function.
29946 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29948 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29949 segment length for dr_b and compute it in wide_int.
29951 2017-05-31 Richard Biener <rguenther@suse.de>
29953 PR tree-optimization/80906
29954 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29955 and pass through iv_map.
29956 (copy_bb_and_scalar_dependences): Adjust.
29957 (translate_pending_phi_nodes): Likewise.
29958 (copy_loop_close_phi_args): Handle code-generating IVs instead
29961 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29963 * diagnostic-color.c (color_dict): Add "type-diff".
29964 (parse_gcc_colors): Update comment.
29965 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29966 -fdiagnostics-show-template-tree and -fno-elide-type.
29967 (GCC_COLORS): Add type-diff to example.
29969 (-fdiagnostics-show-template-tree): New.
29970 (-fno-elide-type): New.
29971 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29972 the pp_format_decoder callback. Call any m_format_postprocessor's
29974 (pretty_printer::pretty_printer): Initialize
29975 m_format_postprocessor.
29976 (pretty_printer::~pretty_printer): Delete any
29977 m_format_postprocessor.
29978 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29979 (class format_postprocessor): New class.
29980 (struct pretty_printer::format_decoder): Document the new parameters.
29981 (struct pretty_printer::m_format_postprocessor): New field.
29982 * tree-diagnostic.c (default_tree_printer): Update for new
29983 bool and const char ** params.
29984 * tree-diagnostic.h (default_tree_printer): Likewise.
29986 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29988 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29989 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29990 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29991 rs6000_gen_cell_microcode code.
29992 (rs6000_final_prescan_insn): Delete.
29993 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29994 "warn-cell-microcode" entries.
29995 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29996 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29997 throughout. Change cc_reg_not_micro_cr0_operand to
29998 cc_reg_not_cr0_operand throughout.
29999 (*extendhi<mode>2_noload): Delete.
30000 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30001 (mwarn-cell-microcode): Delete.
30002 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30003 -mgen-cell-microcode and -mwarn-cell-microcode.
30005 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30008 * config/i386/constraints.md (Yd): New constraint.
30010 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30011 and (?Yd, r) alternatives. Update insn attributes.
30012 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30013 and (?*Yd, r) alternatives. Update insn attributes.
30014 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30016 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30018 * gimplify.c (gimplify_modify_expr): Don't create a
30019 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30022 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30024 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30026 2017-05-30 Richard Biener <rguenther@suse.de>
30028 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30029 and reduc_def fields.
30030 (STMT_VINFO_REDUC_TYPE): New define.
30031 (STMT_VINFO_REDUC_DEF): Likewise.
30032 (vect_force_simple_reduction): Adjust prototype.
30033 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30034 (vect_is_simple_reduction): Remove check_reduction argument.
30035 (vect_force_simple_reduction): Adjust and set
30036 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30037 (vectorizable_reduction): Do not re-do reduction analysis
30038 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30039 * tree-parloops.c (gather_scalar_reductions): Adjust.
30041 2017-05-30 Richard Biener <rguenther@suse.de>
30043 PR middle-end/80901
30044 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30047 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30049 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30050 Introduce unknown_misalignment parameter and remove vf.
30051 (vect_peeling_hash_get_lowest_cost):
30052 Pass unknown_misalignment parameter.
30053 (vect_enhance_data_refs_alignment):
30054 Fix unsupportable data ref treatment.
30056 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30058 * tree-vect-data-refs.c (vect_get_data_access_cost):
30059 Workaround for SLP handling.
30060 (vect_enhance_data_refs_alignment):
30061 Compute costs for doing no peeling at all, compare to the best
30062 peeling costs so far and avoid peeling if cheaper.
30064 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30066 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30067 Return peeling info and set costs to zero for unlimited cost
30069 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30070 with unknown misalignment. Compute and costs for unknown
30071 misalignment, compare them to the costs for known misalignment
30072 and choose the cheapest for peeling.
30074 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30076 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30077 (vect_get_peeling_costs_all_drs): Create function.
30078 (vect_peeling_hash_get_lowest_cost):
30079 Use vect_get_peeling_costs_all_drs.
30080 (vect_peeling_supportable): Create function.
30081 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30083 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30085 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30086 DR_HAS_NEGATIVE_STEP.
30087 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30088 (vect_enhance_data_refs_alignment): Use.
30089 (vect_duplicate_ssa_name_ptr_info): Use.
30090 * tree-vectorizer.h (dr_misalignment): Use.
30091 (known_alignment_for_access_p): Use.
30093 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30096 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30098 (has_section_name): New function.
30100 2017-05-30 Martin Liska <mliska@suse.cz>
30103 * auto-profile.c (get_function_decl_from_block): Fix
30106 2017-05-30 Richard Biener <rguenther@suse.de>
30108 PR middle-end/80876
30109 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30111 2017-05-30 Martin Liska <mliska@suse.cz>
30113 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30114 * dumpfile.h (struct dump_file_info): Remove ctors.
30116 2017-05-30 Martin Liska <mliska@suse.cz>
30118 * predict.def: Fix GNU coding style.
30120 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30122 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30123 Mark 'to' argument with ATTRIBUTE_UNUSED.
30125 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30127 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30128 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30129 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30132 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30134 * doc/install.texi (Options specification): Restore entry of
30135 --enable-sjlj-exceptions.
30137 2017-05-27 Michael Eager <eager@eagercon.com>
30140 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30142 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30144 * config/microblaze/microblaze.h
30145 (FIXED_REGISTERS): Update in macro.
30146 (CALL_USED_REGISTERS): Update in macro.
30148 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30150 * doc/install.texi: Add links to macOS binary distributions.
30152 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30156 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30158 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30160 2017-05-26 Martin Liska <mliska@suse.cz>
30162 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30164 2017-05-26 Martin Liska <mliska@suse.cz>
30166 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30167 always leading ';; '.
30168 (dump_bb_info): Likewise.
30169 (brief_dump_cfg): Likewise.
30170 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30171 * dumpfile.c: Remove usage of TDF_VERBOSE.
30172 * dumpfile.h (enum dump_kind): Likewise.
30173 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30174 * print-tree.c (debug_verbose): Remove.
30175 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30176 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30177 * tree-diagnostic.c (default_tree_printer): Replace
30178 TDF_DIAGNOSTIC with TDF_SLIM.
30180 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30182 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30183 in parameter loop, rather than loop_vinfo.
30184 (create_intersect_range_checks): Ditto.
30185 (vect_create_cond_for_alias_checks): Update call to above functions.
30187 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30189 PR tree-optimization/80815
30190 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30191 for merging runtime alias checks. Handle negative DR_STEPs.
30193 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30195 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30197 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30198 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30199 out code pruning runtime alias checks.
30200 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30201 factored out from above.
30202 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30204 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30206 (prune_runtime_alias_test_list): New decalaration.
30208 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30210 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30211 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30212 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30213 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30214 (operator==, comp_dr_with_seg_len_pair): Ditto.
30215 (vect_prune_runtime_alias_test_list): Ditto.
30217 2017-05-26 Martin Liska <mliska@suse.cz>
30220 * params.def: Bound partial-inlining-entry-probability param.
30222 2017-05-26 Marek Polacek <polacek@redhat.com>
30225 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30228 2017-05-26 Richard Biener <rguenther@suse.de>
30230 PR tree-optimization/80842
30231 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30234 2017-05-26 Richard Biener <rguenther@suse.de>
30236 PR tree-optimization/80844
30237 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30239 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30241 * doc/md.texi (Machine Constraints): Update x86 family
30242 machine constraints section to match 'config/i386/constraints.md'.
30244 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30246 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30248 2017-05-25 Nathan Sidwell <nathan@acm.org>
30250 * doc/invoke.texi (--enable-languages): Update documentation.
30252 2017-05-25 Martin Liska <mliska@suse.cz>
30254 * dumpfile.c: Add TDF_FOLDING.
30255 * dumpfile.h (enum dump_kind): Likewise.
30256 * genmatch.c (dt_simplify::gen_1): Use it.
30258 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30260 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30262 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30264 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30265 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30267 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30269 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30270 * match.pd (X == C): Rewrite it here.
30271 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30272 with_certain_nonzero_bits2): New predicates.
30273 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30275 2017-05-24 Nathan Sidwell <nathan@acm.org>
30277 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30280 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30283 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30285 * config/powerpcspe: New port. Files are copied from the rs6000
30286 port, with "rs6000" in filenames replaced by "powerpcspe".
30288 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30290 PR rtl-optimization/80754
30291 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30293 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30295 * config/sparc/sparc.md (length): Return the correct value for -mflat
30296 sibcalls to match output_sibcall.
30298 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30302 * config/rs6000/rs6000.c (struct machine_function): Add new field
30304 (rs6000_get_separate_components): Init that field, use it.
30305 (rs6000_components_for_bb): Use the field.
30307 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30309 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30311 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30313 PR middle-end/80823
30314 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30316 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30319 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30320 address against address_operand predicate.
30321 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30323 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30325 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30326 parameters passed indirectly.
30328 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30330 * config/i386/i386.md (*movdi_internal): Remove SSE4
30331 alternative 18 (?r, *v). Update insn attributes.
30332 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30333 Update insn attributes.
30334 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30335 Update insn attributes.
30336 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30337 alternative 1 (r, v). Remove isa attribute.
30338 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30339 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30340 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30342 2017-05-23 Tom de Vries <tom@codesourcery.com>
30344 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30347 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30349 * cgraphunit.c (symbol_table::process_new_functions): Update.
30350 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30351 (inline_generate_summary): Rename to ...
30352 (ipa_fn_summary_generate): ... this one.
30353 (inline_read_summary): Rename to ...
30354 (ipa_fn_summary_read): ... this one.
30355 (inline_write_summary): Rename to ...
30356 (ipa_fn_summary_write): ... this one.
30357 (inline_free_summary): Rename to ...
30358 (ipa_free_fn_summary): ... this one.
30359 (pass_data_local_fn_summary, pass_local_fn_summary,
30360 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30361 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30362 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30363 make_pass_ipa_fn_summary): New.
30364 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30365 inline_write_summary, inline_free_summary): Remove.
30366 (ipa_free_fn_summary) : New.
30367 * ipa-inline.c (ipa_inline): Update.
30368 (pass_ipa_inline): Do not generate summaries.
30369 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30371 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30372 and add pass_ipa_fn_summary.
30373 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30375 (make_pass_inline_parameters): Remove.
30377 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30379 * omp-low.c (struct omp_context): Remove "default_kind" member.
30382 * omp-offload.c (execute_oacc_device_lower): Remove the
30383 parallelism dimensions function attributes for unparallelized
30384 OpenACC kernels constructs.
30386 2017-05-23 Martin Liska <mliska@suse.cz>
30388 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30390 (cgraph_edge::make_speculative): Likewise.
30391 (cgraph_edge::resolve_speculation): Likewise.
30392 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30393 (cgraph_node::dump): Likewise.
30394 * cgraph.h: Likewise.
30395 * cgraphunit.c (analyze_functions): Likewise.
30396 (symbol_table::compile): Likewise.
30397 * ipa-cp.c (print_all_lattices): Likewise.
30398 (determine_versionability): Likewise.
30399 (initialize_node_lattices): Likewise.
30400 (ipcp_verify_propagated_values): Likewise.
30401 (estimate_local_effects): Likewise.
30402 (update_profiling_info): Likewise.
30403 (create_specialized_node): Likewise.
30404 (perhaps_add_new_callers): Likewise.
30405 (decide_about_value): Likewise.
30406 (decide_whether_version_node): Likewise.
30407 (identify_dead_nodes): Likewise.
30408 (ipcp_store_bits_results): Likewise.
30409 * ipa-devirt.c (dump_targets): Likewise.
30410 (ipa_devirt): Likewise.
30411 * ipa-icf.c (sem_item::dump): Likewise.
30412 (sem_function::equals): Likewise.
30413 (sem_variable::equals): Likewise.
30414 (sem_item_optimizer::read_section): Likewise.
30415 (sem_item_optimizer::execute): Likewise.
30416 (congruence_class::dump): Likewise.
30417 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30418 (dump_inline_summary): Likewise.
30419 (estimate_node_size_and_time): Likewise.
30420 (inline_analyze_function): Likewise.
30421 * ipa-inline-transform.c (inline_call): Likewise.
30422 * ipa-inline.c (report_inline_failed_reason): Likewise.
30423 (want_early_inline_function_p): Likewise.
30424 (edge_badness): Likewise.
30425 (update_edge_key): Likewise.
30426 (inline_small_functions): Likewise.
30427 * ipa-profile.c (ipa_profile): Likewise.
30428 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30429 (ipa_make_edge_direct_to_target): Likewise.
30430 (remove_described_reference): Likewise.
30431 (ipa_impossible_devirt_target): Likewise.
30432 (propagate_controlled_uses): Likewise.
30433 (ipa_print_node_params): Likewise.
30434 (ipcp_transform_function): Likewise.
30435 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30436 (propagate_pure_const): Likewise.
30437 * ipa-reference.c (generate_summary): Likewise.
30438 (read_write_all_from_decl): Likewise.
30439 (propagate): Likewise.
30440 (ipa_reference_read_optimization_summary): Likewise.
30441 * ipa-utils.c (ipa_merge_profiles): Likewise.
30442 * ipa.c (walk_polymorphic_call_targets): Likewise.
30443 (symbol_table::remove_unreachable_nodes): Likewise.
30444 (ipa_single_use): Likewise.
30445 * passes.c (execute_todo): Likewise.
30446 * predict.c (drop_profile): Likewise.
30447 * symtab.c (symtab_node::get_dump_name): New function.
30448 (symtab_node::dump_name): Likewise.
30449 (symtab_node::dump_asm_name): Likewise.
30450 (symtab_node::dump_references): Likewise.
30451 (symtab_node::dump_referring): Likewise.
30452 (symtab_node::dump_base): Likewise.
30453 (symtab_node::debug_symtab): Likewise.
30454 * tree-sra.c (convert_callers_for_node): Likewise.
30455 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30456 * value-prof.c (init_node_map): Likewise.
30458 2017-05-23 Martin Liska <mliska@suse.cz>
30460 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30461 and symtab_node::debug_symtab to symbol_table::debug.
30462 * cgraphunit.c (analyze_functions): Use the renamed function.
30463 (symbol_table::compile): Likewise.
30464 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30465 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30466 * passes.c (execute_todo): Likewise.
30467 * symtab.c (symbol_table::dump): New function.
30468 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30470 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30472 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30473 that nonconst implies exec.
30475 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30477 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30478 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30479 (inline_edge_summary_vec): Turn into ...
30480 (ipa_call_summaries): ... this one.
30481 (redirect_to_unreachable, edge_set_predicate,
30482 evaluate_properties_for_edge, inline_summary_alloc,
30483 reset_ipa_call_summary, reset_inline_summary,
30484 inline_summary_t::duplicate): Update.
30485 (inline_edge_duplication_hook): Turn to ...
30486 (ipa_call_summary_t::duplicate): ... this one.
30487 (inline_edge_removal_hook): Turn to ...
30488 (ipa_call_summary_t::remove): ... this one.
30489 (dump_inline_edge_summary): Turn to ...
30490 (dump_ipa_call_summary): ... this one.
30491 (estimate_function_body_sizes): Update.
30492 (inline_update_callee_summaries): Update.
30493 (remap_edge_change_prob): Update.
30494 (remap_edge_summaries): Update.
30495 (inline_merge_summary): Update.
30496 (do_estimate_edge_time): Update.
30497 (inline_generate_summary): Update.
30498 (inline_read_section): Update.
30499 (inline_read_summary): Update.
30500 (inline_free_summary): Update.
30501 * ipa-inline.c (can_inline_edge_p): Update.
30502 (compute_inlined_call_time): Update.
30503 (want_inline_small_function_p): Update.
30504 (edge_badness): Update.
30505 (early_inliner): Update.
30506 * ipa-inline.h (inline_edge_summary): Turn to ...
30507 (ipa_call_summary): ... this one.
30508 (ipa_call_summary_t): New class.
30509 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30510 (ipa_call_summaries): New.
30511 (inline_edge_summary): Remove.
30512 (estimate_edge_growth): Update.
30513 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30514 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30515 * ipa-split.c (execute_split_functions): Update.
30516 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30518 2017-05-23 Tom de Vries <tom@codesourcery.com>
30520 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30521 attributes): Document rdrand effective target.
30523 2017-05-23 Tom de Vries <tom@codesourcery.com>
30525 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30526 attributes): Sort alphabetically.
30528 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30530 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30532 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30535 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30536 V2DF/V2DI splat into two separate patterns, one that handles
30537 registers, and the other that only handles memory. Drop support
30538 for splatting from a GPR on ISA 2.07 and then splitting the
30539 splat into direct move and splat.
30540 (vsx_splat_<mode>_reg): Likewise.
30541 (vsx_splat_<mode>_mem): Likewise.
30543 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30545 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30547 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30549 PR middle-end/80809
30550 * omp-low.c (finish_taskreg_remap): New function.
30551 (finish_taskreg_scan): If unit size of ctx->record_type
30552 is non-constant, unshare the size expression and replace
30553 decls in it with possible outer var refs.
30555 PR middle-end/80809
30556 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30557 GOVD_SHARED rather than GOVD_PRIVATE with it.
30558 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30559 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30561 PR middle-end/80853
30562 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30563 as last argument to build_outer_var_ref for pointer bases of array
30564 section reductions.
30566 2017-05-19 Martin Sebor <msebor@redhat.com>
30568 * print-tree.c (print_node): Print DECL_READ_P flag.
30570 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30572 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30573 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30574 * cgraph.c: Likewise.
30575 * cgraphunit.c: Likewise.
30576 * gengtype.c: Likewise.
30577 * ipa-cp.c: Likewise.
30578 * ipa-devirt.c: Likewise.
30579 * ipa-icf.c: Likewise.
30580 * ipa-predicate.c: Likewise.
30581 * ipa-profile.c: Likewise.
30582 * ipa-prop.c: Likewise.
30583 * ipa-split.c: Likewise.
30585 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30586 edge_predicate_pool, dump_inline_hints,
30587 inline_summary::account_size_time, redirect_to_unreachable,
30588 edge_set_predicate, set_hint_predicate,
30589 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30590 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30591 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30592 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30593 ipa_call_summary_t::remove, initialize_growth_caches,
30594 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30595 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30596 mark_modified, unmodified_parm_1, unmodified_parm,
30597 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30598 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30599 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30600 will_be_nonconstant_predicate, record_modified_bb_info,
30601 get_minimal_bb, record_modified, param_change_prob,
30602 phi_result_unknown_predicate, predicate_for_phi_result,
30603 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30604 estimate_function_body_sizes, compute_inline_parameters,
30605 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30606 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30607 inline_update_callee_summaries, remap_edge_change_prob,
30608 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30609 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30610 inline_analyze_function, inline_summary_t::insert,
30611 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30612 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30613 inline_free_summary): Move to ipa-fnsummary.h
30614 (predicate_t): Remove.
30615 * ipa-fnsummary.c: New file.
30616 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30617 (enum inline_hints_vals, inline_hints, agg_position_info,
30618 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30619 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30620 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30621 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30622 inline_read_summary, inline_write_summary, inline_free_summary,
30623 inline_analyze_function, initialize_inline_failed,
30624 inline_merge_summary, inline_update_overall_summary,
30625 compute_inline_parameters): Move to ipa-fnsummary.h
30626 * ipa-fnsummary.h: New file.
30627 * ipa-inline-transform.h: Include ipa-inline.h.
30628 * ipa-inline.c: LIkewise.
30630 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30632 * ipa-inline.c (edge_badness): Use inlined_time instead of
30633 inline_summaries->get.
30635 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30637 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30639 2017-05-22 Nathan Sidwell <nathan@acm.org>
30641 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30642 (fdump-lang): Document 'raw' option.
30643 * dumpfile.h (TDI_tu): Delete.
30644 * dumpfile.c (dump_files): Remove translation-unit.
30645 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30647 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30649 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30650 command option from $(AWK) call.
30651 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30652 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30653 [FORMAT]: Remove handling of variable.
30654 * config/avr/t-multilib: Regenerate.
30656 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30658 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30660 (dump_inline_summary): Do not print self_time.
30661 (estimate_function_body_sizes): Do not set self_time.
30662 (compute_inline_parameters): Likewise.
30663 (inline_read_section, inline_write_summary): Do not stream self_time.
30664 * ipa-inline.h (inline_summary): Drop self_time.
30666 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30668 * ipa-inline-analysis.c (account_size_time): Rename to ...
30669 (inline_summary::account_size_time): ... this one.
30670 (reset_ipa_call_summary): Turn to ...
30671 (ipa_call_summary::reset): ... this one.
30672 (reset_inline_summary): Turn to ...
30673 (inline_summary::reset): ... this one.
30674 (inline_summary_t::remove): Update.
30675 (inline_summary_t::duplicate): Update.
30676 (ipa_call_summary_t::remove): Update.
30677 (dump_inline_summary): Update.
30678 (estimate_function_body_sizes): Update.
30679 (compute_inline_parameters): Update.
30680 (estimate_node_size_and_time): Update.
30681 (inline_merge_summary): Update.
30682 (inline_update_overall_summary): Update.
30683 (inline_read_section): Update.
30684 (inline_write_summary): Update.
30685 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30686 add account_size_time and reset member functions.
30687 (ipa_call_summary): Add reset function.
30688 * ipa-predicate.h (predicate::operator &): Constify.
30690 2017-05-22 Richard Biener <rguenther@suse.de>
30692 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30694 2017-05-19 Jason Merrill <jason@redhat.com>
30696 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30698 2017-05-19 Marek Polacek <polacek@redhat.com>
30701 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30702 TYPE_OVERFLOW_WRAPS checks.
30704 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30706 * tree-core.h (enum omp_clause_default_kind): Add
30707 "OMP_CLAUSE_DEFAULT_PRESENT".
30708 * tree-pretty-print.c (dump_omp_clause): Handle it.
30709 * gimplify.c (enum gimplify_omp_var_data): Add
30710 "GOVD_MAP_FORCE_PRESENT".
30711 (gimplify_adjust_omp_clauses_1): Map it to
30712 "GOMP_MAP_FORCE_PRESENT".
30713 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30715 * gimplify.c (oacc_default_clause): Clarify.
30717 2017-05-19 Nathan Sidwell <nathan@acm.org>
30719 LANG_HOOK_REGISTER_DUMPS
30720 * toplev.c (general_init): Call register dump lang hook.
30721 * doc/invoke.texi: Document -fdump-lang option family.
30722 * dumpfile.c (dump_files): Remove class dump here.
30723 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30724 * dumpfile.h (tree_dump_index): Remove TDI_class.
30725 * langhooks-def.h (lhd_register_dumps): Declare.
30726 (LANG_HOOKS_REGISTER_DUMPS): Define.
30727 (LANG_HOOKS_INITIALIZER): Add it.
30728 * langhooks.c (lhd_register_dumps): Define.
30729 * langhooks.h (struct lang_hooks): Add register_dumps.
30731 2017-05-19 Nathan Sidwell <nathan@acm.org>
30733 * context.h (context::set_passes): New.
30734 * context.c (context::context): Do not create pass manager.
30735 * toplev.c (general_init): Create pass manager here.
30737 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30739 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30740 use this splitter if two add or or instructions would also work for
30741 the constant we want to generate.
30743 2017-05-19 Richard Biener <rguenther@suse.de>
30746 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30747 predicate evaluation.
30749 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30751 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30753 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30756 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30758 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30759 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30760 (inline_edge_summary_vec): Turn into ...
30761 (ipa_call_summaries): ... this one.
30762 (redirect_to_unreachable, edge_set_predicate,
30763 evaluate_properties_for_edge, inline_summary_alloc,
30764 reset_ipa_call_summary, reset_inline_summary,
30765 inline_summary_t::duplicate): Update.
30766 (inline_edge_duplication_hook): Turn to ...
30767 (ipa_call_summary_t::duplicate): ... this one.
30768 (inline_edge_removal_hook): Turn to ...
30769 (ipa_call_summary_t::remove): ... this one.
30770 (dump_inline_edge_summary): Turn to ...
30771 (dump_ipa_call_summary): ... this one.
30772 (estimate_function_body_sizes): Update.
30773 (inline_update_callee_summaries): Update.
30774 (remap_edge_change_prob): Update.
30775 (remap_edge_summaries): Update.
30776 (inline_merge_summary): Update.
30777 (do_estimate_edge_time): Update.
30778 (inline_generate_summary): Update.
30779 (inline_read_section): Update.
30780 (inline_read_summary): Update.
30781 (inline_free_summary): Update.
30782 * ipa-inline.c (can_inline_edge_p): Update.
30783 (compute_inlined_call_time): Update.
30784 (want_inline_small_function_p): Update.
30785 (edge_badness): Update.
30786 (early_inliner): Update.
30787 * ipa-inline.h (inline_edge_summary): Turn to ...
30788 (ipa_call_summary): ... this one.
30789 (ipa_call_summary_t): New class.
30790 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30791 (ipa_call_summaries): New.
30792 (inline_edge_summary): Remove.
30793 (estimate_edge_growth): Update.
30794 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30795 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30796 * ipa-split.c (execute_split_functions): Update.
30797 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30799 2017-05-19 Richard Biener <rguenther@suse.de>
30801 PR middle-end/80764
30802 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30804 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30806 * config/rs6000/rs6000.c (struct machine_function): Add field
30807 fpr_is_wrapped_separately.
30808 (rs6000_get_separate_components): Use 64 components. Handle the
30809 new FPR components.
30810 (rs6000_components_for_bb): Handle the FPR components.
30811 (rs6000_emit_prologue_components): Handle the FPR components.
30812 (rs6000_emit_epilogue_components): Handle the FPR components.
30813 (rs6000_set_handled_components): Handle the FPR components.
30814 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30815 that are already separately shrink-wrapped.
30816 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30817 that are already separately shrink-wrapped.
30819 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30822 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30825 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30826 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30827 cases where the register allocator uses a move and an offsettable
30828 memory operation to/from a FPR register on ISA 2.06/2.07.
30829 (define_peephole2 for Altivec d-form store): Likewise.
30831 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30834 * config/i386/mmx.md (*mov<mode>_internal): Enable
30835 alternatives 11, 12, 13 and 14 also for 32bit targets.
30836 Remove alternatives 15, 16, 17 and 18.
30837 * config/i386/sse.md (vec_concatv2di): Change
30838 alternative (!x, *y) to (x, ?!*Yn).
30840 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30842 * dumpfile.h (enum dump_kind): Remove stray comma.
30844 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30846 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30847 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30848 predicate::num_conditions
30849 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30850 (CHANGED): turn into predicate::changed.
30851 (agg_position_info): Move to ipa-predicate.h
30852 (add_condition, predicate::add_clause, predicate::operator &=,
30853 predicate::or_with, predicate::evaluate, predicate::probability,
30854 dump_condition, dump_clause, predicate::dump,
30855 predicate::remap_after_duplication, predicate::remap_after_inlining,
30856 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30857 (evaluate_conditions_for_known_args): Update.
30858 (set_cond_stmt_execution_predicate): Update.
30859 * ipa-inline.h: Include ipa-predicate.h
30860 (condition, inline_param_summary, conditions, agg_position_info,
30861 predicate): Move to ipa-predicate.h
30862 * ipa-predicate.c: New file.
30863 * ipa-predicate.h: New file.
30865 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30867 * final.c (leaf_function_p): Check we are not in a sequence.
30869 2017-05-18 Martin Liska <mliska@suse.cz>
30871 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30872 * dumpfile.c (dump_register): Use new enum dump_kind.
30873 (get_dump_file_name): Likewise.
30874 (dump_enable_all): Likewise.
30875 (dump_switch_p_1): Likewise.
30876 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30877 * dumpfile.h (enum dump_kind): New enum type.
30878 (struct dump_file_info): Create constructor and
30879 format fields and comments.
30880 * passes.c (pass_manager::register_one_dump_file):
30882 * statistics.c (statistics_early_init): Likewise.
30883 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30884 TDF_TREE with TDF_SLIM.
30885 (gather_memory_references_ref): Likewise.
30887 2017-05-18 Martin Liska <mliska@suse.cz>
30889 * vec.h (struct vnull): Use it.
30891 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30893 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30894 (true_predicate, false_predicate, true_predicate_p,
30895 false_predicate_p): Remove.
30896 (single_cond_predicate, not_inlined_predicate): Turn to member function
30898 (add_condition): Update.
30899 (add_clause): Turn to...
30900 (predicate::add_clause): ... this one; update; allow passing NULL
30902 (and_predicates): Turn to ...
30903 (predicate::operator &=): ... this one.
30904 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30905 (or_predicates): Turn to ...
30906 (predicate::or_with): ... this one.
30907 (evaluate_predicate): Turn to ...
30908 (predicate::evaluate): ... this one.
30909 (predicate_probability): Turn to ...
30910 (predicate::probability): ... this one.
30911 (dump_condition): Update.
30912 (dump_predicate): Turn to ...
30913 (predicate::dump): ... this one.
30914 (account_size_time): Update.
30915 (edge_set_predicate): Update.
30916 (set_hint_predicate): UPdate.
30917 (evaluate_conditions_for_known_args): Update.
30918 (evaluate_properties_for_edge): Update.
30919 (remap_predicate_after_duplication): Turn to...
30920 (predicate::remap_after_duplication): ... this one.
30921 (remap_hint_predicate_after_duplication): Update.
30922 (inline_summary_t::duplicate): UPdate.
30923 (dump_inline_edge_summary): Update.
30924 (dump_inline_summary): Update.
30925 (set_cond_stmt_execution_predicate): Update.
30926 (set_switch_stmt_execution_predicate): Update.
30927 (compute_bb_predicates): Update.
30928 (will_be_nonconstant_expr_predicate): Update.
30929 (will_be_nonconstant_predicate): Update.
30930 (phi_result_unknown_predicate): Update.
30931 (predicate_for_phi_result): Update.
30932 (array_index_predicate): Update.
30933 (estimate_function_body_sizes): Update.
30934 (estimate_node_size_and_time): Update.
30935 (estimate_ipcp_clone_size_and_time): Update.
30936 (remap_predicate): Rename to ...
30937 (predicate::remap_after_inlining): ... this one.
30938 (remap_hint_predicate): Update.
30939 (inline_merge_summary): Update.
30940 (inline_update_overall_summary): Update.
30941 (estimate_size_after_inlining): Update.
30942 (read_predicate): Rename to ...
30943 (predicate::stream_in): ... this one.
30944 (read_inline_edge_summary): Update.
30945 (write_predicate): Rename to ...
30946 (predicate::stream_out): ... this one.
30947 (write_inline_edge_summary): Update.
30948 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30949 (clause_t): Turn to uint32_t
30950 (predicate): Turn to class; implement constructor and operators
30952 (size_time_entry): Update.
30953 (inline_summary): Update.
30954 (inline_edge_summary): Update.
30956 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30958 * fold-const.c (fold_binary_loc): Move transformation...
30959 * match.pd (C - X CMP X): ... here.
30961 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30963 * config/sparc/sparc.c (sparc_option_override): Set function
30964 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30965 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30967 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30969 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30971 2017-05-18 Marek Polacek <polacek@redhat.com>
30974 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30975 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30976 gimple_assign_load_p.
30978 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30980 PR middle-end/80692
30981 * real.c (do_compare): Give decimal_do_compare preference over
30982 comparing just the signs.
30984 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30986 * doc/md.texi (Canonicalization of Instructions): Describe the
30987 canonical form of instructions that inherently set a condition
30990 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30992 PR middle-end/80775
30993 * tree-cfg.c: Move deletion of unreachable case statements to after
30994 the merging of consecutive case labels.
30996 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30998 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30999 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31000 restoring of callee-saved registers.
31002 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31004 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31005 * config/visium/visium.c (single_set_and_flags): Likewise.
31006 * config/visium/visium.md (Substitutions): Likewise.
31008 2017-05-17 Martin Liska <mliska@suse.cz>
31010 * cfg.c: Introduce dump_flags_t type and
31011 use it instead of int type.
31013 * cfghooks.c: Likewise.
31014 * cfghooks.h (struct cfg_hooks): Likewise.
31015 * cfgrtl.c: Likewise.
31016 * cfgrtl.h: Likewise.
31017 * cgraph.c (cgraph_node::get_body): Likewise.
31018 * coretypes.h: Likewise.
31019 * domwalk.c: Likewise.
31020 * domwalk.h: Likewise.
31021 * dumpfile.c (struct dump_option_value_info): Likewise.
31022 (dump_enable_all): Likewise.
31023 (dump_switch_p_1): Likewise.
31024 (opt_info_switch_p): Likewise.
31025 * dumpfile.h (enum tree_dump_index): Likewise.
31026 (struct dump_file_info): Likewise.
31027 * genemit.c: Likewise.
31028 * generic-match-head.c: Likewise.
31029 * gengtype.c (open_base_files): Likewise.
31030 * gimple-pretty-print.c: Likewise.
31031 * gimple-pretty-print.h: Likewise.
31032 * graph.c (print_graph_cfg): Likewise.
31033 * graphite-scop-detection.c (dot_all_sese): Likewise.
31034 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31035 * loop-unroll.c (report_unroll): Likewise.
31036 * passes.c (pass_manager::register_one_dump_file): Likewise.
31037 * print-tree.c: Likewise.
31038 * statistics.c: Likewise.
31039 * tree-cfg.c: Likewise.
31040 * tree-cfg.h: Likewise.
31041 * tree-dfa.c: Likewise.
31042 * tree-dfa.h: Likewise.
31043 * tree-dump.c (dump_function): Likewise.
31044 * tree-dump.h (struct dump_info): Likewise.
31045 * tree-pretty-print.c: Likewise.
31046 * tree-pretty-print.h: Likewise.
31047 * tree-ssa-live.c: Likewise.
31048 * tree-ssa-live.h: Likewise.
31049 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31050 * tree-vect-loop.c: Likewise.
31051 * tree-vect-slp.c: Likewise.
31053 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31054 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31056 PR tree-optimization/80457
31057 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31058 of all arguments to a statement as scalar_to_vec operations.
31059 (vectorizable_call): Adjust call to vect_model_simple_cost for
31061 (vectorizable_conversion): Likewise.
31062 (vectorizable_assignment): Likewise.
31063 (vectorizable_shift): Likewise.
31064 (vectorizable_operation): Likewise.
31065 (vectorizable_comparison): Likewise.
31066 (vect_is_simple_cond): Record the def types for operands.
31067 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31068 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31069 for statement argument count.
31071 2017-05-16 Carl Love <cel@us.ibm.com>
31073 * config/rs6000/rs6000-c: Add support for built-in functions
31074 vector unsigned long long vec_bperm (vector unsigned long long,
31075 vector unsigned char)
31076 vector signed long long vec_mule (vector signed int,
31078 vector unsigned long long vec_mule (vector unsigned int,
31079 vector unsigned int)
31080 vector signed long long vec_mulo (vector signed int,
31082 vector unsigned long long vec_mulo (vector unsigned int,
31083 vector unsigned int)
31084 vector signed char vec_sldw (vector signed char,
31085 vector signed char,
31087 vector unsigned char vec_sldw (vector unsigned char,
31088 vector unsigned char,
31090 vector signed short vec_sldw (vector signed short,
31091 vector signed short,
31093 vector unsigned short vec_sldw (vector unsigned short,
31094 vector unsigned short,
31096 vector signed int vec_sldw (vector signed int,
31099 vector unsigned int vec_sldw (vector unsigned int,
31100 vector unsigned int,
31102 vector signed long long vec_sldw (vector signed long long,
31103 vector signed long long,
31105 vector unsigned long long vec_sldw (vector unsigned long long,
31106 vector unsigned long long,
31108 * config/rs6000/rs6000-c: Add support for built-in functions
31109 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31110 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31111 * doc/extend.texi: Update the built-in documentation for the
31112 new built-in functions.
31114 2017-05-16 Marek Polacek <polacek@redhat.com>
31118 * tree.c (save_expr): Don't fold the expression.
31120 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31122 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31123 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31124 and (?*y,m). Update insn attributes.
31126 2017-05-16 Martin Liska <mliska@suse.cz>
31128 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31129 flags argument of print_gimple_stmt, print_gimple_expr,
31130 print_generic_stmt and print_generic_expr.
31131 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31132 * coretypes.h: Likewise.
31133 * except.c (dump_eh_tree): Likewise.
31134 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31135 * gimple-pretty-print.h: Likewise.
31136 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31137 (backprop::push_to_worklist): Likewise.
31138 (backprop::pop_from_worklist): Likewise.
31139 (backprop::process_use): Likewise.
31140 (backprop::intersect_uses): Likewise.
31141 (note_replacement): Likewise.
31142 * gimple-ssa-store-merging.c
31143 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31144 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31145 (pass_store_merging::execute): Likewise.
31146 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31147 (ssa_base_cand_dump_callback): Likewise.
31148 (dump_incr_vec): Likewise.
31149 (replace_refs): Likewise.
31150 (replace_mult_candidate): Likewise.
31151 (create_add_on_incoming_edge): Likewise.
31152 (create_phi_basis): Likewise.
31153 (insert_initializers): Likewise.
31154 (all_phi_incrs_profitable): Likewise.
31155 (introduce_cast_before_cand): Likewise.
31156 (replace_one_candidate): Likewise.
31157 * gimplify.c (gimplify_expr): Likewise.
31158 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31159 (set_rename): Likewise.
31160 (rename_uses): Likewise.
31161 (copy_loop_phi_nodes): Likewise.
31162 (add_close_phis_to_merge_points): Likewise.
31163 (copy_loop_close_phi_args): Likewise.
31164 (copy_cond_phi_args): Likewise.
31165 (graphite_copy_stmts_from_block): Likewise.
31166 (translate_pending_phi_nodes): Likewise.
31167 * graphite-poly.c (print_pdr): Likewise.
31168 (dump_gbb_cases): Likewise.
31169 (dump_gbb_conditions): Likewise.
31170 (print_scop_params): Likewise.
31171 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31172 (build_cross_bb_scalars_use): Likewise.
31173 (gather_bbs::before_dom_children): Likewise.
31174 * hsa-dump.c (dump_hsa_immed): Likewise.
31175 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31176 (get_replacement_map): Likewise.
31177 * ipa-inline-analysis.c (dump_condition): Likewise.
31178 (estimate_function_body_sizes): Likewise.
31179 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31180 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31181 * ipa-prop.c (ipa_dump_param): Likewise.
31182 (ipa_print_node_jump_functions_for_edge): Likewise.
31183 (ipa_modify_call_arguments): Likewise.
31184 (ipa_modify_expr): Likewise.
31185 (ipa_dump_param_adjustments): Likewise.
31186 (ipa_dump_agg_replacement_values): Likewise.
31187 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31188 * ipa-pure-const.c (check_stmt): Likewise.
31189 (pass_nothrow::execute): Likewise.
31190 * ipa-split.c (execute_split_functions): Likewise.
31191 * omp-offload.c (dump_oacc_loop_part): Likewise.
31192 (dump_oacc_loop): Likewise.
31193 * trans-mem.c (tm_log_emit): Likewise.
31194 (tm_memopt_accumulate_memops): Likewise.
31195 (dump_tm_memopt_set): Likewise.
31196 (dump_tm_memopt_transform): Likewise.
31197 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31198 (print_loop): Likewise.
31199 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31200 (chkp_gather_checks_info): Likewise.
31201 (chkp_get_check_result): Likewise.
31202 (chkp_remove_check_if_pass): Likewise.
31203 (chkp_use_outer_bounds_if_possible): Likewise.
31204 (chkp_reduce_bounds_lifetime): Likewise.
31205 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31206 (chkp_mark_completed_bounds): Likewise.
31207 (chkp_register_incomplete_bounds): Likewise.
31208 (chkp_mark_invalid_bounds): Likewise.
31209 (chkp_maybe_copy_and_register_bounds): Likewise.
31210 (chkp_build_returned_bound): Likewise.
31211 (chkp_get_bound_for_parm): Likewise.
31212 (chkp_build_bndldx): Likewise.
31213 (chkp_get_bounds_by_definition): Likewise.
31214 (chkp_generate_extern_var_bounds): Likewise.
31215 (chkp_get_bounds_for_decl_addr): Likewise.
31216 * tree-chrec.c (chrec_apply): Likewise.
31217 * tree-data-ref.c (dump_data_reference): Likewise.
31218 (dump_subscript): Likewise.
31219 (dump_data_dependence_relation): Likewise.
31220 (analyze_overlapping_iterations): Likewise.
31221 * tree-inline.c (expand_call_inline): Likewise.
31222 (tree_function_versioning): Likewise.
31223 * tree-into-ssa.c (dump_defs_stack): Likewise.
31224 (dump_currdefs): Likewise.
31225 (dump_names_replaced_by): Likewise.
31226 (dump_update_ssa): Likewise.
31227 (update_ssa): Likewise.
31228 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31229 * tree-parloops.c (build_new_reduction): Likewise.
31230 (try_create_reduction_list): Likewise.
31231 (ref_conflicts_with_region): Likewise.
31232 (oacc_entry_exit_ok_1): Likewise.
31233 (oacc_entry_exit_single_gang): Likewise.
31234 * tree-pretty-print.h: Likewise.
31235 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31236 (get_scalar_evolution): Likewise.
31237 (add_to_evolution): Likewise.
31238 (get_loop_exit_condition): Likewise.
31239 (analyze_evolution_in_loop): Likewise.
31240 (analyze_initial_condition): Likewise.
31241 (analyze_scalar_evolution): Likewise.
31242 (instantiate_scev): Likewise.
31243 (number_of_latch_executions): Likewise.
31244 (gather_chrec_stats): Likewise.
31245 (final_value_replacement_loop): Likewise.
31246 (scev_const_prop): Likewise.
31247 * tree-sra.c (dump_access): Likewise.
31248 (disqualify_candidate): Likewise.
31249 (create_access): Likewise.
31250 (reject): Likewise.
31251 (maybe_add_sra_candidate): Likewise.
31252 (create_access_replacement): Likewise.
31253 (analyze_access_subtree): Likewise.
31254 (analyze_all_variable_accesses): Likewise.
31255 (sra_modify_assign): Likewise.
31256 (initialize_constant_pool_replacements): Likewise.
31257 (find_param_candidates): Likewise.
31258 (decide_one_param_reduction): Likewise.
31259 (replace_removed_params_ssa_names): Likewise.
31260 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31261 * tree-ssa-copy.c (dump_copy_of): Likewise.
31262 (copy_prop_visit_cond_stmt): Likewise.
31263 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31264 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31265 (record_equivalences_from_stmt): Likewise.
31266 * tree-ssa-dse.c (compute_trims): Likewise.
31267 (delete_dead_call): Likewise.
31268 (delete_dead_assignment): Likewise.
31269 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31270 (forward_propagate_into_cond): Likewise.
31271 (pass_forwprop::execute): Likewise.
31272 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31273 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31275 (move_computations_worker): Likewise.
31276 (execute_sm): Likewise.
31277 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31278 (remove_exits_and_undefined_stmts): Likewise.
31279 (remove_redundant_iv_tests): Likewise.
31280 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31281 (adjust_iv_update_pos): Likewise.
31282 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31283 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31284 (value_replacement): Likewise.
31285 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31286 * tree-ssa-pre.c (print_pre_expr): Likewise.
31287 (get_representative_for): Likewise.
31288 (create_expression_by_pieces): Likewise.
31289 (insert_into_preds_of_block): Likewise.
31290 (eliminate_insert): Likewise.
31291 (eliminate_dom_walker::before_dom_children): Likewise.
31292 (eliminate): Likewise.
31293 (remove_dead_inserted_code): Likewise.
31294 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31295 * tree-ssa-reassoc.c (get_rank): Likewise.
31296 (eliminate_duplicate_pair): Likewise.
31297 (eliminate_plus_minus_pair): Likewise.
31298 (eliminate_not_pairs): Likewise.
31299 (undistribute_ops_list): Likewise.
31300 (eliminate_redundant_comparison): Likewise.
31301 (update_range_test): Likewise.
31302 (optimize_range_tests_var_bound): Likewise.
31303 (optimize_vec_cond_expr): Likewise.
31304 (rewrite_expr_tree): Likewise.
31305 (rewrite_expr_tree_parallel): Likewise.
31306 (linearize_expr): Likewise.
31307 (break_up_subtract): Likewise.
31308 (linearize_expr_tree): Likewise.
31309 (attempt_builtin_powi): Likewise.
31310 (attempt_builtin_copysign): Likewise.
31311 (transform_stmt_to_copy): Likewise.
31312 (transform_stmt_to_multiply): Likewise.
31313 (dump_ops_vector): Likewise.
31314 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31315 (print_scc): Likewise.
31316 (set_ssa_val_to): Likewise.
31317 (visit_reference_op_store): Likewise.
31318 (visit_use): Likewise.
31319 (sccvn_dom_walker::before_dom_children): Likewise.
31320 (run_scc_vn): Likewise.
31321 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31323 (expr_hash_elt::print): Likewise.
31324 (const_and_copies::pop_to_marker): Likewise.
31325 (const_and_copies::record_const_or_copy_raw): Likewise.
31326 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31327 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31328 (dump_predicates): Likewise.
31329 (find_uninit_use): Likewise.
31330 (warn_uninitialized_phi): Likewise.
31331 (pass_late_warn_uninitialized::execute): Likewise.
31332 * tree-ssa.c (verify_vssa): Likewise.
31333 (verify_ssa): Likewise.
31334 (maybe_optimize_var): Likewise.
31335 * tree-vrp.c (dump_value_range): Likewise.
31336 (dump_all_value_ranges): Likewise.
31337 (dump_asserts_for): Likewise.
31338 (register_edge_assert_for_2): Likewise.
31339 (vrp_visit_cond_stmt): Likewise.
31340 (vrp_visit_switch_stmt): Likewise.
31341 (vrp_visit_stmt): Likewise.
31342 (vrp_visit_phi_node): Likewise.
31343 (simplify_cond_using_ranges_1): Likewise.
31344 (fold_predicate_in): Likewise.
31345 (evrp_dom_walker::before_dom_children): Likewise.
31346 (evrp_dom_walker::push_value_range): Likewise.
31347 (evrp_dom_walker::pop_value_range): Likewise.
31348 (execute_early_vrp): Likewise.
31350 2017-05-16 Richard Biener <rguenther@suse.de>
31352 * dwarf2out.c (loc_list_from_tree_1): Do not create
31353 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31355 2017-05-16 Richard Biener <rguenther@suse.de>
31357 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31359 (note_variable_value_in_expr): If we resolved the decl ref
31360 do not push to the stack.
31362 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31364 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31365 operations in fast-math mode.
31366 (vaddq_f16): Likewise.
31367 (vmul_f16): Likewise.
31368 (vmulq_f16): Likewise.
31369 (vsub_f16): Likewise.
31370 (vsubq_f16): Likewise.
31371 * config/arm/neon.md (add<mode>3): New.
31373 (fma:<VH:mode>3): New. Also remove outdated comment.
31376 2017-05-16 Martin Liska <mliska@suse.cz>
31380 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31381 (odr_types_equivalent_p): Likewise.
31383 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31385 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31387 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31390 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31391 non-interunit SSE move alternatives with '?'.
31392 (zero-extendsidi peephole2): New peephole to skip intermediate
31393 general register in SSE zero-extend sequence.
31395 2017-05-15 Jeff Law <law@redhat.com>
31397 * reorg.c (relax_delay_slots): Create a new variable to hold
31398 the temporary target rather than clobbering TARGET_LABEL.
31400 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31401 missing argument to extract_bit_field call.
31402 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31404 2017-05-15 Martin Liska <mliska@suse.cz>
31407 * gcc.c (process_command): Do not allow empty argument of -o option.
31409 2017-05-15 Renlin Li <renlin.li@arm.com>
31411 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31412 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31413 * config/aarch64/constraints.md (Usf): Add long call check.
31414 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31415 (call_value): Likewise.
31416 (sibcall): Likewise.
31417 (sibcall_value): Likewise.
31419 (call_value_insn): New.
31420 (sibcall_insn): Update rtx pattern.
31421 (sibcall_value_insn): Likewise.
31422 (call_internal): Remove.
31423 (call_value_internal): Likewise.
31424 (sibcall_internal): Likewise.
31425 (sibcall_value_internal): Likewise.
31426 (call_reg): Likewise.
31427 (call_symbol): Likewise.
31428 (call_value_reg): Likewise.
31429 (call_value_symbol): Likewise.
31431 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31434 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31436 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31438 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31439 compatible with CCGOCmode and with CCZmode.
31441 2017-05-14 Martin Sebor <msebor@redhat.com>
31443 PR middle-end/77671
31444 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31445 (gimple_fold_builtin_snprintf): Same.
31446 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31447 (gimple_fold_builtin_snprintf): Same.
31448 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31449 of character types.
31450 (is_call_safe): New function.
31451 (try_substitute_return_value): Call it.
31452 (try_simplify_call): New function.
31453 (pass_sprintf_length::handle_gimple_call): Call it.
31455 2017-05-14 Martin Sebor <msebor@redhat.com>
31457 PR middle-end/80669
31458 * builtins.c (expand_builtin_stpncpy): Simplify.
31460 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31462 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31463 * config/i386/i386.h
31464 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31465 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31466 (struct machine_function): Add new members call_ms2sysv,
31467 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31468 (struct machine_frame_state): New fields sp_realigned and
31469 sp_realigned_offset.
31470 * config/i386/i386.c
31471 (enum xlogue_stub): New enum.
31472 (enum xlogue_stub_sets): New enum.
31473 (class xlogue_layout): New class.
31474 (struct ix86_frame): New fields stack_realign_allocate_offset,
31475 stack_realign_offset and outlined_save_offset. Modify comments to
31476 detail stack layout when using out-of-line stubs.
31477 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31478 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31479 -mcall-ms2sysv-xlogues.
31480 (stub_managed_regs): New static variable.
31481 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31482 registers managed by out-of-line stub.
31483 (disable_call_ms2sysv_xlogues): New function.
31484 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31485 m->call_ms2sysv when appropriate and compute frame layout for
31487 (sp_valid_at, fp_valid_at): New inline functions.
31488 (choose_basereg): New function.
31489 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31491 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31492 Use align parameter of choose_baseaddr to generated aligned SSE movs
31494 (pro_epilogue_adjust_stack): Modify to track
31495 machine_frame_state::sp_realigned.
31496 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31497 (ix86_nsaved_sseregs): Likewise.
31498 (ix86_emit_save_regs): Likewise.
31499 (ix86_emit_save_regs_using_mov): Likewise.
31500 (ix86_emit_save_sse_regs_using_mov): Likewise.
31501 (get_scratch_register_on_entry): Likewise.
31502 (gen_frame_set): New function.
31503 (gen_frame_load): Likewise.
31504 (gen_frame_store): Likewise.
31505 (emit_outlined_ms2sysv_save): Likewise.
31506 (emit_outlined_ms2sysv_restore): Likewise.
31507 (ix86_expand_prologue): Modify stack re-alignment code and call
31508 emit_outlined_ms2sysv_save when appropriate.
31509 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31510 parameter rtx_insn *insn, which allows the function to be used to only
31511 generate the notes.
31512 (ix86_expand_epilogue): Modify validity checks of frame and stack
31513 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31514 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31515 * config/i386/predicates.md
31516 (save_multiple): New predicate.
31517 (restore_multiple): Likewise.
31518 * config/i386/sse.md
31519 (save_multiple<mode>): New pattern.
31520 (save_multiple_realign<mode>): Likewise.
31521 (restore_multiple<mode>): Likewise.
31522 (restore_multiple_and_return<mode>): Likewise.
31523 (restore_multiple_leave_return<mode>): Likewise.
31524 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31526 2017-05-14 Julia Koval <julia.koval@intel.com>
31528 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31529 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31530 (__builtin_ia32_xsetbv): New builtins.
31531 * config/i386/i386.c (ix86_expand_special_args_builtin):
31533 (ix86_expand_builtin): Special expand for new intrinsics.
31534 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31535 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31536 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31538 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31540 * cfganal.c (inverted_post_order_compute): Change argument type
31542 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31543 * df-core.c (rest_of_handle_df_initialize): Adjust.
31544 (rest_of_handle_df_finish): Likewise.
31545 (df_analyze_1): Likewise.
31546 (df_analyze): Likewise.
31547 (loop_inverted_post_order_compute): Change argument to be a vec *.
31548 (df_analyze_loop): Adjust.
31549 (df_get_n_blocks): Likewise.
31550 (df_get_postorder): Likewise.
31551 * df.h (struct df_d): Change field to be a vec.
31552 * lcm.c (compute_laterin): Adjust.
31553 (compute_available): Likewise.
31554 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31555 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31556 * tree-ssa-pre.c (compute_antic): Likewise.
31558 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31560 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31561 (depth_first_search::depth_first_search): Change structure init
31562 function to this constructor.
31563 (depth_first_search::add_bb): Rename function to this member.
31564 (depth_first_search::execute): Likewise.
31565 (flow_dfs_compute_reverse_finish): Adjust.
31567 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31569 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31570 (longest_simple_path): Likewise.
31571 * shrink-wrap.c (spread_components): Likewise.
31572 (disqualify_problematic_components): Likewise.
31573 (emit_common_heads_for_components): Likewise.
31574 (emit_common_tails_for_components): Likewise.
31575 (insert_prologue_epilogue_for_components): Likewise.
31577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31579 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31582 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31584 * df-core.c (df_set_blocks): Start using auto_bitmap.
31585 (df_compact_blocks): Likewise.
31586 * df-problems.c (df_rd_confluence_n): Likewise.
31587 * df-scan.c (df_insn_rescan_all): Likewise.
31588 (df_process_deferred_rescans): Likewise.
31589 (df_update_entry_block_defs): Likewise.
31590 (df_update_exit_block_uses): Likewise.
31591 (df_entry_block_bitmap_verify): Likewise.
31592 (df_exit_block_bitmap_verify): Likewise.
31593 (df_scan_verify): Likewise.
31594 * lra-constraints.c (lra_constraints): Likewise.
31595 (undo_optional_reloads): Likewise.
31596 (lra_undo_inheritance): Likewise.
31597 * lra-remat.c (calculate_gen_cands): Likewise.
31598 (do_remat): Likewise.
31599 * lra-spills.c (assign_spill_hard_regs): Likewise.
31600 (spill_pseudos): Likewise.
31601 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31602 (bitmap_set_subtract_values): Likewise.
31604 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31606 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31607 management with auto_bitmap.
31608 (fix_inter_tick): Likewise.
31609 (fix_recovery_deps): Likewise.
31610 * ira.c (add_store_equivs): Likewise.
31611 (find_moveable_pseudos): Likewise.
31612 (split_live_ranges_for_shrink_wrap): Likewise.
31613 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31614 (rtx_reuse_manager::seen_def_p): Likewise.
31615 (rtx_reuse_manager::set_seen_def): Likewise.
31616 * print-rtl.h (class rtx_reuse_manager): Likewise.
31618 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31620 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31622 (migrate_btr_def): Likewise.
31623 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31624 * df-core.c (loop_post_order_compute): Likewise.
31625 (loop_inverted_post_order_compute): Likewise.
31626 * hsa-common.h: Likewise.
31627 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31628 * init-regs.c (initialize_uninitialized_regs): Likewise.
31629 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31630 (inline_small_functions): Likewise.
31631 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31632 * ira.c (combine_and_move_insns): Likewise.
31633 (build_insn_chain): Likewise.
31634 * loop-invariant.c (find_invariants): Likewise.
31635 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31636 * predict.c (tree_predict_by_opcode): Likewise.
31637 (predict_paths_leading_to): Likewise.
31638 (predict_paths_leading_to_edge): Likewise.
31639 (estimate_loops_at_level): Likewise.
31640 (estimate_loops): Likewise.
31641 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31642 (spread_components): Likewise.
31643 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31644 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31645 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31646 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31647 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31648 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31649 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31650 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31651 (mark_threaded_blocks): Likewise.
31652 (thread_through_all_blocks): Likewise.
31653 * tree-ssa.c (verify_ssa): Likewise.
31654 (execute_update_addresses_taken): Likewise.
31655 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31657 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31659 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31661 (post_order_compute): Likewise.
31662 (inverted_post_order_compute): Likewise.
31663 (pre_and_rev_post_order_compute_fn): Likewise.
31665 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31667 * genrecog.c (int_set::int_set): Explicitly construct our
31668 auto_vec base class.
31669 * vec.h (auto_vec::auto_vec): New constructor.
31671 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31673 * bitmap.h (class auto_bitmap): New constructor taking
31674 bitmap_obstack * argument.
31676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31678 * bitmap.h (class auto_bitmap): Change type of m_bits to
31679 bitmap_head, and adjust ctor / dtor and member operators.
31681 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31683 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31684 when returned register mode doesn't match original mode.
31686 2017-05-12 Jeff Law <law@redhat.com>
31687 Jakub Jelinek <jakub@redhat.com>
31689 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31690 we look for cc setter after the compare-elim changes.
31691 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31692 within the vector to match what compare-elim now expects.
31693 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31694 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31696 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31697 after the compare-elim changes.
31698 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31699 the vector to match what compare-elim now expects.
31700 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31701 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31702 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31703 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31704 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31706 * config/visium/visium.c (single_set_and_flags): Fix where
31707 we look for cc setter after the compare-elim changes.
31708 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31709 with the vector to match what compare-elim now expects.
31710 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31711 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31712 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31713 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31714 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31715 (neg<mode>2_insn_set_overflow): Likewise.
31717 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31719 PR middle-end/79794
31720 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31721 maybe_expand_insn call, set ops[0].target. If still set after call,
31722 set alt_rtl. Add extra arg to recursive calls.
31723 (extract_bit_field): Add alt_rtl argument. Pass to
31725 * expmed.h (extract_bit_field): Fix prototype.
31726 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31727 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31728 to extract_bit_field_calls.
31729 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31730 Pass alt_rtl to extract_bit_field calls.
31731 * calls.c (store_unaligned_arguments_into_psuedos)
31732 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31733 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31735 * optabs.h (struct expand_operand): Add target bitfield.
31737 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31739 * compare-elim.c (try_eliminate_compare): Canonicalize
31740 operation with embedded compare to
31741 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31742 (set (reg) (operation)].
31744 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31746 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31749 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31750 cost of adding a carry flag for ADC instruction.
31751 [case MINUS]: Ignore the cost of subtracting a carry flag
31752 for SBB instruction.
31754 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31756 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31758 * config/rs6000/bmiintrin.h: New file.
31759 * config/rs6000/bmi2intrin.h: New file.
31760 * config/rs6000/x86intrin.h: New file.
31762 2017-05-12 Jeff Law <law@redhat.com>
31764 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31767 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31769 PR middle-end/80707
31770 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31772 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31775 early expansion of vector divide builtins.
31776 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31777 builtins identified as having unsigned arguments.
31779 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31781 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31782 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31783 expansion of vector logical operations (and, andc, or, xor,
31786 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31788 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31789 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31791 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31793 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31794 early GIMPLE expansion of vector multiplies.
31796 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31798 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31799 TARGET_HAVE_MOVT conditional.
31800 (movt splitter): Likewise.
31802 2017-05-12 Richard Biener <rguenther@suse.de>
31804 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31805 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31806 Fold all stmts not inplace.
31808 2017-05-12 Richard Biener <rguenther@suse.de>
31810 PR tree-optimization/80713
31811 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31812 inserted_exprs bit for not removed stmts.
31814 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31816 PR middle-end/69921
31817 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31818 parallelized" attribute for parallelized OpenACC kernels.
31819 * omp-offload.c (execute_oacc_device_lower): Use it.
31821 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31822 Set "oacc kernels" attribute.
31823 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31824 parameter. Adjust all users.
31825 (oacc_fn_attrib_kernels_p): Remove function.
31826 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31827 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31828 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31829 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31830 assert "oacc kernels" attribute is set.
31832 2017-05-11 Carl Love <cel@us.ibm.com>
31834 * config/rs6000/rs6000-c: Add support for built-in functions
31835 vector unsigned char vec_popcnt (vector signed char)
31836 vector unsigned char vec_popcnt (vector unsigned char)
31837 vector unsigned short vec_popcnt (vector signed short)
31838 vector unsigned short vec_popcnt (vector unsigned short)
31839 vector unsigned int vec_popcnt (vector signed int)
31840 vector unsigned int vec_popcnt (vector unsigned int)
31841 vector unsigned long long vec_popcnt (vector signed long long)
31842 vector unsigned long long vec_popcnt (vector unsigned long long)
31843 vector signed long long vec_slo (vector signed long long,
31844 vector signed char)
31845 vector signed long long vec_slo (vector signed long long,
31846 vector unsigned char)
31847 vector unsigned long long vec_slo (vector unsigned long long,
31848 vector signed char)
31849 vector unsigned long long vec_slo (vector unsigned long long,
31850 vector unsigned char)
31851 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31852 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31853 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31854 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31855 * doc/extend.texi: Update the built-in documentation file for the
31856 new built-in functions.
31858 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31860 * attribs.h (sorted_attr_string): Move machine independent
31861 functions for target clone support from the i386 port to common
31862 code. Rename ix86_function_versions to common_function_versions.
31863 Rename make_name to make_unique_name.
31864 (common_function_versions): Likewise.
31865 (make_unique_name): Likewise.
31866 (make_dispatcher_decl): Likewise.
31867 (is_function_default_version): Likewise.
31868 * attribs.c (attr_strcmp): Likewise.
31869 (sorted_attr_string): Likewise.
31870 (common_function_versions): Likewise.
31871 (make_unique_name): Likewise.
31872 (make_dispatcher_decl): Likewise.
31873 (is_function_default_version): Likewise.
31874 * config/i386/i386.c (attr_strcmp): Likewise.
31875 (sorted_attr_string): Likewise.
31876 (ix86_function_versions): Likewise.
31877 (make_name): Likewise.
31878 (make_dispatcher_decl): Likewise.
31879 (is_function_default_version): Likewise.
31880 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31882 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31885 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31886 Account for direct move costs for vec_construct of integer
31889 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31892 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31893 (UNSPEC_STX_ATOMIC): Ditto.
31894 (loaddi_via_sse): New insn.
31895 (storedi_via_sse): Ditto.
31896 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31897 Update corresponding peephole2 patterns.
31898 (atomic_storedi_fpu): Ditto.
31900 2017-05-11 Julia Koval <julia.koval@intel.com>
31902 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31903 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31905 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31906 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31907 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31909 2017-05-11 Nathan Sidwell <nathan@acm.org>
31911 * graphite-poly.c: Include dumpfile.h.
31913 * dumpfle.h (dump_function): Declare here ...
31914 * tree-dump.h (dump_function): ... not here.
31915 * dumpfile.c: #include tree-cfg.h.
31916 (dump_function): Move here from ...
31917 * tree-dump.c (dump_function): ... here.
31918 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31919 * graphite-poly.c: Don't include tree-dump.h.
31920 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31921 * print-tree.c: Likewise.
31922 * stor-layout.c: Likewise.
31923 * tree-nested.c: Likewise.
31925 * dumpfile.c (dump_start): Use TDF_FLAGS.
31926 (dump_enable_all): Fix TDF_KIND check thinko.
31928 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31930 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31931 array entries to represent two legal parameterizations of the
31932 overloaded __builtin_cmpb function, as represented by the
31933 P6_OV_BUILTIN_CMPB constant.
31934 (altivec_resolve_overloaded_builtin): Add special case handling
31935 for the __builtin_cmpb function, as represented by the
31936 P6_OV_BUILTIN_CMPB constant.
31937 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31938 (BU_P6_64BIT_2): New macro.
31939 (BU_P6_OVERLOAD_2): New macro
31940 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31941 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31942 (CMPB): Add overload support to represent both 32-bit and 64-bit
31943 compare-bytes function.
31944 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31945 support for TARGET_CMPB.
31946 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31947 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31948 documentation of the __builtin_cmpb overloaded built-in function.
31950 2017-05-11 Richard Biener <rguenther@suse.de>
31952 PR tree-optimization/80705
31953 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31954 bases are not vectorizable.
31956 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31958 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31959 when counting register pressure.
31961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31963 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31964 (iv_ca_more_deps): Renamed to ...
31965 (iv_ca_compare_deps): ... this.
31966 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31968 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31970 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31972 (determine_group_iv_costs): ... here.
31973 (find_inv_vars_cb): Record inv var if it's not recorded before.
31975 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31977 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31978 (get_shiftadd_cost): Ditto.
31980 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31982 * tree-ssa-address.c: Include header file.
31983 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31985 (add_to_parts): Refactor.
31986 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31987 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31990 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31992 PR tree-optimization/53090
31993 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31995 (extract_cond_operands): Detect condition with IV on both sides
31996 and return COMP_IV_EXPR_2.
31997 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31998 (rewrite_use_compare): Simplify by removing call to function
31999 extract_cond_operands.
32001 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32003 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32004 (extract_cond_operands): Detect condition comparing against non-
32005 invariant bound and return appropriate enum value.
32006 (find_interesting_uses_cond): Update use of extract_cond_operands.
32007 Handle its return value accordingly.
32008 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32010 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32012 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32013 nonlinear iv_use computation in loop invariant sensitive way.
32015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32017 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32018 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32022 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32023 (dump_cand): Support iv_cand.inv_exprs.
32024 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32026 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32029 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32031 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32033 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32034 as local function. Include necessary header files.
32035 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32039 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32041 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32043 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32044 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32045 RSHIFT_EXPR and BIT_NOT_EXPR.
32047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32049 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32050 (adjust_setup_cost): New parameter supporting round up adjustment.
32051 (struct address_cost_data): Delete.
32052 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32053 (split_address_cost, ptr_difference_cost): Delete.
32054 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32055 (struct ainc_cost_data): New struct.
32056 (get_address_cost_ainc): New function.
32057 (get_address_cost, get_computation_cost): Reimplement.
32058 (determine_group_iv_cost_address): Record inv_expr for all uses of
32060 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32061 (iv_ca_has_deps): Reimplemented to ...
32062 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32064 (iv_ca_extend): Call iv_ca_more_deps.
32066 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32068 * tree-ssa-address.c (struct mem_address): Move to header file.
32069 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32070 * tree-ssa-address.h (struct mem_address): Move from C file.
32071 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32075 * tree-affine.h (aff_combination_type): New interface.
32076 (aff_combination_zero_p): Remove static.
32077 (aff_combination_const_p): New interface.
32078 (aff_combination_singleton_var_p): New interfaces.
32080 2017-05-11 Richard Biener <rguenther@suse.de>
32082 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32083 Skip unreachable blocks and destinations.
32084 (eliminate): Move stmt removal and fixup ...
32085 (fini_eliminate): ... here. Skip inserted exprs.
32086 (pass_pre::execute): Move fini_pre after fini_eliminate.
32087 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32088 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32089 PRE to get rid of dead code that has invalid SSA form and
32090 split critical edges again.
32092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32094 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32096 2017-05-11 Richard Biener <rguenther@suse.de>
32098 * passes.c (execute_function_todo): Verify loops if they are
32099 said to be up-to-date.
32100 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32101 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32103 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32106 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32107 handle calling assemble_external ourself.
32110 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32111 modes with zero size. Enhance comment.
32113 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32115 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32116 built-ins for vec_xl and vec_xst with short and char pointer
32119 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32121 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32122 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32123 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32124 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32125 (_mm_maskz_min_round_ss): New intrinsics.
32126 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32127 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32128 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32129 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32130 (__builtin_ia32_minss_mask_round): New builtins.
32131 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32132 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32133 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32135 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32136 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32138 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32141 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32143 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32144 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32145 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32146 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32147 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32148 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32149 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32150 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32151 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32152 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32153 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32154 (__builtin_ia32_mulss_mask_round): New builtins.
32155 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32156 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32157 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32159 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32160 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32162 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32165 2017-05-10 Julia Koval <julia.koval@intel.com>
32167 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32168 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32169 (_mm256_setr_m128i): New intrinsics.
32171 2017-05-10 Julia Koval <julia.koval@intel.com>
32173 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32174 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32175 (_mm_maskz_rcp14_ss): New intrinsics.
32176 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32177 (__builtin_ia32_rcp14ss_mask): New builtins.
32178 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32180 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32182 PR tree-optimization/51513
32183 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32184 (assert_unreachable_fallthru_edge_p): Use it.
32185 (group_case_labels_stmt): Likewise.
32186 * tree-cfg.h: Prototype it.
32187 * stmt.c: Include cfghooks.h and tree-cfg.h.
32188 (emit_case_dispatch_table) <gap_label>: New local variable.
32189 Use it to fill dispatch table gaps.
32190 Test for default_label before updating probabilities.
32191 (expand_case) <default_label>: Remove unneeded initialization.
32192 Test for unreachable default case statement and remove its edge.
32193 Set default_label accordingly.
32194 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32196 2017-05-10 Carl Love <cel@us.ibm.com>
32198 * config/rs6000/rs6000-c: Add support for built-in functions
32199 vector signed char vec_neg (vector signed char)
32200 vector signed short int vec_neg (vector short int)
32201 vector signed int vec_neg (vector signed int)
32202 vector signed long long vec_neg (vector signed long long)
32203 vector float vec_neg (vector float)
32204 vector double vec_neg (vector double)
32205 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32207 * config/rs6000/altivec.h: Add define for vec_neg
32208 * doc/extend.texi: Update the built-in documentation for the
32209 new built-in functions.
32211 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32213 PR tree-optimization/77644
32214 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32216 2017-05-10 Nathan Sidwell <nathan@acm.org>
32218 * dumpfile.h (TDI_lang_all): New.
32219 (TDF_KIND): New. Renumber others
32220 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32222 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32224 (get_dump_file_name): Adjust suffix generation.
32225 (dump_enable_all): Use TDF_KIND.
32226 * doc/invoke.texi (-fdump-lang-all): Document.
32228 * dumpfile.h: Tabify.
32230 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32233 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32234 Move member access before delete.
32236 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32238 * tree-inline.c (expand_call_inline): Split block at stmt
32241 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32244 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32245 are now unused after splitting mov{sf,sd}_hardfloat.
32246 (f32_lr2): Likewise.
32247 (f32_lm): Likewise.
32248 (f32_lm2): Likewise.
32249 (f32_li): Likewise.
32250 (f32_li2): Likewise.
32251 (f32_lv): Likewise.
32252 (f32_sr): Likewise.
32253 (f32_sr2): Likewise.
32254 (f32_sm): Likewise.
32255 (f32_sm2): Likewise.
32256 (f32_si): Likewise.
32257 (f32_si2): Likewise.
32258 (f32_sv): Likewise.
32259 (f32_dm): Likewise.
32260 (f32_vsx): Likewise.
32261 (f32_av): Likewise.
32262 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32263 For movsf, order stores so the VSX stores occur before the GPR
32264 store which encourages the register allocator to use a traditional
32265 FPR instead of a GPR. For movsd, order the stores so that the GPR
32266 store comes before the VSX stores to allow the power6 to work.
32267 This is due to the power6 not having a 32-bit integer store
32268 instruction from a FPR.
32269 (movsf_hardfloat): Likewise.
32270 (movsd_hardfloat): Likewise.
32272 2017-05-09 Martin Sebor <msebor@redhat.com>
32274 PR translation/80280
32275 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32278 PR translation/80280
32279 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32280 data member added in r247778.
32281 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32283 2017-05-09 Nathan Sidwell <nathan@acm.org>
32285 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32287 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32290 2017-05-09 Marek Polacek <polacek@redhat.com>
32292 * doc/invoke.texi: Fix typo.
32294 2017-05-09 Richard Biener <rguenther@suse.de>
32296 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32297 (vrp_val_is_min): Likewise.
32298 (set_value_range_to_value): Likewise.
32299 (set_value_range_to_nonnegative): Likewise.
32300 (gimple_assign_nonzero_p): Likewise.
32301 (gimple_stmt_nonzero_p): Likewise.
32302 (vrp_int_const_binop): Likewise. Remove unreachable case.
32303 (adjust_range_with_scev): Adjust comments.
32304 (compare_range_with_value): Likewise.
32305 (extract_range_from_phi_node): Likewise.
32306 (test_for_singularity): Likewise.
32308 2017-05-09 Richard Biener <rguenther@suse.de>
32310 * tree-vrp.c (get_single_symbol): Add assert that we don't
32311 get overflowed constants as invariant part.
32312 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32313 checks. Use wi::cmp instead of recursing for integer constants.
32314 (compare_values): Just ignore whether we assumed undefined
32315 overflow instead of failing the compare.
32316 (extract_range_for_var_from_comparison_expr): Add comment before the
32317 TREE_NO_WARNING sets.
32318 (test_for_singularity): Likewise.
32319 (extract_range_from_comparison): Do not disable optimization
32320 when we assumed undefined overflow.
32321 (extract_range_basic): Remove init of unused var.
32323 2017-05-09 Richard Biener <rguenther@suse.de>
32325 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32326 (extract_range_from_multiplicative_op_1): Adjust.
32327 (extract_range_from_binary_expr_1): Use int_const_binop.
32329 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32332 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32333 rs6000_store_data_bypass_p in seven define_bypass directives and
32334 in several comments.
32335 * config/rs6000/rs6000-protos.h: Add prototype for
32336 rs6000_store_data_bypass_p function.
32337 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32338 function implements slightly different (rs6000-specific) semantics
32339 than store_data_bypass_p, returning false rather than aborting
32340 with assertion error when arguments do not satisfy the
32341 requirements of store data bypass.
32342 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32343 rs6000_store_data_bypass_p.
32345 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32347 * config/xtensa/xtensa-protos.h
32348 (xtensa_initial_elimination_offset): New declaration.
32349 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32350 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32351 macro definition, add case for FRAME_POINTER_REGNUM when
32352 FRAME_GROWS_DOWNWARD.
32353 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32354 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32355 xtensa_initial_elimination_offset.
32357 2017-05-08 Nathan Sidwell <nathan@acm.org>
32359 * doc/invoke.texi: Alphabetize -fdump options.
32361 2017-05-08 Martin Sebor <msebor@redhat.com>
32363 PR translation/80280
32364 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32366 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32368 * target.def (compute_frame_layout): New optional target hook.
32369 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32370 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32371 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32373 * reload1.c (verify_initial_elim_offsets): Likewise.
32374 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32375 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32376 (arm_get_frame_offsets): Split up into this ...
32377 (arm_compute_frame_layout): ... and this function.
32379 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32381 * config/aarch64/constraints.md (Usa): New constraint.
32382 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32384 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32386 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32387 with_multilib_list after it has been checked.
32389 2017-05-08 Richard Biener <rguenther@suse.de>
32391 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32392 (bitmap_set_subtract_values): Likewise.
32394 2017-05-08 Richard Biener <rguenther@suse.de>
32396 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32397 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32399 (gimple_stmt_nonzero_warnv_p): Rename to ...
32400 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32402 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32403 (extract_range_basic): Adjust, do not disable propagation on
32404 strict overflow sensitive simplification.
32405 (vrp_visit_cond_stmt): Likewise.
32407 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32409 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32410 body size unconditionally.
32412 2017-05-07 Jeff Law <law@redhat.com>
32415 2017-05-06 Jeff Law <law@redhat.com>
32416 PR tree-optimization/78496
32417 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32420 PR tree-optimization/78496
32421 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32422 (simplify_stmt_using_ranges): Call it.
32423 (vrp_dom_walker::before_dom_children): Extract equivalences
32424 from an ASSERT_EXPR with an equality comparison against a
32427 2017-05-06 Jeff Law <law@redhat.com>
32429 PR tree-optimization/78496
32430 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32433 PR tree-optimization/78496
32434 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32435 (simplify_stmt_using_ranges): Call it.
32436 (vrp_dom_walker::before_dom_children): Extract equivalences
32437 from an ASSERT_EXPR with an equality comparison against a
32440 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32442 * lra-constraints.c (lra_copy_reg_equiv): New function.
32443 (split_reg): Use it to copy equivalence information from the
32444 original register to the spill register.
32446 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32448 PR rtl-optimization/75964
32449 * simplify-rtx.c (simplify_const_relational_operation): Remove
32450 invalid handling of comparisons of integer ABS.
32452 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32454 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32455 initialize to zero.
32456 (init_regs): Remove declaration.
32457 (function_arg_advance_32): Initialize error_p as boolean variable.
32459 2017-05-05 Nathan Sidwell <nathan@acm.org>
32461 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32462 lines. Use for (;;).
32464 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32466 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32467 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32468 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32469 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32470 VF=2 that require versioning.
32472 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32474 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32477 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32479 * diagnostic.h (diagnostic_override_option_index): Convert from
32480 macro to inline function.
32482 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32484 * diagnostic.c (last_module_changed_p): New function.
32485 (set_last_module): New function.
32486 (diagnostic_report_current_module): Convert macro usage to
32487 the above functions.
32488 * diagnostic.h (diagnostic_context::last_module): Strengthen
32489 from const line_map * to const line_map_ordinary *.
32490 (diagnostic_last_module_changed): Delete macro.
32491 (diagnostic_set_last_module): Delete macro.
32493 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32495 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32496 with diagnostic_report_diagnostic.
32497 (diagnostic_n_impl_richloc): Likewise.
32498 * diagnostic.h (report_diagnostic): Delete macro.
32499 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32500 with diagnostic_report_diagnostic.
32501 * substring-locations.c (format_warning_va): Likewise.
32503 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32505 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32506 save/restor of format_spec. Move option-printing code to...
32507 (print_option_information): ...this new function, and
32508 reimplement by simply printing to the pretty_printer,
32509 rather than appending to the format string.
32511 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32513 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32514 handling logic into...
32515 (update_effective_level_from_pragmas): ...this new function.
32517 2017-05-04 Andrew Waterman <andrew@sifive.com>
32519 * config/riscv/riscv.opt (mstrict-align): New option.
32520 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32521 (SLOW_UNALIGNED_ACCESS): Define.
32522 (riscv_slow_unaligned_access): Declare.
32523 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32525 (riscv_slow_unaligned_access): New variable.
32526 (rocket_tune_info): Set slow_unaligned_access to true.
32527 (optimize_size_tune_info): Set slow_unaligned_access to false.
32528 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32529 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32530 (riscv_option_override): Set riscv_slow_unaligned_access.
32531 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32533 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32535 * config/riscv/riscv.md: Unify indentation.
32537 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32542 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32544 (extendsi<mode>2): Add support for doing sign extension via
32545 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32546 don't have ISA 3.0 instructions.
32547 (extendsi<mode>2 splitter): Likewise.
32548 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32549 generate the normal insns since SImode can now go in vector
32550 registers. Disallow the special UNSPECs needed for previous
32551 machines to hide SImode being used. Add new insns
32552 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32553 (fix_trunc<mode>si2_stfiwx): Likewise.
32554 (fix_trunc<mode>si2_internal): Likewise.
32555 (fixuns_trunc<mode>si2): Likewise.
32556 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32557 (fctiw<u>z_<mode>_smallint): Likewise.
32558 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32559 of floating point to 32-bit integer from doing a direct move to
32560 the GPR registers to do a store.
32561 (fctiwz_<mode>): Break long line.
32563 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32565 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32566 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32567 (addr_list, addr_offset_valid_p): New.
32568 (split_address_groups): Check offset validity with above function.
32569 (gt-tree-ssa-loop-ivopts.h): Include header file.
32571 2017-05-05 Nathan Sidwell <nathan@acm.org>
32573 * config.gcc (arm*-*-*): Add missing 'fi'.
32575 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32577 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32578 included in -fopt-info does not matter.
32579 * doc/optinfo.texi (-fopt-info): Fix description of default
32580 behavour. Explicitly say order of options included in -fopt-info
32583 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32585 * config.gcc: Allow combinations of aprofile and rmprofile values for
32586 --with-multilib-list.
32587 * config/arm/t-multilib: New file.
32588 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32589 variables. Remove setting of ISA and floating-point ABI in
32590 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32591 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32592 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32593 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32595 * config/arm/t-rmprofile: Likewise except for the matches changes.
32596 * doc/install.texi (--with-multilib-list): Document the combination of
32597 aprofile and rmprofile values and warn about pitfalls in doing that.
32599 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32601 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32602 (movdi_aarch64): Likewise.
32604 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32606 PR tree-optimization/80632
32607 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32609 (build_arrays): Initialize it for virtual phis.
32610 (fix_phi_nodes): Use it for virtual phis.
32612 PR tree-optimization/80558
32613 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32614 [x, y] op z into [x op, y op z] for op & or | if conditions
32617 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32618 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32621 * config/arm/arm.md (use_literal_pool): Remove.
32622 (64-bit immediate split): No longer takes cost into consideration
32623 if arm_disable_literal_pool is enabled.
32624 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32625 used when arm_disable_literal_pool is enabled.
32626 (arm_max_const_double_inline_cost): Remove use of
32627 arm_disable_literal_pool.
32628 (push_minipool_fix): Add assert.
32629 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32630 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32631 (no_literal_pool_sf_immediate): New.
32633 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32635 PR tree-optimization/80613
32636 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32637 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32639 2017-05-05 Richard Biener <rguenther@suse.de>
32641 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32643 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32645 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32646 of this flag from insn conditions due to removal from r247495.
32648 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32650 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32652 (arm_early_store_addr_dep_ptr): Likewise.
32653 * config/arm/aarch-common-protos.h
32654 (arm_early_load_addr_dep_ptr): Add prototype.
32655 (arm_early_store_addr_dep_ptr): Likewise.
32656 * config/arm/cortex-a53.md: Add new bypasses.
32658 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32660 * tree.c (next_type_uid): Change type to unsigned.
32661 (type_hash_canon): Decrement back next_type_uid if
32662 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32663 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32666 2017-05-04 Martin Sebor <msebor@redhat.com>
32668 * builtins.c: Fix a trivial typo in a comment.
32670 PR middle-end/79234
32671 * builtins.c (check_sizes): Adjust to handle reading past the end.
32672 Avoid printing excessive upper bound of ranges. Use %E to print
32673 tree nodes instead of converting them to %wu.
32674 (expand_builtin_memchr): New function.
32675 (compute_dest_size): Rename...
32676 (compute_objsize): ...to this.
32677 (expand_builtin_memcpy): Adjust.
32678 (expand_builtin_mempcpy): Adjust.
32679 (expand_builtin_strcat): Adjust.
32680 (expand_builtin_strcpy): Adjust.
32681 (check_strncat_sizes): Adjust.
32682 (expand_builtin_strncat): Adjust.
32683 (expand_builtin_strncpy): Adjust and simplify.
32684 (expand_builtin_memset): Adjust.
32685 (expand_builtin_bzero): Adjust.
32686 (expand_builtin_memcmp): Adjust.
32687 (expand_builtin): Handle memcmp.
32688 (maybe_emit_chk_warning): Check strncat just once.
32690 2017-05-04 Martin Sebor <msebor@redhat.com>
32692 PR preprocessor/79214
32693 PR middle-end/79222
32694 PR middle-end/79223
32695 * builtins.c (check_sizes): Add inlining context and issue
32696 warnings even when -Wno-system-headers is set.
32697 (check_strncat_sizes): Same.
32698 (expand_builtin_strncat): Same.
32699 (expand_builtin_memmove): New function.
32700 (expand_builtin_stpncpy): Same.
32701 (expand_builtin): Handle memmove and stpncpy.
32703 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32705 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32706 which is not used any more.
32708 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32710 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32712 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32714 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32715 (cortexa53_tunings): Likewise.
32716 (cortexa57_tunings): Likewise.
32717 (cortexa72_tunings): Likewise.
32718 (cortexa73_tunings): Likewise.
32720 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32722 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32723 Set loop alignment to 8.
32725 2017-05-04 Martin Sebor <msebor@redhat.com>
32727 PR translation/80280
32728 * builtins.c (expand_builtin_object_size): Add missing quoting to
32729 %D and like directives.
32730 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32731 (hsa_type_for_tree_type): Same.
32732 (verify_function_arguments): Same.
32733 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32734 * varasm.c (get_section): Same.
32737 2017-05-04 Martin Sebor <msebor@redhat.com>
32739 PR translation/80280
32740 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32742 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32744 * config/aarch64/aarch64.c (generic_addrcost_table):
32745 Change HI/TI mode setting.
32747 2017-05-04 Martin Jambor <mjambor@suse.cz>
32749 PR tree-optimization/80622
32750 * tree-sra.c (comes_initialized_p): New function.
32751 (build_accesses_from_assign): Only set write lazily when
32752 comes_initialized_p is false.
32753 (analyze_access_subtree): Use comes_initialized_p.
32754 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32755 instead of testing for PARM_DECL.
32757 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32759 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32760 constraint on operand 0 to allow more general addressing modes.
32761 Adjust output template.
32762 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32764 * config/aarch64/aarch64-protos.h
32765 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32766 * config/aarch64/constraints.md (Dp): New address constraint.
32767 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32770 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32772 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32773 update use of estimate_ipcp_clone_size_and_time.
32774 (estimate_local_effects): Update use of
32775 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32776 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32777 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32778 Return nonspecialized time.
32780 2017-05-04 Richard Biener <rguenther@suse.de>
32782 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32783 for the last VUSE which def dominates the PHI. Directly call
32785 (get_continuation_for_phi_1): Remove.
32787 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32789 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32790 to explain the use of truncating division. Cap the number of
32791 iterations to the maximum given by nb_iterations_upper_bound,
32794 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32796 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32797 * configure: Regenerate.
32798 * config.in: Regenerate.
32799 * config/i386/driver-mingw32.c: new file.
32800 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32801 * config.host: Link driver-mingw32.o on MinGW host.
32802 * doc/install.texi: Document new --enable-mingw-wildcard configure
32805 2017-05-04 Marek Polacek <polacek@redhat.com>
32807 PR tree-optimization/80612
32808 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32810 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32811 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32813 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32814 (movt splitter): Likewise.
32815 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32816 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32817 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32818 block for Thumb-1 with MOVT.
32819 (thumb2_legitimate_address_p): Move code block ...
32820 (can_avoid_literal_pool_for_label_p): ... into this new function.
32821 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32823 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32824 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32825 "M-profile targets with the MOVT instruction".
32827 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32829 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32830 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32831 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32833 2017-05-04 Martin Liska <mliska@suse.cz>
32835 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32836 variable cond_code.
32838 2017-05-04 Richard Biener <rguenther@suse.de>
32840 * tree.c (array_at_struct_end_p): Handle arrays at struct
32841 end with flexarrays more conservatively. Refactor and treat
32842 arrays of arrays or aggregates more strict. Fix
32843 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32844 * tree.c (array_at_struct_end_p): Adjust prototype.
32845 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32846 * gimple-fold.c (get_range_strlen): Likewise.
32847 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32849 2017-05-04 Richard Biener <rguenther@suse.de>
32851 PR tree-optimization/31130
32852 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32854 (supports_overflow_infinity): Likewise.
32855 (is_negative_overflow_infinity): Likewise.
32856 (is_positive_overflow_infinity): Likewise.
32857 (is_overflow_infinity): Likewise.
32858 (stmt_overflow_infinity): Likewise.
32859 (overflow_infinity_range_p): Likewise.
32860 (usable_range_p): Remove as always returning true.
32861 (make_overflow_infinity): Remove.
32862 (negative_overflow_infinity): Likewise.
32863 (positive_overflow_infinity): Likewise.
32864 (avoid_overflow_infinity): Likewise.
32865 (set_value_range): Adjust accordingly.
32866 (set_value_range_to_nonnegative): Likewise, remove now unused
32867 overflow_infinity arg.
32868 (vrp_operand_equal_p): Adjust.
32869 (update_value_range): Likewise.
32870 (range_int_cst_singleton_p): Likewise.
32871 (operand_less_p): Likewise.
32872 (compare_values_warnv): Likewise.
32873 (extract_range_for_var_from_comparison_expr): Likewise.
32874 (vrp_int_const_binop): Likewise.
32875 (zero_nonzero_bits_from_vr): Likewise.
32876 (extract_range_from_multiplicative_op_1): Likewise.
32877 (extract_range_from_binary_expr_1): Likewise.
32878 (extract_range_from_unary_expr): Likewise.
32879 (extract_range_from_comparison): Likewise.
32880 (extract_range_basic): Likewise.
32881 (adjust_range_with_scev): Likewise.
32882 (compare_ranges): Likewise.
32883 (compare_range_with_value): Likewise.
32884 (dump_value_range): Likewise.
32885 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32887 (simplify_cond_using_ranges): Adjust.
32889 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32891 * brig-builtins.def: Added a builtin for class_f64.
32892 * builtin-types.def: Added a builtin type needed by class_f64.
32894 2017-05-03 Jason Merrill <jason@redhat.com>
32896 * timevar.def: Add TV_CONSTEXPR.
32898 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32900 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32902 2017-05-03 Martin Jambor <mjambor@suse.cz>
32904 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32905 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32906 * ipa-cp.c (ipcp_propagate_stage): Do not call
32907 ipa_update_after_lto_read.
32908 * ipa-inline.c (ipa_inline): Likewise.
32910 2017-05-03 Martin Jambor <mjambor@suse.cz>
32912 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32913 tag. Added a default constructor and a destructor.
32914 (ipa_edge_args_sum_t): New class;
32915 (ipa_edge_args_sum): Declare.
32916 (ipa_edge_args_vector): Remove declaration.
32917 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32918 (ipa_free_edge_args_substructures): Remove declaration.
32919 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32920 (ipa_edge_args_info_available_for_edge_p): Likewise.
32921 * ipa-prop.c (ipa_edge_args_vector): Removed.
32922 (edge_removal_hook_holder): Likewise.
32923 (edge_duplication_hook_holder): Likewise.
32924 (ipa_edge_args_sum): New variable.
32925 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32926 ipa_edge_args_vector.
32927 (ipa_free_edge_args_substructures): Likewise.
32928 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32929 ipa_edge_args_vector.
32930 (ipa_edge_removal_hook): Turned into method
32931 ipa_edge_args_sum_t::remove.
32932 (ipa_edge_duplication_hook): Turned into method
32933 ipa_edge_args_sum_t::duplicate.
32934 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32935 registering edge hooks.
32936 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32937 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32938 ipa_edge_args_sum instead of ipa_edge_args_vector.
32939 * ipa-profile.c (ipa_profile): Likewise.
32941 2017-05-03 Martin Jambor <mjambor@suse.cz>
32943 * symbol-summary.h (function_summary): New method exists.
32944 (function_summary::symtab_removal): Deallocate through release.
32945 (call_summary): New class.
32946 (gt_ggc_mx): New overload.
32947 (gt_pch_nx): Likewise.
32948 (gt_pch_nx): Likewise.
32950 2017-05-03 Jeff Law <law@redhat.com>
32952 PR tree-optimization/78496
32953 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32954 from simplify_cond_using_ranges. Split off code to walk
32955 backwards through casts into ...
32956 (simplify_cond_using_ranges_2): New function.
32957 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32958 (execute_vrp): After identifying jump threads, call
32959 simplify_cond_using_ranges_2.
32961 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32964 * ipa-inline.h (inline_summary): Add ctor.
32965 (create_ggc): Do not use ggc_cleared_alloc.
32967 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32968 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32970 * gcc.c (handle_braces): Support escaping in switch matching
32972 * doc/invoke.texi (Spec Files): Document it.
32973 Remove superfluous @code markup in items.
32975 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32977 * diagnostic-show-locus.c (struct column_range): New struct.
32978 (get_affected_columns): New function.
32979 (get_printed_columns): New function.
32980 (struct correction): New struct.
32981 (correction::ensure_capacity): New function.
32982 (correction::ensure_terminated): New function.
32983 (struct line_corrections): New struct.
32984 (line_corrections::~line_corrections): New dtor.
32985 (line_corrections::add_hint): New function.
32986 (layout::print_trailing_fixits): Reimplement in terms of the new
32988 (selftest::test_overlapped_fixit_printing): New function.
32989 (selftest::diagnostic_show_locus_c_tests): Call it.
32991 2017-05-03 Nathan Sidwell <nathan@acm.org>
32993 Canonicalize canonical type hashing
32994 * tree.h (type_hash_canon_hash): Declare.
32995 * tree.c (type_hash_list, attribute_hash_list): Move into
32996 type_hash_canon_hash.
32997 (build_type_attribute_qual_variant): Break out hash code calc into
32998 type_hash_canon_hash.
32999 (type_hash_canon_hash): New. Generic type hash computation.
33000 (build_range_type_1, build_array_type_1, build_function_type,
33001 build_method_type_directly, build_offset_type, build_complex_type,
33002 make_vector_type): Call it.
33004 2017-05-03 Richard Biener <rguenther@suse.de>
33006 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33007 When all DRs have unknown misaligned do not always peel
33008 when there is a store but apply the same costing model as if
33009 there were only loads.
33011 2017-05-03 Richard Biener <rguenther@suse.de>
33014 PR tree-optimization/80492
33015 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33016 compare_base_decls returning dont-know properly.
33018 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33020 * config/arm/iterators.md (CCSI): New mode iterator.
33021 (arch): New mode attribute.
33022 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33023 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33024 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33025 code iterator for success result mode.
33026 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33027 the corresponding new insn generators.
33029 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33033 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33035 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33037 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33038 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33039 (DDR_A): Wrap DDR argument in brackets.
33040 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33041 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33042 (DDR_REVERSED_P): Likewise.
33044 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33046 PR tree-optimization/79472
33047 * tree-switch-conversion.c (struct switch_conv_info): Add
33048 contiguous_range and default_case_nonstandard fields.
33049 (collect_switch_conv_info): Compute contiguous_range and
33050 default_case_nonstandard fields, don't clear final_bb if
33051 contiguous_range and only the default case doesn't have the required
33053 (check_all_empty_except_final): Set default_case_nonstandard instead
33054 of failing if contiguous_range and the default case doesn't have empty
33056 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33057 and only the default case doesn't have the required constants. Skip
33059 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33060 if default_case_nonstandard.
33061 (build_constructors): Build constant 1 just once. Assert that default
33062 values aren't inserted in between cases if contiguous_range. Skip
33064 (build_arrays): Skip virtual phis.
33065 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33066 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33067 Handle virtual phis.
33068 (gen_inbound_check): Handle default_case_nonstandard case.
33069 (process_switch): Adjust check_final_bb caller. Call
33070 gather_default_values with the first non-default case instead of
33071 default case if default_case_nonstandard.
33073 2017-05-02 Nathan Sidwell <nathan@acm.org>
33075 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33076 check. Fix formatting.
33078 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33080 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33081 errors when comparing specialized and unspecialized times.
33083 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33085 * diagnostic-show-locus.c
33086 (layout::should_print_annotation_line_p): Make private.
33087 (layout::print_annotation_line): Make private.
33088 (layout::annotation_line_showed_range_p): Make private.
33089 (layout::show_ruler): Make private.
33090 (layout::print_source_line): Make private. Pass in line and
33091 line_width, rather than calling location_get_source_line. Drop
33093 (layout::print_leading_fixits): New method.
33094 (layout::print_any_fixits): Rename to...
33095 (layout::print_trailing_fixits): ...this, and make private.
33096 Don't print newline fixits.
33097 (diagnostic_show_locus): Move logic for printing one row into...
33098 (layout::print_line): ...this new function. Move the
33099 location_get_source_line call and error-handling from
33100 print_source_line to here. Call print_leading_fixits, and rename
33101 print_any_fixits to print_trailing_fixits.
33102 (selftest::test_fixit_insert_containing_newline): Update now that
33103 newlines are partially supported.
33104 (selftest::test_fixit_insert_containing_newline_2): New test.
33105 (selftest::test_fixit_replace_containing_newline): Update comments.
33106 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33107 * edit-context.c (class added_line): New class.
33108 (class edited_line): Describe newline handling in comment.
33109 (edited_line::actually_edited_p): New method.
33110 (edited_line::print_content): Delete redundant decl.
33111 (edited_line::m_predecessors): New field.
33112 (edited_file::print_content): Call edited_line::print_content.
33113 (edited_file::print_diff): Update to support newlines.
33114 (edited_file::print_diff_hunk): Likewise.
33115 (edited_file::print_run_of_changed_lines): New function.
33116 (edited_file::print_diff_line): Convert to...
33117 (print_diff_line): ...this.
33118 (edited_file::get_effective_line_count): New function.
33119 (edited_line::edited_line): Initialize new field m_predecessors.
33120 (edited_line::~edited_line): Clean up m_predecessors.
33121 (edited_line::apply_fixit): Handle newlines.
33122 (edited_line::get_effective_line_count): New function.
33123 (edited_line::print_content): New function.
33124 (edited_line::print_diff_lines): New function.
33125 (selftest::test_applying_fixits_insert_containing_newline): New
33127 (selftest::test_applying_fixits_replace_containing_newline): New
33129 (selftest::insert_line): New function.
33130 (selftest::test_applying_fixits_multiple_lines): Add example of
33132 (selftest::edit_context_c_tests): Call the new tests.
33134 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33136 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33137 parameter cand. Update dump information.
33138 (get_computation_cost): Update uses.
33140 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33142 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33143 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33144 (get_computation_at, rewrite_use_address): Update use of
33145 get_computation_aff.
33147 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33149 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33150 (get_computation): Delete.
33151 (get_computation_cost): Implement like get_computation_cost_at.
33152 Use get_computation_at.
33153 (get_computation_cost_at): Delete.
33154 (rewrite_use_nonlinear_expr): Use get_computation_at.
33155 (rewrite_use_compare, remove_unused_ivs): Ditto.
33157 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33159 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33161 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33163 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33164 (ivopts_global_cost_for_size): Rename parameter and update uses.
33165 (iv_ca_recount_cost): Update uses.
33166 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33167 candidates seperately in n_invs and n_cands.
33168 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33172 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33173 (find_inv_vars_cb): New.
33174 (find_depends): Renamed to ...
33175 (find_inv_vars): ... this.
33176 (add_candidate_1, force_var_cost): Call find_inv_vars.
33177 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33179 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33181 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33182 inv_vars. Add inv_exprs.
33183 (struct iv_cand): Rename depends_on to inv_vars.
33184 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33185 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33186 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33187 (dump_cand): Dump inv_vars.
33188 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33189 (record_invariant, find_depends, add_candidate_1): Ditto.
33190 (set_group_iv_cost, force_var_cost): Ditto.
33191 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33192 (get_computation_cost_at, get_computation_cost): Ditto.
33193 (determine_group_iv_cost_generic): Ditto.
33194 (determine_group_iv_cost_address): Ditto.
33195 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33196 (determine_group_iv_costs): Ditto.
33197 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33198 (iv_ca_set_remove_invariants): Renamed to ...
33199 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33200 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33201 (iv_ca_set_add_invariants): Renamed to ...
33202 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33203 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33204 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33205 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33206 (create_new_ivs): Remove useless dump.
33208 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33210 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33212 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33213 (iv_ca_set_no_cp, create_new_iv): Ditto.
33215 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33217 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33219 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33221 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33222 function tree_check2.
33224 2017-05-02 Martin Liska <mliska@suse.cz>
33226 * doc/gcov.texi: Add missing preposition.
33227 * gcov.c (function_info::function_info): Properly fill up
33228 all member variables.
33230 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33232 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33234 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33236 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33238 2017-05-02 Martin Liska <mliska@suse.cz>
33241 * lto-streamer-in.c (lto_read_tree_1): Remove
33242 LTO_STREAMER_DEBUG.
33243 * lto-streamer.c (struct tree_hash_entry): Likewise.
33244 (struct tree_entry_hasher): Likewise.
33245 (tree_entry_hasher::hash): Likewise.
33246 (tree_entry_hasher::equal): Likewise.
33247 (lto_streamer_init): Likewise.
33248 (lto_orig_address_map): Likewise.
33249 (lto_orig_address_get): Likewise.
33250 (lto_orig_address_remove): Likewise.
33251 * lto-streamer.h: Likewise.
33252 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33253 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33255 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33257 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33258 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33259 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33260 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33261 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33262 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33263 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33264 (mm_maskz_sub_ss): New intrinsics.
33265 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33266 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33267 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33268 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33269 (__builtin_ia32_subss_mask_round): New builtins.
33270 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33271 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33272 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33274 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33275 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33277 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33280 2017-05-02 Martin Jambor <mjambor@suse.cz>
33282 PR tree-optimization/78687
33283 * tree-sra.c (access): New field parent.
33284 (process_subtree_disqualification): New function.
33285 (disqualify_candidate): Call it.
33286 (build_accesses_from_assign): Reset write flag if creating an
33288 (build_access_subtree): Fill in parent field and also prpagate
33289 down grp_write flag.
33290 (create_artificial_child_access): New parameter set_grp_write, set
33291 grp_write to its value.
33292 (propagate_subaccesses_across_link): Also propagate grp_write flag
33294 (propagate_all_subaccesses): Push the closest parent back to work
33295 queue if add_access_to_work_queue returned true.
33297 2017-05-02 Richard Biener <rguenther@suse.de>
33299 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33300 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33301 -fstrict-overflow documentation.
33302 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33303 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33304 flag_strict_overflow.
33305 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33306 * lto-opts.c (lto_write_options): Do not stream it.
33307 * lto-wrapper.c (merge_and_complain): Do not handle it.
33308 * opts.c (default_options_table): Do not set -fstrict-overflow.
33309 (finish_options): Likewise do not clear it when sanitizing.
33310 * simplify-rtx.c (simplify_const_relational_operation): Do not
33311 test flag_strict_overflow.
33313 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33315 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33316 using enabled attribute.
33317 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33318 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33319 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33320 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33321 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33322 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33323 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33324 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33325 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33326 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33328 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33330 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33332 2017-05-02 Richard Biener <rguenther@suse.de>
33334 PR tree-optimization/80591
33336 2017-04-10 Richard Biener <rguenther@suse.de>
33338 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33341 2017-05-02 Richard Biener <rguenther@suse.de>
33343 PR tree-optimization/80549
33344 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33345 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33346 headers if they do not have a preheader.
33348 2017-05-02 Martin Liska <mliska@suse.cz>
33351 * common.opt: Fix typo.
33352 * doc/invoke.texi: Likewise.
33354 2017-05-01 Jan Beulich <jbeulich@suse.com>
33356 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33357 swapping, add (x,x,m,x,n) alternative.
33359 2017-05-01 Nathan Sidwell <nathan@acm.org>
33361 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33362 unnecessary unadjusted_alignment check.
33364 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33367 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33369 * gimplify.c (gimplify_cilk_detach): New function.
33370 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33371 * tree-core.h: Document EXPR_CILK_SPAWN.
33372 * tree.h (EXPR_CILK_SPAWN): Define.
33374 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33376 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33377 to use new fixit_hint representation, using the "replace" logic.
33378 (get_line_span_for_fixit_hint): Likewise.
33379 (layout::print_any_fixits): Likewise.
33380 (selftest::test_one_liner_many_fixits): Rename to...
33381 (selftest::test_one_liner_many_fixits_1): ...this, and update
33382 comment and expected output to reflect that the multiple fix-it
33383 hints are now consolidated into one insertion.
33384 (selftest::test_one_liner_many_fixits_2): New test.
33385 (selftest::test_diagnostic_show_locus_one_liner): Update for
33387 (selftest::test_fixit_consolidation): Update for fix-it API
33389 * diagnostic.c (print_parseable_fixits): Likewise.
33390 * edit-context.c (edited_line::m_line_events): Convert from
33391 auto_vec <line_event *> to auto_vec <line_event>.
33392 (class line_event): Convert from abstract base class to a concrete
33393 class, taking over the role of replace_event.
33394 (class insert_event): Delete.
33395 (class replace_event): Rename to class line_event. Convert to
33397 (edit_context::add_fixits): Reimplement.
33398 (edit_context::apply_insert): Delete.
33399 (edit_context::apply_replace): Rename to...
33400 (edit_context::apply_fixit): ...this. Convert to half-open range.
33401 (edited_file::apply_insert): Delete.
33402 (edited_file::apply_replace): Rename to...
33403 (edited_file::apply_fixit): ...this.
33404 (edited_line::~edited_line): Drop deletion of events.
33405 (edited_line::apply_insert): Delete.
33406 (edited_line::apply_replace): Rename to...
33407 (edited_line::apply_fixit): ...this. Convert to half-open range.
33408 Update for change to type of m_line_events.
33409 * edit-context.h (edit_context::apply_insert): Delete.
33410 (edit_context::apply_replace): Rename to...
33411 (edit_context::apply_fixit): ...this.
33413 2017-05-01 Martin Sebor <msebor@redhat.com>
33415 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33418 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33421 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33422 __get_cpuid_max returns 0.
33423 (__get_cpuid_count): Ditto.
33425 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33427 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33428 replacement expression is another instance of one of its arguments.
33430 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33433 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33434 check for stack push/pop autoinc.
33435 * config/i386/i386.c (ix86_agi_dependent): Return false
33436 if the only reason why modified_in_p returned true is that
33437 addr is SP based and set_insn is a push or pop.
33439 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33441 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33444 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33447 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33448 (account_size_time): Use two predicates - exec_pred and
33450 (evaluate_conditions_for_known_args): Compute both clause and
33452 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33453 (inline_summary_t::duplicate): Update.
33454 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33456 (compute_inline_parameters): Likewise.
33457 (estimate_edge_size_and_time): Update caluclation of time.
33458 (estimate_node_size_and_time): Compute both time and nonspecialized
33460 (estimate_ipcp_clone_size_and_time): Update.
33461 (inline_merge_summary): Update.
33462 (do_estimate_edge_time): Update.
33463 (do_estimate_edge_size): Update.
33464 (do_estimate_edge_hints): Update.
33465 (inline_read_section, inline_write_summary): Stream both new predicates.
33466 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33468 (compute_inlined_call_time): Cleanup.
33469 (big_speedup_p): Update.
33470 (edge_badness): Update.
33471 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33472 (size_time_entry): Replace predicate by exec_predicate and
33473 nonconst_predicate.
33474 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33475 (estimate_edge_time): Return also nonspec_time.
33476 (reset_edge_growth_cache): Update.
33478 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33480 PR rtl-optimization/80491
33481 * ifcvt.c (noce_process_if_block): When looking for x setter
33482 with missing else_bb, don't check only the insn right before
33483 cond_earliest, but look for the last insn that x is modified in
33484 within the same bb.
33486 PR rtl-optimization/80491
33487 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33489 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33491 PR tree-optimization/80487
33492 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33494 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33496 PR tree-optimization/79697
33497 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33498 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33499 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33501 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33502 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33504 2017-04-28 Martin Sebor <msebor@redhat.com>
33506 PR tree-optimization/80523
33507 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33508 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33510 (maybe_warn, format_directive, parse_directive): Use new functions.
33511 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33513 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33515 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33517 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33519 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33520 target_header_dir): Set correctly.
33521 * configure: Regenerated.
33522 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33523 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33524 instead of SYSTEM_HEADER_DIR.
33526 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33528 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33529 (estimate_local_effects): Likewise.
33530 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33531 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33532 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33533 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33534 do_estimate_edge_time, estimate_edge_time): Likewise.
33535 * ipa-inline-analysis.c (estimate_node_size_and_time,
33536 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33537 (estimate_time_after_inlining): Remove.
33539 2017-04-28 Martin Liska <mliska@suse.cz>
33541 * doc/gcov.texi: Enhance documentation of gcov.
33543 2017-04-28 Martin Liska <mliska@suse.cz>
33545 * doc/gcov.texi: Sort options in alphabetic order.
33546 * doc/gcov-dump.texi: Likewise.
33547 * doc/gcov-tool.texi: Likewise.
33548 * gcov.c (print_usage): Likewise.
33549 * gcov-dump.c (print_usage): Likewise.
33550 * gcov-tool.c (print_merge_usage_message): Likewise.
33551 (print_rewrite_usage_message): Likewise.
33552 (print_overlap_usage_message): Likewise.
33554 2017-04-28 Martin Liska <mliska@suse.cz>
33556 PR gcov-profile/53915
33557 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33559 2017-04-28 Martin Liska <mliska@suse.cz>
33561 PR gcov-profile/79891
33562 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33563 is marked by compiler as living on a line.
33564 (get_cycles_count): Remove usage of the union.
33565 (output_intermediate_file): Likewise.
33566 (find_source): Fix GNU coding style.
33567 (accumulate_line_counts): Remove old non-all block mode.
33568 (output_lines): Remove usage of the union.
33569 * profile.c (output_location): Include all BBs, even if
33570 belonging to a same line (and file) as a previous BB.
33572 2017-04-28 Martin Liska <mliska@suse.cz>
33574 * gcov.c (process_args): Handle new argument 'w'.
33575 (read_graph_file): Assign ID to BBs.
33576 (output_branch_count): Display BB # if verbose flag is set.
33577 (output_lines): Likewise for arcs.
33578 (print_usage): Add '--verbose' option help.
33579 * doc/gcov.texi: Document --verbose (-w) option.
33581 2017-04-28 Martin Liska <mliska@suse.cz>
33583 * gcov.c (struct block_location_info): New struct.
33584 (process_file): Fill up the new structure.
33585 (read_graph_file): Replace usage of encoding by the newly added
33587 (add_line_counts): Likewise.
33588 (accumulate_line_counts): Remove usage of the union.
33589 (function_info::function_info): New function.
33590 (function_info::~function_info): Likewise.
33591 (process_file): Call delete instead of release_function.
33592 (release_function): Release the function.
33593 (release_structures): Call delete instead of release_function.
33594 (solve_flow_graph): Replace usage of num_blocks.
33595 (find_exception_blocks): Likewise.
33596 (output_lines): Fix GNU coding style.
33598 2017-04-28 Martin Liska <mliska@suse.cz>
33601 * coverage.c (coverage_remove_note_file): New function.
33602 * coverage.h: Declare the function.
33603 * toplev.c (finalize): Clean if an error has been seen.
33605 2017-04-28 Martin Liska <mliska@suse.cz>
33607 PR gcov-profile/80031
33608 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33609 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33610 * gcov.c (read_graph_file): Read just number of blocks.
33611 * profile.c (branch_prob): Do not stream 0 flags per a basic
33614 2017-04-28 Martin Liska <mliska@suse.cz>
33616 * gcov-dump.c (tag_*): Add new argument to declarations.
33617 (dump_gcov_file): Likewise.
33618 (tag_blocks): Add and use new argument depth.
33619 (tag_arcs): Likewise.
33620 (tag_lines): Likewise.
33621 (tag_counters): Likewise.
33622 (tag_summary): Likewise.
33623 (dump_working_sets): Use depth to do a proper indentation.
33625 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33628 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33629 * symtab.c (symtab_node::debug_symtab): Move definition here.
33631 2017-04-28 Richard Biener <rguenther@suse.de>
33633 * lto-streamer.h (LTO_major_version): Bump to 7.
33635 2017-04-28 Richard Biener <rguenther@suse.de>
33637 * tree-vrp.c (assert_info): New struct.
33638 (add_assert_info): New helper.
33639 (register_edge_assert_for_2): Refactor to add asserts to a vector
33641 (register_edge_assert_for_1): Likewise.
33642 (register_edge_assert_for): Likewise.
33643 (finish_register_edge_assert_for): New helper actually registering
33644 asserts where live on edge.
33645 (find_conditional_asserts): Adjust.
33646 (find_switch_asserts): Likewise.
33647 (evrp_dom_walker::try_find_new_range): Generalize.
33648 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33650 2017-04-27 Marek Polacek <polacek@redhat.com>
33653 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33654 arg10 and arg11 to itype.
33656 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33658 * doc/extend.texi (Object Size Checking): Improve grammar.
33660 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33663 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33664 that the logic for permitting reciprocal estimates matches that
33667 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33670 * tree.c (type_cache_hasher::equal): Only compare
33671 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33672 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33673 non-aggregate element types.
33674 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33675 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33677 2017-04-27 Richard Biener <rguenther@suse.de>
33679 PR middle-end/80533
33680 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33681 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33682 keeping a reference to a trailing array.
33684 2017-04-27 Richard Biener <rguenther@suse.de>
33686 PR middle-end/80539
33687 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33688 being in loop-closed SSA form conservatively.
33689 (chrec_fold_multiply_poly_poly): Likewise.
33691 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33693 PR middle-end/79665
33694 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33695 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33697 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33700 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33701 (aarch64_function_arg_alignment): Return unsigned int again, but still
33702 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33703 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33704 Don't emit -Wpsabi note.
33705 (aarch64_function_arg_boundary): Likewise.
33706 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33709 2017-04-26 Nathan Sidwell <nathan@acm.org>
33711 * tree.h (crc32_unsigned_n): Declare.
33712 (crc32_unsigned, crc32_unsigned): Make inline.
33713 * tree.c (crc32_unsigned_bits): Replace with ...
33714 (crc32_unsigned_n): ... this.
33715 (crc32_unsigned, crc32_byte): Remove.
33716 (crc32_string): Remove unnecessary braces.
33718 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33720 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33721 * ipa-inline-analysis.c (MAX_TIME): Remove.
33722 (account_size_time): Use sreal for time.
33723 (dump_inline_summary): Update.
33724 (estimate_function_body_sizes): Update.
33725 (estimate_edge_size_and_time): Update.
33726 (estimate_calls_size_and_time): Update.
33727 (estimate_node_size_and_time): Update.
33728 (inline_merge_summary): Update.
33729 (inline_update_overall_summary): Update.
33730 (estimate_time_after_inlining): Update.
33731 (inline_read_section): Update.
33732 (inline_write_summary): Update.
33733 * ipa-inline.c (compute_uninlined_call_time): Update.
33734 (compute_inlined_call_time): Update.
33735 (recursive_inlining): Update.
33736 (inline_small_functions): Update.
33737 (dump_overall_stats): Update.
33738 * ipa-inline.h: Include sreal.h.
33739 (size_time_entry): Turn time to sreal.
33740 (inline_summary): Turn self_time nad time to sreal.
33742 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33744 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33746 (sreal::stream_out, sreal::stream_in): New.
33747 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33749 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33751 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33754 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33757 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33759 (dimode_scalar_chain::compute_convert_gain): Ditto.
33760 (dimode_scalar_chain::make_vector_copies): Ditto.
33761 (dimode_scalar_chain::convert_reg): Ditto.
33762 (dimode_scalar_chain::convert_insn): Ditto.
33763 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33764 (VI248_AVX512BW_1): New mode iterator.
33765 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33766 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33769 2017-04-25 Martin Sebor <msebor@redhat.com>
33771 PR tree-optimization/80497
33772 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33773 constants are representable in HOST_WIDE_INT.
33774 (parse_directive): Ditto.
33776 2017-04-25 Martin Sebor <msebor@redhat.com>
33779 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33780 (new_zero_array): Adjust signature.
33781 (dom_info::dom_init): Used unsigned rather that size_t.
33782 (dom_info::dom_info): Same.
33784 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33785 Jakub Jelinek <jakub@redhat.com>
33788 * config/arm/arm.c: Include gimple.h.
33789 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33790 returns negative, increment ncrn only if it returned positive.
33791 (arm_needs_doubleword_align): Return int instead of bool,
33792 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33793 members, but if there is any such non-FIELD_DECL
33794 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33795 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33796 returns negative, increment nregs only if it returned positive.
33797 (arm_setup_incoming_varargs): Likewise.
33798 (arm_function_arg_boundary): Emit -Wpsabi note if
33799 arm_needs_doubleword_align returns negative, return
33800 DOUBLEWORD_ALIGNMENT only if it returned positive.
33802 2017-04-25 Marek Polacek <polacek@redhat.com>
33805 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33806 first argument to type.
33808 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33811 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33812 type checks to test for compatibility instead of equality.
33814 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33815 Jakub Jelinek <jakub@redhat.com>
33818 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33820 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33821 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33822 the alignment computation, but return their maximum in warn_alignment.
33823 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33824 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33826 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33827 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33830 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33832 * config/arc/simdext.md (dmpyh): Fix typo.
33834 2017-04-25 Richard Biener <rguenther@suse.de>
33836 PR tree-optimization/80492
33837 * alias.c (compare_base_decls): Handle registers with asm
33838 specification conservatively.
33839 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33840 compare_base_decls returning dont-know properly.
33842 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33844 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33845 (legitimate_offset_address_p): New function.
33846 (arc_legitimate_address_p): Use above function.
33848 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33850 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33852 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33854 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33855 ACCH registers whenever they are available.
33857 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33859 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33860 double regs fix when not used.
33862 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33864 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33866 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33867 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33869 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33871 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33872 when emitting short ADD instructions.
33874 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33876 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33878 (cmpsi_cc_c_insn): Likewise.
33879 (cbranchsi4_scratch): Compute proper instruction length using
33880 compact_hreg_operand.
33881 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33883 2017-04-25 Richard Biener <rguenther@suse.de>
33885 PR middle-end/80509
33886 * passes.c (pass_manager::pass_manager): Initialize
33887 m_name_to_pass_map.
33889 2017-04-25 Richard Biener <rguenther@suse.de>
33891 PR tree-optimization/79201
33892 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33894 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33897 * config/s390/vector.md: Split MEM->GPR vector moves for
33898 non-s_operand addresses.
33900 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33903 * config/s390/predicates.md (reload_const_wide_int_operand): New
33905 * config/s390/s390.md ("movti"): Remove d/P alternative.
33906 ("movti_bigconst"): New pattern definition.
33908 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33911 * s390-protos.h (s390_expand_cs_hqi): Removed.
33912 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33913 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33914 modes as well as CCZ1mode and CCZmode.
33915 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33916 signature of s390_emit_compare_and_swap.
33917 (s390_expand_cs_hqi): Likewise, make static.
33918 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33919 compare-and-swap, in some cases.
33920 (s390_expand_cs): Wrapper function.
33921 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33923 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33924 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33925 patterns for small and large integers. Forbid symref memory operands.
33926 Move expander to s390.c. Require cc register.
33927 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33928 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33929 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33930 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33931 symref memory operands. Remove CC mode and call s390_match_ccmode
33933 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33935 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33937 * config/s390/s390.md (define_peephole2): New peephole to help
33938 combining the load-and-test pattern with volatile memory.
33940 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33942 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33943 with CCZmode for TARGET_Z196.
33945 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33947 PR rtl-optimization/80501
33948 * combine.c (make_compound_operation_int): Set subreg_code to SET
33949 even for AND with mask of the sign bit of mode.
33951 PR rtl-optimization/80500
33952 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33953 sum's initial value.
33955 2017-04-25 Julian Brown <julian@codesourcery.com>
33956 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33958 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33960 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33962 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33964 2017-04-25 Julian Brown <julian@codesourcery.com>
33965 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33967 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33968 (thunderx2t99_sha): New Reservation.
33970 2017-04-25 Julian Brown <julian@codesourcery.com>
33971 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33973 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33974 type for 1-element load.
33976 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33978 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33980 2017-04-24 Martin Jambor <mjambor@suse.cz>
33982 PR tree-optimization/80293
33983 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33984 char arrays not totally scalarizable if it is false.
33985 (analyze_all_variable_accesses): Pass correct value in the new
33986 parameter. Add a statistics counter.
33988 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33990 PR middle-end/79931
33991 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33993 2017-04-24 Richard Biener <rguenther@suse.de>
33995 PR tree-optimization/80494
33996 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33997 out for complex types.
33999 2017-04-24 Richard Biener <rguenther@suse.de>
34001 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34002 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34003 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34004 (DFS): Adjust and never fail.
34005 (sccvn_dom_walker::fail): Remove.
34006 (sccvn_dom_walker::before_dom_children): Adjust.
34007 (run_scc_vn): Likewise and never fail.
34008 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34009 (pass_fre::execute): Likewise.
34011 2017-04-24 Richard Biener <rguenther@suse.de>
34013 PR tree-optimization/79725
34014 * tree-ssa-sink.c (statement_sink_location): Return whether
34015 failure reason was zero uses. Move that check later.
34016 (sink_code_in_bb): Deal with zero uses by removing the stmt
34019 2017-04-24 Richard Biener <rguenther@suse.de>
34022 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34023 pointer-based references.
34025 2017-04-24 Richard Biener <rguenther@suse.de>
34028 * pass_manager.h (pass_manager::operator new): Remove.
34029 (pass_manager::operator delete): Likewise.
34030 * passes.c (pass_manager::operator new): Remove.
34031 (pass_manager::operator delete): Likewise.
34032 (pass_manager::pass_manager): Zero individual pass members.
34034 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34037 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34038 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34039 Check "XEXP (src, 1)" operand here.
34040 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34041 Check "XEXP (src, 1)" operand here.
34042 (dimode_scalar_chain::make_vector_copies): Detect count register
34043 of a shift instruction. Zero extend count register from QImode
34044 to DImode to satisfy vector shift pattern count operand predicate.
34045 Substitute vector shift count operand with a DImode copy.
34046 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34049 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34051 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34052 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34053 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34054 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34055 (UNSPEC_NOREX_MEM): Remove definition.
34057 2017-04-21 Richard Biener <rguenther@suse.de>
34059 PR tree-optimization/79547
34060 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34061 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34062 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34063 without any constraints.
34065 2017-04-21 Richard Biener <rguenther@suse.de>
34067 PR tree-optimization/78847
34068 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34070 2017-04-21 Richard Biener <rguenther@suse.de>
34072 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34073 (build_distinct_type_copy): Likewise.
34074 (build_variant_type_copy): Likewise.
34075 * tree.c (build_qualified_type): Pass down mem-stat info.
34076 (build_distinct_type_copy): Likewise.
34077 (build_variant_type_copy): Likewise.
34079 2017-04-21 Richard Biener <rguenther@suse.de>
34081 PR tree-optimization/80237
34082 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34084 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34085 for a simplified result.
34087 2016-04-21 Richard Biener <rguenther@suse.de>
34089 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34090 sth as strict as a simple_iv but a chrec without symbols and an
34091 operand defined in the loop we are peeling (and not some subloop).
34092 (propagate_constants_for_unrolling): Propagate all constants.
34094 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34097 * config/i386/i386.c (print_reg): Remove assert for disalowed
34098 regno values, call output_operand_lossage instead.
34100 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34103 * config/i386/constraints.md (Yc): New register constraint.
34104 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34105 Use Yc constraint for alternative 2 of operand 0. Remove
34106 preferred_for_speed attribute.
34108 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34110 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34111 lastprivate clauses in SIMT case.
34113 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34115 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34117 2017-04-20 Richard Biener <rguenther@suse.de>
34119 PR tree-optimization/57796
34120 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34121 as N scalar stores.
34122 (vect_model_load_cost): Cost gathers as N scalar loads.
34124 2017-04-20 Richard Biener <rguenther@suse.de>
34126 * ggc-page.c (ggc_allocated_p): Rename to ...
34127 (safe_lookup_page_table_entry): ... this and return the lookup
34129 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34131 2017-04-20 Richard Biener <rguenther@suse.de>
34133 PR tree-optimization/80453
34134 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34135 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34136 from the conditions.
34137 (vn_phi_eq): Pass them down.
34138 (vn_phi_lookup): Record them.
34139 (vn_phi_insert): Likewise.
34141 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34143 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34144 uninitialized variable warning to avoid buffer overrun.
34146 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34149 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34150 is suppressed for '{ 0 }' in C.
34152 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34154 * BASE-VER: Set to 8.0.0.
34156 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34158 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34159 priority .init_array and .fini_array section with SECTION_NOTYPE
34162 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34164 PR middle-end/80423
34165 * tree.h (build_array_type): Add typeless_storage default argument.
34166 * tree.c (type_cache_hasher::equal): Also compare
34167 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34168 (build_array_type): Add typeless_storage argument, set
34169 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34171 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34172 (build_array_type): Likewise. Add typeless_storage argument.
34174 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34175 Jakub Jelinek <jakub@redhat.com>
34177 PR tree-optimization/80426
34178 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34179 operation on symbolic operands, also compute the overflow for the
34180 invariant part when the operation degenerates into a negation.
34182 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34185 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34186 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34189 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34191 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34194 * config/avr/avr.c (tree.h): Include it.
34195 (cgraph.h): Include it.
34196 (avr_encode_section_info): Don't warn for uninitialized progmem
34197 variable if it's just an alias.
34199 2017-04-19 Richard Biener <rguenther@suse.de>
34202 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34203 when needed by AutoPGO.
34205 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34208 * doc/lto.texi: Remove an extra 'that'.
34210 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34212 PR rtl-optimization/80429
34213 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34214 are only used in debug insns.
34216 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34217 Vladimir Makarov <vmakarov@redhat.com>
34219 * config/sparc/predicates.md (input_operand): Add comment. Return
34220 true for any memory operand when LRA is in progress.
34221 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34223 2017-04-18 Jeff Law <law@redhat.com>
34226 * mips.md ({return,simple_return}_internal): Do not overwrite
34229 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34231 PR tree-optimization/80443
34232 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34233 instead of adding 1, subtract -1 and similarly instead of subtracting
34236 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34238 PR rtl-optimization/80357
34239 * haifa-sched.c (tmp_bitmap): New variable.
34240 (model_recompute): Handle duplicate use records.
34241 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34242 (free_global_sched_pressure_data): Free it.
34244 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34247 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34248 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34249 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34250 instead of SYSTEM_HEADER_DIR.
34252 2017-04-18 Jeff Law <law@redhat.com>
34254 PR middle-end/80422
34255 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34256 predecessors after walking up the insn chain.
34258 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34261 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34262 sizetype type into debug info.
34264 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34267 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34268 unneeded test for TARGET_UPPER_REGS_SF.
34269 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34271 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34274 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34275 instead of gsi_after_labels.
34277 2017-04-18 Jeff Law <law@redhat.com>
34279 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34283 2017-04-13 Jeff Law <law@redhat.com>
34284 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34285 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34287 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34290 * config/avr/avr.c (intl.h): Include it.
34291 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34293 2017-04-18 Martin Liska <mliska@suse.cz>
34295 PR gcov-profile/78783
34296 * gcov-tool.c (gcov_output_files): Validate that destination
34297 file is either removed by the tool or by a user.
34299 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34300 Guy Benyei <guybe@mellanox.com>
34302 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34303 block, and do not negate it, the stored id is already negative.
34305 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34307 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34309 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34312 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34313 masks of options that should be turned off if the VSX vector
34314 options are turned off.
34315 (OTHER_P8_VECTOR_MASKS): Likewise.
34316 (OTHER_VSX_VECTOR_MASKS): Likewise.
34317 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34318 rs6000_disable_incompatible_switches to validate no type switches
34320 (rs6000_incompatible_switch): New function to disallow turning on
34321 other vector options if -mno-vsx, -mno-power8-vector, or
34322 -mno-power9-vector are specified.
34324 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34326 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34328 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34330 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34331 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34332 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34333 (ARG_POINTER_CFA_OFFSET): Likewise.
34335 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34337 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34338 conditions to take advantage of various optimizations.
34340 2017-04-13 Jeff Law <law@redhat.com>
34342 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34343 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34344 (zero_extendsidi2_dext): Likewise.
34346 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34349 * fold-const.c (fold_ternary_loc): Revert
34350 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34353 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34355 PR rtl-optimization/80343
34356 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34357 new scratch pseudo.
34359 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34362 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34363 to ubsan_encode_value.
34365 2017-04-13 Jeff Law <law@redhat.com>
34367 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34368 appearing in DEBUG_INSNs.
34370 2017-04-13 Martin Liska <mliska@suse.cz>
34372 PR gcov-profile/80413
34373 * gcov-io.c (gcov_write_string): Copy to buffer just when
34374 allocated size is greater than zero.
34376 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34379 * dwarf2out.c (decls_for_scope): Ignore declarations of
34380 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34382 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34385 * ipa-visibility.c (non_local_p): Fix typos.
34386 (localize_node): When localizing symbol in same comdat group,
34387 dissolve the group only when we know external symbols are going
34389 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34391 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34393 PR tree-optimization/79390
34394 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34395 order does not result in usable sequence, retry with reversed operand
34401 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34402 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34403 op0 instead of fold_convert_loc (loc, type, arg0).
34405 2017-04-12 Jeff Law <law@redhat.com>
34407 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34408 has a delay slot in the generated code.
34410 * config/cris/cris.md (cris_preferred_reload_class): Return
34411 GENNONACR_REGS rather than GENERAL_REGS.
34413 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34416 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34417 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34418 signedness of the result type.
34420 2017-04-12 Richard Biener <rguenther@suse.de>
34421 Jeff Law <law@redhat.com>
34423 PR tree-optimization/80359
34424 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34425 trim stores to TARGET_MEM_REFs.
34427 2017-04-12 Richard Biener <rguenther@suse.de>
34429 PR tree-optimization/79390
34430 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34431 threading case even more.
34433 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34436 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34437 for quad_address_p for TImode, instead of just not indexed_address.
34439 2017-04-12 Richard Biener <rguenther@suse.de>
34440 Bernd Edlinger <bernd.edlinger@hotmail.de>
34442 PR middle-end/79671
34443 * alias.c (component_uses_parent_alias_set_from): Handle
34444 TYPE_TYPELESS_STORAGE.
34445 (get_alias_set): Likewise.
34446 * tree-core.h (tree_type_common): Add typeless_storage flag.
34447 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34448 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34449 for types containing members with TYPE_TYPELESS_STORAGE.
34450 (place_field): Likewise.
34451 (layout_type): Likewise for ARRAY_TYPE.
34452 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34453 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34454 TYPE_TYPELESS_STORAGE.
34455 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34457 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34460 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34461 first argument to type.
34463 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34467 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34468 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34469 (rs6000_expand_binop_builtin): Likewise.
34470 (rs6000_expand_ternop_builtin): Likewise; also add missing
34471 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34472 vshasigma built-ins.
34473 * doc/extend.texi: Document that vec_xxpermdi's third argument
34474 must be a constant.
34476 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34478 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34479 Use shift_const cost parameter when calculating gain of STV shifts.
34481 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34483 PR rtl-optimization/70478
34484 * lra-constraints.c (process_alt_operands): Check memory for
34485 disfavoring memory insn operand.
34487 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34489 PR middle-end/80100
34490 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34491 left shift in unsigned HOST_WIDE_INT type.
34493 PR rtl-optimization/80385
34494 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34495 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34498 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34499 if they have any depend clauses.
34501 2017-04-11 Martin Liska <mliska@suse.cz>
34504 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34505 * ipa-split.c (split_function): Create a local comdat symbol
34506 if caller is in a comdat group.
34508 2017-04-11 Martin Liska <mliska@suse.cz>
34511 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34514 2017-04-11 Martin Sebor <msebor@redhat.com>
34516 PR middle-end/80364
34517 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34518 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34520 (directive::set_width, directive::set_precision, format_character):
34522 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34525 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34528 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34529 conflict, set target->arch_name instead of target->cpu_name.
34531 2017-04-11 Richard Biener <rguenther@suse.de>
34533 PR tree-optimization/80374
34534 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34535 build_zero_cst, remove fold_convertible_p check again.
34537 2017-04-11 Martin Liska <mliska@suse.cz>
34540 * ubsan.c (instrument_object_size): Do not instrument register
34543 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34546 * config/i386/i386-builtin-types.def
34547 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34548 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34549 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34550 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34551 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34552 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34553 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34554 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34555 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34556 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34557 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34558 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34559 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34560 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34561 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34562 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34563 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34564 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34565 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34566 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34567 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34568 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34569 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34570 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34571 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34572 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34573 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34574 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34575 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34576 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34577 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34578 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34579 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34580 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34581 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34582 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34583 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34584 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34585 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34586 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34587 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34588 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34589 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34590 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34591 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34592 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34594 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34595 flag to second_arg_count, handle 4 argument function type _COUNT
34596 aliases, handle second_arg_count on second argument rather than last.
34598 2017-04-10 Jeff Law <law@redhat.com>
34600 PR tree-optimization/80374
34601 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34602 record anything if we can not convert integer_zero_node to the
34605 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34608 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34609 Enhance special handling given to the TARGET_P9_MINMAX option in
34610 relation to certain other options.
34612 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34614 PR tree-optimization/80153
34615 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34616 remove POINTER_PLUS_EXPR's base part directly, rather than through
34619 2017-04-10 Richard Biener <rguenther@suse.de>
34620 Bin Cheng <bin.cheng@arm.com>
34622 PR tree-optimization/80153
34623 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34624 the first element of pointer type aff_tree. Build result expr in
34626 (add_elt_to_tree): Convert to type unconditionally. Remove other
34627 fold_convert calls.
34628 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34629 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34631 2017-04-10 Richard Biener <rguenther@suse.de>
34633 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34636 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34638 PR rtl-optimization/70478
34639 * lra-constraints.c (curr_small_class_check): New.
34640 (update_and_check_small_class_inputs): New.
34641 (process_alt_operands): Update curr_small_class_check. Disfavor
34642 alternative insn memory operands. Check available regs for small
34645 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34648 * config/mips/mips.opt (-mvirt): Update description.
34649 * doc/invoke.texi (-mvirt): Likewise.
34651 2017-04-10 Richard Biener <rguenther@suse.de>
34653 PR middle-end/80362
34654 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34655 looking for NEGATE_EXPR in -A / -B to A / B folding.
34657 2017-04-10 Martin Liska <mliska@suse.cz>
34659 PR gcov-profile/80224
34660 * gcov.c (print_usage): Fix usage string.
34661 (get_gcov_intermediate_filename): Remove.
34662 (output_gcov_file): Use both for normal and intermediate format.
34663 (generate_results): Do not initialize special file for
34664 intermediate format.
34666 2017-04-10 Richard Biener <rguenther@suse.de>
34668 PR tree-optimization/80304
34669 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34672 2017-04-10 Nathan Sidwell <nathan@acm.org>
34675 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34676 (rs6000_init_builtins): Use it.
34678 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34680 * config/arm/arm.md (<mrc>): Add mode to SET source.
34681 (<mrrc>): Likewise.
34683 2017-04-10 Richard Biener <rguenther@suse.de>
34685 PR middle-end/80344
34686 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34688 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34691 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34692 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34693 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34694 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34695 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34696 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34697 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34698 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34699 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34700 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34701 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34702 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34703 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34704 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34705 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34706 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34707 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34708 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34709 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34710 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34711 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34712 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34713 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34715 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34717 PR rtl-optimization/70478
34718 * lra-constraints.c: Reverse the last patch.
34720 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34722 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34723 Add comment for WCHAR_T.
34725 2017-04-08 Martin Liska <mliska@suse.cz>
34728 2017-04-07 Martin Liska <mliska@suse.cz>
34731 * ipa-split.c (split_function): Add function part to a same comdat
34734 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34737 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34739 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34741 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34742 (vec_pairing): ...to this.
34743 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34744 (rs6000_sched_init): Adjust for name change.
34745 (struct rs6000_sched_context): Likewise.
34746 (rs6000_init_sched_context): Likewise.
34747 (rs6000_set_sched_context): Likewise.
34749 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34755 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34757 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34758 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34759 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34761 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34763 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34765 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34767 PR rtl-optimization/70703
34768 * ira-color.c (update_conflict_hard_regno_costs): Use
34769 int64_t instead of HOST_WIDE_INT.
34771 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34773 PR rtl-optimization/70478
34774 * lra-constraints.c (process_alt_operands): Disfavor alternative
34775 insn memory operands.
34777 2017-04-07 Jeff Law <law@redhat.com>
34779 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34780 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34782 2017-04-07 Martin Liska <mliska@suse.cz>
34785 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34786 Show error message instead of an ICE.
34788 2017-04-07 Martin Liska <mliska@suse.cz>
34791 * ipa-split.c (split_function): Add function part to a same comdat
34794 2017-04-07 Richard Biener <rguenther@suse.de>
34796 PR middle-end/80341
34797 * tree.c (get_unwidened): Also handle ! for_type case for
34799 * convert.c (do_narrow): Split out from ...
34800 (convert_to_integer_1): ... here. Do not pass final truncation
34801 type to get_unwidened for TRUNC_DIV_EXPR.
34803 2017-04-07 Richard Biener <rguenther@suse.de>
34805 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34807 (aff_combination_const): Adjust.
34808 (aff_combination_scale): Likewise.
34809 (aff_combination_add_elt): Likewise.
34810 (aff_combination_add_cst): Likewise.
34811 (aff_combination_convert): Likewise.
34812 (add_elt_to_tree): Likewise. Remove unused argument.
34813 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34815 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34817 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34819 * config/arm/arm.c (arm_default_short_enums): Use
34820 ARM_DEFAULT_SHORT_ENUMS.
34821 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34823 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34826 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34827 members with redundant out-of-class redeclaration.
34829 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34832 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34833 * config/i386/i386.md (*zero_extendsidi2):
34834 Add (?*x,*x) and (?*v,*v) alternatives.
34836 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34839 * config/i386/i386.c (ix86_expand_builtin)
34840 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34841 mode from insn data. Convert operands to insn operand mode.
34842 Copy operands that don't satisfy insn predicate to a register.
34844 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34846 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34849 2017-04-06 Richard Biener <rguenther@suse.de>
34851 PR tree-optimization/80334
34852 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34853 preserve alignment of accesses.
34855 2017-04-06 Richard Biener <rguenther@suse.de>
34857 PR tree-optimization/80262
34858 * tree-sra.c (build_ref_for_offset): Preserve address-space
34860 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34861 Drop useless address-space information on MEM_REF offsets.
34863 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34865 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34867 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34869 PR rtl-optimization/70703
34870 * ira-color.c (update_conflict_hard_regno_costs): Use
34871 HOST_WIDE_INT instead of long.
34873 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34876 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34877 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34879 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34880 for x86_64 target. Handle -m3dnowa option.
34882 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34884 PR rtl-optimization/70703
34885 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34886 (update_conflict_hard_regno_costs): Use long instead of unsigned
34887 arithmetic for cost calculation.
34889 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34890 Bernd Edlinger <bernd.edlinger@hotmail.de>
34893 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34896 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34899 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34900 ptr_mode with Pmode throughout.
34901 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34902 into probe_stack_range and use DImode.
34904 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34907 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34908 call_eh_return is true.
34910 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34912 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34913 Initialize last_match_fntype_index.
34915 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34918 * tree-nvr.c: Include internal-fn.h.
34919 (pass_return_slot::execute): Ignore internal calls without
34922 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34923 Richard Biener <rguenther@suse.de>
34926 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34927 captures used multiple times, except for the last use.
34928 * generic-match-head.c: Include gimplify.h.
34930 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34932 PR tree-optimization/79390
34933 * target.h (struct noce_if_info): Declare.
34934 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34935 * target.def (noce_conversion_profitable_p): New target hook.
34936 * ifcvt.h (struct noce_if_info): New type, moved from ...
34937 * ifcvt.c (struct noce_if_info): ... here.
34938 (noce_conversion_profitable_p): Renamed to ...
34939 (default_noce_conversion_profitable_p): ... this. No longer
34941 (noce_try_store_flag_constants, noce_try_addcc,
34942 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34943 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34944 instead of noce_conversion_profitable_p.
34945 * config/i386/i386.c: Include ifcvt.h.
34946 (ix86_option_override_internal): Don't override
34947 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34948 (ix86_noce_conversion_profitable_p): New function.
34949 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34950 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34951 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34952 * doc/tm.texi: Regenerated.
34954 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34956 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34959 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34962 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34963 instructions for small multiply cores.
34965 2017-04-04 Jeff Law <law@redhat.com>
34967 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34969 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34970 that are not set by the loop over the elements.
34972 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34975 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34976 int mode, convert_modes it to mode as unsigned, otherwise use
34977 lowpart_subreg to mode rather than SImode.
34978 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34979 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34980 Use DImode instead of SImode for the shift count operand.
34981 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34984 2017-04-04 Richard Biener <rguenther@suse.de>
34986 PR middle-end/80281
34987 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34988 arithmetic done for the negate or the plus. Simplify.
34989 (A - (-B) -> A + B): Likewise.
34990 * fold-const.c (split_tree): Make sure to not negate pointers.
34992 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34994 PR rtl-optimization/60818
34995 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34996 a compare of comparisons with the thing compared if this results
34997 in a different machine mode.
34999 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35001 * alias.c (base_alias_check): Fix typo in comment.
35002 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35003 * cgraphunit.c (symbol_table::compile): Likewise.
35004 * collect2.c (maybe_run_lto_and_relink): Likewise.
35005 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35006 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35007 * config/avr/avr.c (avr_map_op_t): Likewise.
35008 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35009 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35010 * config/epiphany/epiphany.md (movcc): Likewise.
35011 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35012 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35014 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35015 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35016 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35017 * config/sh/sh.c (sh_rtx_costs): Likewise.
35018 * fold-const.c (fold_truth_andor): Likewise.
35019 * genautomata.c (collapse_flag): Likewise.
35020 * gengtype.h (struct type::u::s): Likewise.
35021 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35022 * input.c (FORMAT_AMOUNT): Likewise.
35023 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35024 (known_aggs_to_agg_replacement_list): Likewise.
35025 * ipa-inline-analysis.c: Likewise.
35026 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35027 * ipa-polymorphic-call.c
35028 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35029 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35030 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35032 * modulo-sched.c (apply_reg_moves): Likewise.
35033 * omp-expand.c (build_omp_regions_1): Likewise.
35034 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35035 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35036 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35037 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35038 * value-prof.c: Likewise.
35039 * var-tracking.c (val_reset): Likewise.
35041 2017-04-03 Richard Biener <rguenther@suse.de>
35043 PR tree-optimization/80275
35044 * fold-const.c (split_address_to_core_and_offset): Handle
35047 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35049 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35050 descriptors is at least equal to that of functions.
35052 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35054 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35056 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35059 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35060 (mov<IMOD4:mode>): New expander.
35061 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35063 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35065 PR rtl-optimization/79405
35066 * fwprop.c (propagations_left): New variable.
35067 (forward_propagate_into): Decrement it.
35068 (fwprop_init): Initialize it.
35069 (fw_prop): If the variable has reached zero, stop propagating.
35070 (fwprop_addr): Ditto.
35072 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35075 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35076 a FUNCTION_DECL, pass it as decl instead of origin to
35079 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35081 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35084 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35087 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35088 TARGET_VSX_SMALL_INTEGER.
35090 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35092 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35093 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35095 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35097 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35098 extraction from odd-numbered MSA register.
35100 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35102 PR middle-end/80173
35103 * expmed.c (store_bit_field_1): Don't attempt to create
35104 a word subreg out of hard registers wider than word if they
35105 have HARD_REGNO_NREGS of 1 for their mode.
35107 PR middle-end/80163
35108 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35109 conversions to integer types wider than word and pointer.
35112 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35113 (rtx_equal_for_cselib_p): Pass 0 to it.
35114 * cselib.c (cselib_hasher::equal): Likewise.
35115 (rtx_equal_for_cselib_1): Add depth argument. If depth
35116 is 128, don't look up VALUE locs and punt. Increment
35117 depth in recursive calls when walking VALUE locs.
35119 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35121 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35122 (make_gcov_file_name): Use the canonical path name for generating
35124 (read_line): Fix handling of files with ascii null bytes.
35126 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35128 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35129 to initialise a vector register instead
35130 of using a const_int.
35132 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35134 PR translation/80189
35135 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35136 diagnostic messages.
35138 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35141 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35142 (dfp_diex_<mode>): Update mode of operand 1.
35143 * doc/extend.texi (dxex, dxexq): Document change to return type.
35144 (diex, diexq): Document change to argument type.
35146 2017-03-30 Martin Jambor <mjambor@suse.cz>
35149 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35150 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35151 it reflects the signature changes performed at the callee side.
35152 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35153 to cgraph_build_function_type_skip_args.
35154 (build_function_decl_skip_args): Adjust call to the above function.
35156 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35159 * config/i386/sse.md
35160 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35161 register as dest whenever it is a MEM not rtx_equal_p to the
35162 corresponding dup operand, and when forcing into reg move the
35163 reg into the memory afterwards.
35164 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35165 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35166 for the force_reg mode.
35167 (avx512vl_vextractf128<mode>): Use register as dest either
35168 always when a MEM, or when it is a MEM not rtx_equal_p to the
35169 corresponding dup operand, or even not when it is a CONST_VECTOR
35170 depending on the mode and lo vs. hi.
35171 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35173 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35174 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35175 Likewise. Require that operands[2] is even.
35176 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35177 Remove extraneous parens. Require that operands[2] is a multiple
35179 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35180 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35181 disallow memory then.
35183 2017-03-30 Richard Biener <rguenther@suse.de>
35185 PR tree-optimization/77498
35186 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35187 to non-constants over backedges.
35189 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35191 PR rtl-optimization/80233
35192 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35193 as last_combined_insn. Do not test for BARRIER_P separately.
35195 2017-03-29 Andreas Schwab <schwab@suse.de>
35198 * calls.c (prepare_call_address): Convert funexp to Pmode before
35199 copying to temp reg.
35201 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35203 PR tree-optimization/80158
35204 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35205 Handle possible future case of more than one alternate
35207 (replace_rhs_if_not_dup): Likewise.
35208 (replace_one_candidate): Likewise.
35210 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35212 PR rtl-optimization/80193
35213 * ira.c (ira): Do not check allocation for LRA.
35215 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35217 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35218 (nvptx_output_simt_exit): Declare.
35219 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35220 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35221 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35222 (nvptx_declare_function_name): ...here. Emit declaration of local
35223 memory space buffer for omp_simt_enter insn.
35224 (nvptx_output_unisimt_switch): New.
35225 (nvptx_output_softstack_switch): New.
35226 (nvptx_output_simt_enter): New.
35227 (nvptx_output_simt_exit): New.
35228 * config/nvptx/nvptx.h (struct machine_function): New fields
35229 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35230 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35231 (UNSPECV_SIMT_EXIT): Ditto.
35232 (omp_simt_enter_insn): New insn.
35233 (omp_simt_enter): New expansion.
35234 (omp_simt_exit): New insn.
35235 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35237 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35238 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35239 (expand_GOMP_SIMT_EXIT): New.
35240 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35241 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35242 (GOMP_SIMT_EXIT): Ditto.
35243 * target-insns.def (omp_simt_enter): New insn.
35244 (omp_simt_exit): Ditto.
35245 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35247 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35248 (lower_rec_input_clauses): Likewise.
35249 (lower_lastprivate_clauses): Handle SIMT privatization.
35251 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35252 (ompdevlow_adjust_simt_enter): New.
35253 (find_simtpriv_var_op): New.
35254 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35255 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35257 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35258 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35259 (copy_decl_for_dup_finish): Ditto.
35261 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35263 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35266 * config/i386/i386.c (ix86_option_override_internal): Always
35267 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35269 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35271 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35273 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35275 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35276 mark new edge's irreducible flag accordign to it.
35277 (vect_do_peeling): Check loop preheader edge's irreducible flag
35278 and pass it to function slpeel_add_loop_guard.
35280 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35282 PR tree-optimization/80218
35283 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35284 Update block frequencies and counts.
35286 2017-03-28 Richard Biener <rguenther@suse.de>
35288 PR tree-optimization/78644
35289 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35290 of a simplification result we may not use it at all.
35292 2017-03-28 Richard Biener <rguenther@suse.de>
35295 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35296 without arguments, generate default definition of a SSA name.
35298 2017-03-28 Richard Biener <rguenther@suse.de>
35300 PR middle-end/80222
35301 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35302 TYPE_REF_CAN_ALIAS_ALL references.
35303 * fold-const.c (fold_indirect_ref_1): Likewise.
35305 2017-03-28 Martin Liska <mliska@suse.cz>
35308 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35309 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35311 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35312 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35314 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35315 (EXTRA_SPECS): Define.
35316 (SUBTARGET_EXTRA_SPECS): Likewise.
35317 (SUBTARGET_CPP_SPEC): Likewise.
35318 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35319 SUBTARGET_EXTRA_SPECS.
35320 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35322 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35324 * config/arc/simdext.md (vst64_insn): Update pattern.
35325 (vld32wh_insn): Likewise.
35326 (vld32wl_insn): Likewise.
35327 (vld64_insn): Likewise.
35328 (vld32_insn): Likewise.
35330 2017-03-28 Marek Polacek <polacek@redhat.com>
35333 * fold-const.c (fold_comparison): Use protected_set_expr_location
35334 instead of SET_EXPR_LOCATION.
35336 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35338 * tree.c (add_expr): Avoid name lookup warning.
35340 2017-03-27 Jeff Law <law@redhat.com>
35342 PR tree-optimization/80216
35343 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35344 function name. Limit recursion depth.
35345 (record_temporary_equivalences): Corresponding changes.
35347 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35349 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35352 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35355 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35357 * cfgcleanup.c (reg_note_cfa_p): New array.
35358 (insns_have_identical_cfa_notes): New function.
35359 (old_insns_match_p): Don't cross-jump in between /f
35360 and non-/f instructions. If both i1 and i2 are frame related,
35361 verify all CFA notes, their order and content.
35363 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35366 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35367 HImode and SImode with zero extend to DImode to one insn.
35368 (bswap<mode>2_extenddi): Likewise.
35369 (bswapsi2_extenddi): Likewise.
35370 (bswaphi2_extendsi): Likewise.
35371 (bswaphi2): Combine bswap HImode and SImode into one insn.
35372 Separate memory insns from swapping register.
35373 (bswapsi2): Likewise.
35374 (bswap<mode>2): Likewise.
35375 (bswaphi2_internal): Delete, no longer used.
35376 (bswapsi2_internal): Likewise.
35377 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35378 store, and gpr<-gpr swap insns.
35379 (bswap<mode>2_store): Likewise.
35380 (bswaphi2_reg): Register only splitter, combine with the splitter.
35381 (bswaphi2 splitter): Likewise.
35382 (bswapsi2_reg): Likewise.
35383 (bswapsi2 splitter): Likewise.
35384 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35385 the insns into load, store, and register/register insns.
35386 (bswapdi2_ldbrx): Likewise.
35387 (bswapdi2_load): Likewise.
35388 (bswapdi2_store): Likewise.
35389 (bswapdi2_reg): Likewise.
35391 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35393 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35394 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35396 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35399 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35402 special handling for target option conflicts between dform
35403 options (-mpower9-dform, -mpower9-dform-vector,
35404 -mpower9-dform-scalar) and -mno-direct-move.
35406 2017-03-27 Richard Biener <rguenther@suse.de>
35408 PR tree-optimization/80181
35409 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35411 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35413 * config/arc/predicates.md (move_double_src_operand): Replace the
35414 call to move_double_src_operand with a call to address_operand.
35416 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35418 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35419 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35420 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35422 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35424 * config/arc/predicates.md (long_immediate_loadstore_operand):
35425 Consider scaled addresses cases.
35427 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35429 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35430 restored when in interrupt.
35431 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35432 doesn't have delay slot.
35434 2017-03-27 Richard Biener <rguenther@suse.de>
35437 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35438 inlined thunk clones.
35440 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35443 * asan.c (instrument_derefs): Copy over last operand from
35444 original COMPONENT_REF to the new COMPONENT_REF with
35445 DECL_BIT_FIELD_REPRESENTATIVE.
35446 * ubsan.c (instrument_object_size): Likewise.
35448 2017-03-27 Richard Biener <rguenther@suse.de>
35450 PR tree-optimization/80170
35451 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35452 sure DR/SCEV didnt fold in constants we do not see when looking
35453 at the reference base alignment.
35455 2017-03-27 Richard Biener <rguenther@suse.de>
35457 PR middle-end/80171
35458 * gimple-fold.c (fold_ctor_reference): Properly guard against
35459 NULL return value from canonicalize_constructor_val.
35461 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35464 * config/i386/i386.c (ix86_expand_builtin)
35465 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35466 flags reg setting and flags reg using instructions.
35467 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35468 clobbering instructions to zero extend op2.
35470 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35472 * doc/install.texi (Configuration) <--with-aix-soname>:
35473 Update link to AIX ld.
35475 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35477 PR rtl-optimization/80160
35478 PR rtl-optimization/80159
35479 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35480 reg_alternate_class into account.
35482 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35485 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35486 to consider in curr_insn_transform.
35488 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35490 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35491 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35492 and emit_mode_inner.
35494 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35496 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35497 argument to the overloaded builtin variants. Use the new flag to
35498 deprecate certain builtin variants.
35499 * config/s390/s390-builtin-types.def: Add new builtin types.
35500 * config/s390/s390-builtins.h: Support new flags field for
35501 overloaded builtins.
35502 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35503 (s390_macro_to_expand): Enable vector float data type.
35504 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35505 builtins by incrementing the __VEC__ version number.
35506 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35508 (s390_resolve_overloaded_builtin): Emit error messages depending
35509 on the builtin flags.
35510 * config/s390/s390.c (s390_expand_builtin): Support additional
35511 flags argument. Change error message to match the messages
35512 emitted in s390-c.c.
35513 * config/s390/s390.md: New UNSPEC_* constants.
35514 (op_type): Add new instruction types.
35515 * config/s390/vecintrin.h: Add new builtins and test data class
35517 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35518 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35519 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35520 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35521 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35522 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35524 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35525 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35526 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35527 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35529 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35530 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35531 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35532 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35534 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35535 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35536 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35537 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35538 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35539 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35540 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35542 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35543 ("vec_scatter_element<V_HW_4:mode>_DI")
35544 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35545 ("vec_fpint<mode>", "vflls")
35546 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35547 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35548 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35549 ("*vec_cmphe<mode>_cc"): ... these.
35551 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35552 mode constant instead of magic value.
35554 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35556 * config/s390/s390.c (s390_expand_vec_compare): Support other
35557 vector floating point modes than just V2DF.
35558 (s390_expand_vcond): Likewise.
35559 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35560 (s390_cannot_change_mode_class): Prevent mode changes between TF
35561 and V1TF in vector registers.
35562 * config/s390/s390.md (DF, SF): New mode attributes.
35563 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35564 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35565 SFmode support for VRs.
35566 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35568 (VFT, VF_HW): New mode iterators.
35569 (vw, sdx): New mode attributes.
35570 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35571 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35572 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35573 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35574 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35575 also the new vector floating point modes. Renaming to ...
35577 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35578 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35579 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35580 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35581 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35582 ("vec_unordered<mode>"): ... these.
35584 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35585 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35586 ("*vec_extendv2df"): New insn definitions.
35588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35590 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35591 ("mulditi3_2", "*muldi3_sign"): New patterns.
35592 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35593 rename the pattern definition.
35595 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35597 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35599 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35601 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35603 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35604 instruction if possible.
35605 * config/s390/vector.md (vec_halfnumelts): New mode
35607 ("*vec_vllezlf<mode>"): New pattern.
35609 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35611 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35612 ("popcountv4si2", "popcountv2di2"): Rename to ...
35613 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35614 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35616 ("popcount<mode>2_vxe"): New pattern.
35618 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35620 * common/config/s390/s390-common.c (processor_flags_table): Add
35622 * config.gcc: Add arch12.
35623 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35624 Default to arch12 for unknown CPU model numbers.
35625 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35626 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35627 PROCESSOR_max sanity check.
35628 * config/s390/s390-opts.h (enum processor_type): Add
35630 * config/s390/s390.c (processor_table): Add arch12.
35631 (s390_expand_builtin): Add check for B_VXE flag.
35632 (s390_issue_rate): Add PROCESSOR_ARCH12.
35633 (s390_get_sched_attrmask): Likewise.
35634 (s390_get_unit_mask): Likewise.
35635 (s390_sched_score): Enable z13 scheduling for arch12.
35636 (s390_sched_reorder): Likewise.
35637 (s390_sched_variable_issue): Likewise.
35638 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35640 (s390_tune_attr): Use z13 scheduling also for arch12.
35641 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35642 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35643 (TARGET_VXE_P): New macros.
35644 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35645 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35646 * config/s390/s390.opt: Add arch12 as processor_type.
35648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35650 * config/s390/s390.md
35651 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35652 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35653 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35655 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35656 Rename expanders to ...
35658 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35659 ("fixuns_truncdddi2_emu"): ... these.
35661 ("fixuns_trunc<mode>si2_emu"): New expander.
35663 ("*fixuns_truncdfdi2_z13"): Rename to ...
35664 ("*fixuns_truncdfdi2_vx"): ... this.
35666 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35668 * config/s390/2964.md: Remove the single element vector compare
35669 instructions which are no longer used.
35670 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35672 (s390_canonicalize_comparison): Remove handling of DFmode
35674 (s390_expand_vec_compare_scalar): Remove function.
35675 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35676 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35678 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35680 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35682 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35684 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35685 will anyway by matched by mov<mode>_64dfp.
35687 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35689 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35690 vlef/vstef. Add missing operand to vleif.
35692 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35694 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35695 pair for all vector types with 64 bit elements.
35696 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35697 * config/s390/vector.md (V_HW_64): ... here.
35698 (V_128_NOSINGLE): New mode iterator.
35699 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35700 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35701 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35702 ("*vec_load_pairv2di"): Change to ...
35703 ("*vec_load_pair<mode>"): ... this one.
35705 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35707 * config/s390/constraints.md: Add comments.
35708 (jKK): Reject element sizes > 8 bytes.
35709 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35711 * config/s390/s390.md: Add the s_operand checks formerly in
35712 s390_split_ok_p to various splitters where they are still
35714 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35715 for 128 bit vectors. Plus two splitters.
35717 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35719 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35722 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35725 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35726 error if the boundary argument is not constant.
35728 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35730 PR rtl-optimization/80112
35731 * loop-doloop.c (doloop_condition_get): Don't check condition
35732 if cmp isn't SET with IF_THEN_ELSE src.
35734 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35736 PR tree-optimization/80158
35737 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35738 replacing a candidate statement, also replace it for the
35739 candidate's alternate interpretation.
35740 (replace_rhs_if_not_dup): Likewise.
35741 (replace_one_candidate): Likewise.
35743 2017-03-24 Richard Biener <rguenther@suse.de>
35745 PR tree-optimization/80167
35746 * graphite-isl-ast-to-gimple.c
35747 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35749 (translate_isl_ast_to_gimple::get_rename): Likewise.
35751 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35754 handling of certain combinations of target options, including the
35755 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35756 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35758 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35761 * config/arm/arm.md (*load_multiple): Add reload_completed to
35762 matching condition.
35764 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35765 Richard Biener <rguenth@suse.de>
35767 PR tree-optimization/79908
35768 PR tree-optimization/80136
35769 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35770 been cast away, gimplify_and_add suffices.
35772 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35774 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35776 2017-03-23 Richard Biener <rguenther@suse.de>
35778 PR tree-optimization/80032
35779 * gimplify.c (gimple_push_cleanup): Forced unconditional
35780 cleanups still have to go to the conditional_cleanups
35783 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35785 PR tree-optimization/80072
35786 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35788 (next_operand_entry_id): Change type to unsigned int.
35789 (sort_by_operand_rank): Make sure to return the right return value
35790 even if unsigned fields are bigger than INT_MAX.
35791 (struct oecount): Change cnt and id type to unsigned int.
35792 (oecount_hasher::equal): Formatting fix.
35793 (oecount_cmp): Make sure to return the right return value
35794 even if unsigned fields are bigger than INT_MAX.
35795 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35798 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35799 TREE_READONLY on result if writing it more than once.
35802 * doc/invoke.texi (-fsanitize=thread): Document that with
35803 -fnon-call-exceptions atomics are not able to throw
35807 * tsan.c: Include tree-eh.h.
35808 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35809 maybe_clean_or_replace_eh_stmt where needed.
35810 (instrument_memory_accesses): Add cfg_changed argument.
35811 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35812 if it returned true.
35813 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35815 PR rtl-optimization/63191
35816 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35817 wrapper function, moved the whole old content into ...
35818 (ix86_delegitimize_address_1): ... this. New inline function.
35819 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35820 true as last argument instead of ix86_delegitimize_address.
35822 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35824 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35825 cortexa57_branch_cost.
35827 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35829 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35831 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35834 * doc/md.texi (Constraints): Document wA constraint.
35835 * config/rs6000/constraints.md (wA): New.
35836 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35837 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35838 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35839 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35841 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35844 * gimplify.c (is_oacc_declared): New function.
35845 (oacc_default_clause): Use it to set default flags for acc declared
35846 variables inside parallel regions.
35847 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35848 declared variables.
35849 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35850 declare attribute to any decl as necessary.
35852 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35855 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35856 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35857 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35858 (arm_arch_lpae): This.
35859 * config/arm/arm.c (arm_arch7ve): Rename into ...
35860 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35861 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35864 2017-03-22 Martin Liska <mliska@suse.cz>
35867 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35868 error message instead of an ICE.
35870 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35872 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35874 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35876 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35881 2017-03-21 Martin Sebor <msebor@redhat.com>
35883 * doc/extend.texi: Use "cannot" instead of "can't."
35884 * doc/hostconfig.texi: Same.
35885 * doc/install.texi: Same.
35886 * doc/invoke.texi: Same.
35887 * doc/loop.texi: Same.
35888 * doc/md.texi: Same.
35889 * doc/objc.texi: Same.
35890 * doc/rtl.texi: Same.
35891 * doc/tm.texi: Same.
35892 * doc/tm.texi.in: Same.
35893 * doc/trouble.texi: Same.
35895 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35898 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35899 (collect_checksum_attributes): Set it.
35900 (die_checksum_ordered): Use it.
35902 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35904 PR tree-optimization/79908
35905 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35906 change: For a VA_ARG whose LHS has been cast away, use
35907 force_gimple_operand to construct the side effects.
35909 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35911 PR translation/80001
35912 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35913 more amenable to translation.
35914 (oacc_loop_auto_partitions): Likewise.
35916 2017-03-21 Marek Polacek <polacek@redhat.com>
35917 Martin Sebor <msebor@redhat.com>
35919 PR tree-optimization/80109
35920 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35921 on INTEGRAL_TYPE_P.
35923 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35924 Segher Boessenkool <segher@kernel.crashing.org>
35927 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35928 check reg_used_between_p between insn and one of succ or succ2
35929 depending on if succ is artificial insn not inserted into insn
35932 2017-03-21 Martin Liska <mliska@suse.cz>
35934 PR gcov-profile/80081
35935 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35936 * doc/gcc.texi: Include gcov-dump stuff.
35937 * doc/gcov-dump.texi: New file.
35939 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35941 PR rtl-optimization/79150
35942 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35943 conditional jump, if the jump is the last insn of the loop.
35945 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35946 Richard Biener <rguenth@suse.de>
35948 PR tree-optimization/79908
35949 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35950 been cast away, use force_gimple_operand to construct the side
35953 2017-03-21 Martin Liska <mliska@suse.cz>
35955 PR libfortran/79956
35956 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35959 2017-03-21 Brad Spengler <spender@grsecurity.net>
35962 * plugin.c (htab_hash_plugin): New function.
35963 (add_new_plugin): Use it and adjust.
35964 (parse_plugin_arg_opt): Adjust.
35965 (init_one_plugin): Likewise.
35967 2017-03-21 Richard Biener <rguenther@suse.de>
35969 PR tree-optimization/80032
35970 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35971 if set force the cleanup to happen unconditionally.
35972 (gimplify_target_expr): Push inserted clobbers with force_uncond
35973 to avoid them being removed by control-dependent DCE.
35975 2017-03-21 Richard Biener <rguenther@suse.de>
35977 PR tree-optimization/80122
35978 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35979 va_arg_pack_len when the inlined call stmt requires pack
35981 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35983 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35986 * tsan.c (instrument_builtin_call): If the memory model argument
35987 is not a constant, assume it is valid.
35990 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35993 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35995 PR rtl-optimization/79910
35996 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35997 if its dest is used by an insn before I2 (other than the combined
35998 insns themselves, which are properly handled already).
36000 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36003 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36005 * combine.c (record_used_regs): New static function.
36006 (try_combine): Handle situations where there is an additional
36007 instruction between I2 and I3 which needs to have a LOG_LINK
36011 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36013 * combine.c (try_combine): Delete redundant i1 test. Call
36014 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36016 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36019 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36020 alternatives 13/14.
36022 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36024 PR tree-optimization/80054
36025 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36026 the optimization if a PHI or any of its arguments is not dominated
36027 by the candidate's basis. Use gphi* rather than gimple* as
36029 (replace_profitable_candidates): Clean up a gimple* variable that
36030 should be a gphi* variable.
36032 2017-03-20 Martin Sebor <msebor@redhat.com>
36035 * doc/extend.texi (attribute constructor): Document present limitation.
36037 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36040 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36041 __POWER9_VECTOR__ #ifdef control, change template definition to
36042 use Power9-specific built-in function.
36043 (vec_any_eq): Likewise.
36044 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36045 to control outcomes from this test.
36046 (vector_ae_<mode>p): For VEC_F modes, likewise.
36048 2017-03-20 Ian Lance Taylor <iant@google.com>
36050 * config/i386/i386.c (ix86_function_regparm): Save an extra
36051 register for -fsplit-stack with DECL_STATIC_CHAIN.
36053 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36056 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36057 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36059 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36061 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36063 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36066 2017-03-20 Marek Polacek <polacek@redhat.com>
36069 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36071 2017-03-20 Richard Biener <rguenther@suse.de>
36073 PR tree-optimization/80113
36074 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36075 allocate extra SSA name for PHI def.
36076 (add_close_phis_to_outer_loops): Likewise.
36077 (add_close_phis_to_merge_points): Likewise.
36078 (copy_loop_close_phi_args): Likewise.
36079 (copy_cond_phi_nodes): Likewise.
36081 2017-03-20 Martin Liska <mliska@suse.cz>
36083 PR middle-end/79753
36084 * tree-chkp.c (chkp_build_returned_bound): Do not build
36085 returned bounds for a LHS that's not a BOUNDED_P type.
36087 2017-03-20 Martin Liska <mliska@suse.cz>
36091 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36092 COMPLEX_CST and VECTOR_CST.
36094 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36097 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36098 target operand. A new splitter adds the clobber statement in case
36099 the target operand is dead anyway.
36101 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36103 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36104 to age-old versions of binutils and glibc.
36106 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36108 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36110 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36112 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36114 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36116 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36117 requirement for binutils 2.13.
36119 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36121 * combine.c (try_combine): Delete redundant i1 test. Call
36122 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36124 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36126 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36127 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36129 <riscv64-*-elf>: Re-arrange section
36130 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36131 <riscv32-*-linux>: Likewise.
36132 <riscv64-*-elf>: Likewise
36133 <riscv64-*-linux>: Likewise.
36135 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36138 * aarch64.opt(verbose-cost-dump): Fix typo.
36140 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36143 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36144 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36146 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36148 * reload.c (find_reloads): When reloading a nonoffsettable address,
36149 use RELOAD_OTHER for it and its address reloads.
36151 PR rtl-optimization/79910
36152 * combine.c (record_used_regs): New static function.
36153 (try_combine): Handle situations where there is an additional
36154 instruction between I2 and I3 which needs to have a LOG_LINK
36157 2017-03-17 Jeff Law <law@redhat.com>
36159 PR tree-optimization/71437
36160 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36161 conditional in the hash table first.
36162 (vrp_dom_walker::before_dom_children): Extract condition from
36163 ASSERT_EXPR. Record condition, its inverion and any implied
36164 conditions as well.
36166 2017-03-17 Marek Polacek <polacek@redhat.com>
36167 Markus Trippelsdorf <markus@trippelsdorf.de>
36169 PR tree-optimization/80079
36170 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36173 2017-03-17 Richard Biener <rguenther@suse.de>
36175 PR middle-end/80075
36176 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36177 Properly verify the LHS before the RHS possibly claims to be
36179 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36182 2017-03-17 Martin Jambor <mjambor@suse.cz>
36184 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36185 (List of -O2 options): Likewise.
36186 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36189 2017-03-17 Tom de Vries <tom@codesourcery.com>
36191 * gcov-dump.c (print_usage): Print bug_report_url.
36193 2017-03-17 Richard Biener <rguenther@suse.de>
36195 PR middle-end/80050
36196 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36197 (parser::peek): Likewise.
36199 2017-03-17 Richard Biener <rguenther@suse.de>
36201 PR tree-optimization/80048
36202 * sese.c (free_sese_info): Properly release rename_map and
36203 copied_bb_map elements.
36205 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36207 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36208 Add linked-list forward and backlinks. Insert on
36209 construction, remove on destruction.
36210 (class pass_store_merging): Add m_stores_head field.
36211 (pass_store_merging::terminate_and_process_all_chains):
36212 Iterate over m_stores_head list.
36213 (pass_store_merging::terminate_all_aliasing_chains):
36215 (pass_store_merging::execute): Check for debug stmts first.
36216 Push new chains onto the m_stores_head stack.
36218 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36221 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36222 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36223 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36225 2017-03-16 Jeff Law <law@redhat.com>
36227 PR tree-optimization/71437
36228 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36229 member function. Implementation moved into after_dom_children
36230 member function and into the threader's thread_outgoing_edges
36232 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36233 some code into new thread_outgoing_edges.
36234 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36235 definition. Simplify marker handling (do it here). Assume we always
36236 have the available expression and the const/copies tables.
36237 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36239 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36240 * tree-vrp.c (equiv_stack): No longer file scoped.
36241 (vrp_dom_walker): New class.
36242 (vrp_dom_walker::before_dom_children): New member function.
36243 (vrp_dom_walker::after_dom_children): Likewise.
36244 (identify_jump_threads): Setup domwalker. Use it rather than
36245 walking edges in a random order by hand. Simplify setup/finalization.
36246 (finalize_jump_threads): Remove.
36247 (vrp_finalize): Do not call identify_jump_threads here.
36248 (execute_vrp): Do it here instead and call thread_through_all_blocks
36251 PR tree-optimization/71437
36252 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36254 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36256 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36257 (dom_opt_dom_walker::thread_across_edge): Remove
36258 handle_dominating_asserts argument. All callers changed.
36259 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36260 changes. Remove calls to lhs_of_dominating_assert. Other
36261 uses of handle_dominating_asserts turn into unconditional code
36262 (simplify_control_stmt_condition_1): Likewise.
36263 (simplify_control_stmt_condition): Likewise.
36264 (thread_through_normal_block, thread_across_edge): Likewise.
36265 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36266 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36267 object if it is not an SSA_NAME.
36268 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36269 before calling into the VRP specific simplifiers.
36270 (identify_jump_threads): Remove handle_dominating_asserts
36273 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36276 * tree-diagnostic.c (default_tree_printer): No longer static.
36277 * tree-diagnostic.h (default_tree_printer): New prototype.
36279 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36281 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36282 Change ins into fmov.
36284 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36286 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36287 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36288 Use h_con constraint for operand 1.
36289 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36290 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36292 2017-03-15 Jeff Law <law@redhat.com>
36294 PR tree-optimization/71437
36295 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36296 (record_temporary_equivalences): Use it.
36298 PR tree-optimization/71437
36299 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36300 tree-ssa-scopedtables.
36301 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36302 (record_conditions, record_cond, vuse_eq): Likewise.
36303 (record_edge_info): Adjust to API tweak of record_conditions.
36304 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36305 (record_temporary_equivalences, optimize_stmt): Likewise.
36306 (eliminate_redundant_computations): Likewise.
36307 (record_equivalences_from_stmt): Likewise.
36308 * tree-ssa-scopedtables.c: Include options.h and params.h.
36309 (vuse_eq): New function, moved from tree-ssa-dom.c
36310 (build_and_record_new_cond): Likewise.
36311 (record_conditions): Likewise. Accept vector of conditions rather
36312 than edge_equivalence structure for first argument.
36313 for the first argument.
36314 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36315 from tree-ssa-dom.c.
36316 (avail_exprs_stack::record_cond): Likewise.
36317 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36318 from tree-ssa-dom.c.
36319 (avail_exprs_stack): Add new member functions lookup_avail_expr
36321 (record_conditions): Declare.
36323 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36326 * lra-constraints.c (process_alt_operands): Increase reject for
36327 reloading an input/output operand.
36329 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36332 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36333 insns to convert from signed/unsigned char/short to IEEE 128-bit
36335 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36337 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36340 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36341 subreg of inner mode for values already in registers.
36343 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36345 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36346 iteration reg is used after the loop.
36348 2017-03-14 Martin Sebor <msebor@redhat.com>
36350 PR tree-optimization/79800
36351 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36352 precision in negative-positive range.
36353 (format_floating): Call non-const overload with adjusted precision.
36355 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36358 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36361 2017-03-14 Martin Sebor <msebor@redhat.com>
36363 PR middle-end/80020
36364 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36365 * builtins.def (aligned_alloc): Use it.
36368 * Makefile.in (GTFILES): Add calls.c.
36369 * calls.c: Include "gt-calls.h".
36371 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36373 PR rtl-optimization/79728
36374 * regs.h (struct target_regs): New field
36375 x_contains_allocatable_regs_of_mode.
36376 (contains_allocatable_regs_of_mode): New macro.
36377 * reginfo.c (init_reg_sets_1): Initialize it, and change
36378 contains_reg_of_mode so it includes global regs as well.
36379 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36380 rather than contains_regs_of_mode.
36382 2017-03-14 Martin Liska <mliska@suse.cz>
36384 * doc/invoke.texi: Document options that can't be combined with
36385 -fcheck-pointer-bounds.
36387 2017-03-14 Martin Liska <mliska@suse.cz>
36389 PR middle-end/79831
36390 * doc/invoke.texi (-Wchkp): Document the option.
36392 2017-03-14 Martin Liska <mliska@suse.cz>
36394 * Makefile.in: Install gcov-dump.
36396 2017-03-14 Martin Liska <mliska@suse.cz>
36398 * multiple_target.c (expand_target_clones): Bail out for
36399 an invalid attribute.
36401 2017-03-14 Richard Biener <rguenther@suse.de>
36403 * alias.c (struct alias_set_entry): Pack properly.
36404 * cfgloop.h (struct loop): Likewise.
36405 * cse.c (struct set): Likewise.
36406 * ipa-utils.c (struct searchc_env): Likewise.
36407 * loop-invariant.c (struct invariant): Likewise.
36408 * lra-remat.c (struct cand): Likewise.
36409 * recog.c (struct change_t): Likewise.
36410 * rtl.h (struct address_info): Likewise.
36411 * symbol-summary.h (function_summary): Likewise.
36412 * tree-loop-distribution.c (struct partition): Likewise.
36413 * tree-object-size.c (struct object_size_info): Likewise.
36414 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36415 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36416 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36417 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36418 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36419 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36420 (struct _stmt_vec_info): Likewise.
36422 2017-03-14 Martin Liska <mliska@suse.cz>
36425 * multiple_target.c (create_dispatcher_calls): Check that
36426 a target can create a function dispatcher.
36428 2017-03-14 Martin Liska <mliska@suse.cz>
36431 * multiple_target.c (expand_target_clones): Drop local.local
36432 flag for default implementation.
36434 2017-03-14 Richard Biener <rguenther@suse.de>
36436 PR tree-optimization/80030
36437 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36439 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36441 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36442 gcc_fallthrough() instead of __attribute__((fallthrough));
36444 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36446 * doc/gcc.texi: Remove "up" link to (DIR).
36447 * doc/gccint.texi: Ditto.
36449 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36451 * doc/install.texi (Specific) <avr>: Remove reference to
36454 2017-03-13 Jeff Law <law@redhat.com>
36456 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36457 attribute rather than comments.
36459 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36460 match_scratch operand is highest.
36462 2017-03-13 Martin Liska <mliska@suse.cz>
36464 PR middle-end/78339
36465 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36466 is a CHKP clone, use original declaration.
36468 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36470 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36471 (arc_conditional_register_usage): Use a different allocation order
36472 when optimizing for size.
36473 * common/config/arc/arc-common.c (arc_option_optimization_table):
36474 Section anchors default on when optimizing for size.
36476 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36478 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36480 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36482 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36483 * config/arc/arc.md (cpu_facility): Add cd variant.
36484 (*movqi_insn): Add code density variant.
36485 (*movhi_insn): Likewise.
36486 (*movqi_insn): Likewise.
36487 (*addsi3_mixed): Likewise.
36488 (subsi3_insn): Likewise.
36490 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36492 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36494 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36496 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36497 expressions with MINUS and UNARY ops.
36499 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36502 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36504 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36505 between vec_select and vector argument.
36506 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36507 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36508 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36509 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36510 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36511 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36513 2017-03-13 Richard Biener <rguenther@suse.de>
36516 * params.def (vect-max-peeling-for-alignment): Fix typo.
36518 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36520 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36521 issue that only occurred with binutils below 2.18.
36523 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36525 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36526 refer to binutils 2.11/2.12 minimum.
36528 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36530 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36531 ftp.kernel.org and simplify binutils requirement.
36533 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36535 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36537 (Optimize Options): Ditto. Also remove redundancy.
36539 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36541 PR translation/79848
36542 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36544 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36545 to G_ to avoid double translation.
36547 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36549 PR translation/79923
36550 * auto-profile.c (get_combined_location): Convert leading
36551 character of diagnostics to lower case and remove trailing period.
36552 (read_profile): Likewise for various diagnostics.
36553 * config/arm/arm.c (arm_option_override): Remove trailing period
36554 from various diagnostics.
36555 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36556 (msp430_expand_delay_cycles): Likewise.
36558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36561 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36562 full command-line argument, rather than just "str".
36563 (aarch64_validate_march): Likewise.
36564 (aarch64_validate_mtune): Likewise.
36566 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36568 PR rtl-optimization/78911
36569 * lra-assigns.c (must_not_spill_p): New function.
36570 (spill_for): Use it.
36572 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36574 PR tree-optimization/79981
36575 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36576 ATOMIC_COMPARE_EXCHANGE ifn result.
36577 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36578 IFN_ATOMIC_COMPARE_EXCHANGE.
36580 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36583 * opts.c (parse_sanitizer_options): Add missing question mark to
36584 "did you mean" message.
36586 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36588 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36590 (VMULEUH_UNS): Likewise.
36591 (VMULOUB_UNS): Likewise.
36592 (VMULOUH_UNS): Likewise.
36593 * config/rs6000/rs6000.c (builtin_function_type): Remove
36594 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36596 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36599 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36600 x with result of extra_parsing_for_operand_code_0.
36601 (function_reader::extra_parsing_for_operand_code_0): Convert
36602 return type from void to rtx, returning x. When reading
36603 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36604 larger size containing struct block_symbol.
36606 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36608 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36609 -mfloat128-hardware without -m64.
36611 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36614 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36615 entries to the case statement that marks unsigned arguments to
36616 overloaded functions.
36618 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36620 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36621 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36623 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36626 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36627 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36629 2017-03-10 Martin Liska <mliska@suse.cz>
36633 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36634 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36637 2017-03-10 Olivier Hainque <hainque@adacore.com>
36639 * tree-switch-conversion (array_value_type): Start by resetting
36640 candidate type to it's main variant.
36642 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36644 PR rtl-optimization/79909
36645 * combine.c (try_combine): Use simplify_replace_rtx on individual
36646 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36647 of the whole CALL_INSN_FUNCTION_USAGE.
36649 PR tree-optimization/79972
36650 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36651 get_range_info on SSA_NAMEs. Formatting fixes.
36653 2017-03-10 Richard Biener <rguenther@suse.de>
36654 Jakub Jelinek <jakub@redhat.com>
36656 PR tree-optimization/77975
36657 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36658 edge to be constant.
36659 (get_val_for): For constant x return it. Formatting fix.
36660 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36661 would use the same bases as the current one.
36663 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36665 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36666 instead of vec_select for V1TImode.
36667 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36669 (VSX_LE_128): Add V1TI to this mode iterator.
36670 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36671 (*vsx_le_perm_store_<mode>): Likewise.
36672 (pre-reload splitter for VSX stores): Likewise.
36673 (post-reload splitter for VSX stores): Likewise.
36674 (*vsx_xxpermdi2_le_<mode>): Likewise.
36675 (*vsx_lxvd2x2_le_<mode>): Likewise.
36676 (*vsx_stxvd2x2_le_<mode>): Likewise.
36678 2017-03-09 Michael Eager <eager@eagercon.com>
36680 Correct failures with --enable-checking=yes,rtl.
36682 * config/microblaze/microblaze.c (microblaze_expand_shift):
36683 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36684 test for const0_rtx.
36685 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36686 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36688 2017-03-09 Richard Biener <rguenther@suse.de>
36690 PR tree-optimization/79977
36691 * graphite-scop-detection.c (scop_detection::merge_sese):
36692 Handle the case of extra exits to blocks dominating the entry.
36694 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36696 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36699 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36701 PR rtl-optimization/79949
36702 * lra-constraints.c (process_alt_operands): Check memory when
36703 trying to predict a cycle. Print about the overall increase.
36705 2017-03-09 Richard Biener <rguenther@suse.de>
36707 PR middle-end/79971
36708 * gimple-expr.c (useless_type_conversion_p): Preserve
36709 TYPE_SATURATING for fixed-point types.
36711 2017-03-09 Richard Biener <rguenther@suse.de>
36714 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36715 alignment of BLKmode params.
36717 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36720 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36721 (VALL_NO_V2Q): Likewise.
36723 * config/aarch64/aarch64-simd.md
36724 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36726 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36727 VALL_NO_V2Q mode iterator.
36728 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36730 2017-03-09 Martin Liska <mliska@suse.cz>
36732 PR tree-optimization/79631
36733 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36734 tree_int_cst_sign_bit just for INTEGER constants.
36736 2017-03-09 Martin Liska <mliska@suse.cz>
36740 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36743 2017-03-09 Marek Polacek <polacek@redhat.com>
36746 * tree.c (inchash::add_expr): Handle TREE_VEC.
36748 2017-03-09 Martin Liska <mliska@suse.cz>
36751 (chkp_narrow_size_and_offset): New function.
36752 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36753 (void chkp_parse_bit_field_ref): New function.
36754 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36755 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36757 2017-03-09 Martin Liska <mliska@suse.cz>
36760 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36761 (chkp_find_bounds_1): Remove gcc_unreachable.
36763 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36766 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36767 BUILT_IN_SYNC*, determine the access type from the size suffix and
36768 always build a MEM_REF with that type. Handle forgotten
36769 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36772 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36773 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36774 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36775 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36776 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36777 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36778 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36779 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36780 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36781 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36782 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36783 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36784 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36785 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36786 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36787 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36788 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36789 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36790 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36791 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36792 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36793 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36794 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36795 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36796 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36797 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36798 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36799 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36800 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36801 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36802 definitions outside of __OPTIMIZE__ guarded section.
36805 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36806 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36807 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36808 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36811 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36813 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36814 ("vfenez<mode>"): Add missing constraints.
36816 2017-03-08 Martin Sebor <msebor@redhat.com>
36819 * config/nds32/nds32.c (nds32_option_override):
36820 Fix misspelled diagnostic.
36822 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36825 * gimplify.c (gimplify_omp_for): Replace index var in outer
36826 taskloop statement with an artificial variable and add
36827 OMP_CLAUSE_PRIVATE clause for it.
36829 2017-03-08 Richard Biener <rguenther@suse.de>
36831 PR tree-optimization/79955
36832 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36833 for accesses that are completely outside of the variable.
36835 2017-03-08 Andrew Haley <aph@redhat.com>
36837 PR tree-optimization/79943
36838 * tree-ssa-loop-split.c (compute_new_first_bound): When
36839 calculating the new upper bound, (END-BEG) should be added, not
36842 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36844 * config/avr/avr.md (setmemhi): Make sure match_dup
36845 operand number comes before match_scratch.
36847 2017-03-08 Richard Biener <rguenther@suse.de>
36849 PR tree-optimization/79920
36850 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36851 with ncopies == 1 to ...
36852 (vect_transform_slp_perm_load): ... here. Properly compute
36853 all element loads by iterating VF times over the group. Do
36854 not handle ncopies (computed in a broken way) in
36855 vect_create_mask_and_perm.
36857 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36860 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36861 is a uniform vector, use uniform_vector_p return value instead of
36862 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36864 2017-03-07 Marek Polacek <polacek@redhat.com>
36866 PR middle-end/79809
36867 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36868 (alloca_call_type): Likewise.
36870 2017-03-07 Martin Liska <mliska@suse.cz>
36872 * gcov.c (process_args): Put comment to correct location.
36874 2017-03-07 Martin Liska <mliska@suse.cz>
36876 PR middle-end/68270
36877 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36878 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36879 (chkp_narrow_bounds_for_field): Likewise.
36880 (chkp_parse_array_and_component_ref): Pass one more argument to
36883 2017-03-07 Richard Biener <rguenther@suse.de>
36885 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36888 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36890 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36891 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36893 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36896 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36897 to end of description.
36898 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36900 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36902 PR rtl-optimization/79901
36903 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36905 (*avx512f_<code><mode>3<mask_name>): ... this.
36906 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36907 iterator instead of VI8_AVX2_AVX512BW.
36909 PR rtl-optimization/79901
36910 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36911 min/max expander, expand it using expand_vec_cond_expr.
36914 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36917 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36920 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36921 to void * for PCH reasons.
36922 * dwarf2out.c (output_loc_operands, output_die): Cast
36923 v.val_vec.array to unsigned char *.
36925 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36928 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36931 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36933 PR rtl-optimization/79571
36934 * lra-constraints.c (process_alt_operands): Calculate static
36935 reject and subtract it from overall when only addresses will be
36938 2017-03-06 Julia Koval <julia.koval@intel.com>
36941 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36942 incoming stack boundary to 128 for 64-bit targets.
36944 2017-03-06 Richard Biener <rguenther@suse.de>
36946 PR tree-optimization/79894
36947 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36948 to NULL after folding it.
36950 2017-03-06 Richard Biener <rguenther@suse.de>
36952 PR tree-optimization/79824
36953 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36954 check disabling peeling for gaps.
36956 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36958 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36959 attributes): Document gettimeofday.
36961 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36963 * config/s390/s390.c (s390_option_override_internal): Set
36964 PARAM_MIN_VECT_LOOP_BOUND
36966 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36968 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36969 * config/s390/s390.md: Likewise.
36971 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36974 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36975 (<avx2_avx512>_perm<mode>): Rename to ...
36976 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36978 (<avx512>_perm<mode>_mask): Rename to ...
36979 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36981 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36982 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36983 instead of VI8F_256_512.
36984 (avx512f_perm<mode>): New define_expand.
36985 (avx512f_perm<mode>_mask): Likewise.
36986 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36987 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36989 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36991 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36992 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36994 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36996 2017-03-06 Martin Liska <mliska@suse.cz>
36999 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37000 when having a SSA NAME w/o VAR_DECL assigned to it.
37002 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37004 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37005 msa_dpsub_<su>_d): Fix MODE for vec_select.
37007 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37009 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37011 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37013 2017-03-06 Richard Biener <rguenther@suse.de>
37015 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37016 * plugin.c (register_plugin_info): Likewise.
37017 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37019 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37021 * config/i386/sse.md (sse_storehps, sse_storelps,
37022 avx_<castmode><avxsizesuffix>_<castmode>,
37023 avx512f_<castmode><avxsizesuffix>_<castmode>,
37024 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37025 in condition that at least one operand is not a MEM.
37027 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37029 PR middle-end/79805
37030 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37031 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37033 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37034 gimple_call_nothrow_p flag based on whether original builtin can throw.
37035 If it can, emit following stmts on the fallthrough edge.
37036 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37037 don't create new bb if inserting just debug stmts on the edge, try to
37038 insert them on the fallthru bb or just reset debug stmts.
37040 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37043 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37044 restore recog_data (including the operand rtxes inside it) around
37045 the call to get_insn_template.
37047 2017-03-03 Martin Sebor <msebor@redhat.com>
37049 PR tree-optimization/79699
37050 * context.c (context::~context): Free MPFR caches to avoid
37051 a memory leak on program exit.
37053 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37055 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37056 Use wide_int::ulow () instead of .elt (0).
37058 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37060 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37061 (*pushxf): Limit oF constraint to 32bit targets and add oC
37062 constraint for 64bit targets.
37063 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37064 (*pushdf): Change rmF constraint to rmC.
37066 2017-03-03 Martin Liska <mliska@suse.cz>
37068 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37069 Remove unused variable.
37071 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37074 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37075 is a memory operand, increase num_memory.
37076 (ix86_expand_args_builtin): Likewise.
37078 2017-03-03 Jan Hubicka <jh@suse.cz>
37081 * ipa-devirt.c (maybe_record_node): Properly handle
37082 __cxa_pure_virtual visibility.
37084 2017-03-03 Martin Liska <mliska@suse.cz>
37086 PR tree-optimization/79803
37087 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37089 (pass_loop_prefetch::execute): Disabled optimization if an
37090 assumption about L1 cache size is not met.
37092 2017-03-03 Martin Liska <mliska@suse.cz>
37094 PR rtl-optimization/79574
37095 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37096 (hash_scan_set): Likewise.
37097 (dump_hash_table): Likewise.
37098 (hoist_code): Likewise.
37100 2017-03-03 Richard Biener <rguenther@suse.de>
37102 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37103 in place of uhigh ().
37104 (fixed_convert_from_real): Likewise.
37106 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37109 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37111 2017-03-03 Richard Biener <rguenther@suse.de>
37113 PR middle-end/79818
37114 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37115 TYPE_OVERFLOW_UNDEFINED check.
37117 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37119 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37121 (vector_ae_<mode>_p): Likewise.
37122 (vector_nez_<mode>_p): Likewise.
37123 (vector_ne_v2di_p): Likewise.
37124 (vector_ae_v2di_p): Likewise.
37125 (vector_ne_<mode>_p): Likewise.
37126 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37128 (vsx_tsqrt<mode>2_fe): Likewise.
37130 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37133 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37135 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37137 PR rtl-optimization/79780
37138 * cprop.c (one_cprop_pass): When second and further conditional trap
37139 in a single basic block is turned into an unconditional trap, turn it
37140 into a deleted note to avoid RTL verification failures.
37142 2017-03-02 Richard Biener <rguenther@suse.de>
37144 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37146 2017-03-02 Richard Biener <rguenther@suse.de>
37148 PR tree-optimization/79345
37150 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37151 param and abort the walk, returning -1 if it is hit.
37152 (walk_aliased_vdefs): Take a limit param and pass it on.
37153 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37154 defaulting to 0 and return a signed int.
37155 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37156 (check_defs): New helper.
37157 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37158 about uninitialized memory.
37159 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37160 bogus uninitialized warning.
37161 (fixed_convert_from_real): Likewise.
37163 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37165 PR tree-optimization/66768
37166 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37167 iv_use if base object can't be determined.
37169 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37171 PR tree-optimization/79345
37172 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37173 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37174 (get_pattern_stats): Initialize it.
37175 * genemit.c (gen_expand): Verify match_scratch numbers come after
37176 match_operand/match_dup numbers.
37177 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37178 match_scratch numbers.
37179 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37181 * config/s390/s390.md (trunctdsd2): Likewise.
37183 2017-03-02 Richard Biener <rguenther@suse.de>
37185 * wide-int.h (wide_int_storage::operator=): Implement in terms
37188 2017-03-02 Richard Biener <rguenther@suse.de>
37190 PR tree-optimization/79777
37191 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37192 the to insert expression to sth existing.
37194 2017-03-01 Martin Sebor <msebor@redhat.com>
37196 PR middle-end/79692
37197 * gimple-ssa-sprintf.c
37198 (directive::known_width_and_precision): New function.
37199 (format_integer): Use it.
37200 (get_mpfr_format_length): Consider the full range of precision
37201 when computing %g output with the # flag. Set the likely byte
37202 count to 3 rather than 1 when precision is indeterminate.
37203 (format_floating): Correct the lower bound of precision.
37205 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37207 * doc/invoke.texi: Document default code model for 64-bit Linux.
37209 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37212 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37213 udiv rather than div since input pattern is unsigned.
37215 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37217 * config/i386/i386.c (print_reg): Warn for values of
37218 unsupported size in integer register.
37220 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37223 * config/rs6000/predicates.md (current_file_function_operand): Do
37224 not allow self calls to be local if the function is replaceable.
37226 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37229 * config/rs6000/altivec.h (vec_ctz and others): Change the
37230 preprocessor macro that controls conditional compilation from
37231 _ARCH_PWR9 to __POWER9_VECTOR__.
37232 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37233 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37234 control (instead of _ARCH_PWR9 control) so that template
37235 definition uses power9-specific function.
37236 (vec_any_eq): Likewise.
37237 (vec_all_ne): Change macro definition to use a power9-specific
37238 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37239 _ARCH_PWR9 control).
37240 (vec_any_eq) Likewise.
37241 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37242 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37243 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37244 support for xvcmpnedp instruction.
37245 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37246 macro expansion so that Power9 implementation of vec_all_ne does
37247 not use the AltiVec predicate framework.
37248 (VCMPNEH_P): Likewise.
37249 (VCMPNEW_P): Likewise.
37250 (VCMPNED_P): Likewise.
37251 (VCMPNEFP_P): Likewise.
37252 (VCMPNEDP_P): Likewise.
37253 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37254 implementation of vec_any_eq to not use AltiVec predicate
37256 (VCMPAEH_P): Likewise.
37257 (VCMPAEW_P): Likewise.
37258 (VCMPAED_P): Likewise.
37259 (VCMPAEFP_P): Likewise.
37260 (VCMPAEDP_P): Likewise.
37261 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37262 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37263 not use the AltiVec predicate framework.
37264 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37265 of vec_any_eq to not use AltiVec predicate framework.
37266 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37267 support for predefined __POWER9_VECTOR__ macro to indicate that
37268 Power9 instruction selection is enabled.
37269 (altivec_overloaded_builtins): Remove extraneous
37270 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37271 function argument types RS6000_BTI_bool_V16QI and
37272 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37273 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37274 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37275 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37276 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37277 Power9 for implementations of vec_cmpne. Change the signature for
37278 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37279 (representing vec_all_ne) to remove the previously described first
37280 argument of type RS6000_BTI_INTSI, as this was an artifact of
37281 reliance on the AltiVec predicate framework, which is no longer
37282 used in the implementation of these functions. Add
37283 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37284 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37285 since, unlike the AltiVec predicate framework implementation, we
37286 do not share function descriptors between vec_alle and vec_anyeq.
37287 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37288 set of modes that receive special treatment even when
37289 TARGET_P9_VECTOR is true. The special treatment emits code that
37290 does not depend on Power9 instructions.
37291 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37292 define_expand to not rely on AltiVec predicate framework.
37293 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37295 (vector_ne_v2di_p): Change this define_expand to not rely on
37296 AltiVec predicate framework.
37297 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37299 (vector_ne_<mode>_p): Change this define_expand to not rely on
37300 AltiVec predicate framework.
37301 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37303 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37304 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37305 define_insn pattern.
37306 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37307 define_insn pattern because the xvcmpne<VSs>. instruction is not
37309 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37310 instruction is not supported.
37312 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37314 * config/nvptx/nvptx.c: Include intl.h.
37316 2017-03-01 Martin Jambor <mjambor@suse.cz>
37319 * ipa-prop.h (ipa_bits): Removed field known.
37320 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37321 to pointers. Adjusted their comments to warn about their sharing.
37322 (ipcp_transformation_summary): Change bits to a vector of pointers.
37323 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37324 (ipa_get_ipa_bits_for_value): Declare.
37325 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37326 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37327 (ipa_bits_hash_table): Likewise.
37328 (ipa_vr_ggc_hash_traits): Likewise.
37329 (ipa_vr_hash_table): Likewise.
37330 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37331 being pointers and vr_known being removed.
37332 (ipa_set_jf_unknown): Likewise.
37333 (ipa_get_ipa_bits_for_value): New function.
37334 (ipa_set_jfunc_bits): Likewise.
37335 (ipa_get_value_range): New overloaded functions.
37336 (ipa_set_jfunc_vr): Likewise.
37337 (ipa_compute_jump_functions_for_edge): Use the above functions to
37338 construct bits and vr parts of jump functions.
37339 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37340 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37342 (ipcp_grow_transformations_if_necessary): Also allocate
37343 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37345 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37346 them. Fix too long lines.
37347 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37348 vr_known being removed.
37349 (ipa_read_jump_function): Use new setter functions to construct bits
37350 and vr parts of jump functions or set them to NULL.
37351 (write_ipcp_transformation_info): Adjust for bits being pointers.
37352 (read_ipcp_transformation_info): Likewise.
37353 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37355 Include gt-ipa-prop.h.
37356 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37358 (ipcp_store_bits_results): Likewise.
37359 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37360 Do not write to existing jump functions but use a temporary instead.
37362 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37365 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37366 attempt to use its first operand as BIT_FIELD_REF base.
37368 2017-03-01 Richard Biener <rguenther@suse.de>
37370 PR middle-end/79721
37371 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37372 interpolating formula in wrapping arithmetic.
37373 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37375 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37377 PR tree-optimization/79734
37378 * tree-vect-generic.c (expand_vector_condition): Optimize
37379 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37380 Handle VEC_COND_EXPR where comparison has different inner width from
37381 type's inner width.
37383 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37385 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37386 markup, and similar issues. Remove @opindex entries for things
37387 that aren't options. Add missing -mmpy-option entries.
37389 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37391 PR tree-optimization/79737
37392 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37393 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37394 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37395 instead of byte_size. Formatting fix.
37396 (shift_bytes_in_array_right): Formatting fix.
37398 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37401 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37402 condition on optimize for the leaf function test.
37404 2017-02-28 Martin Liska <mliska@suse.cz>
37407 * read-rtl-function.c (function_reader::handle_unknown_directive):
37408 Bail out when one uses -flto.
37410 2017-02-28 Martin Liska <mliska@suse.cz>
37412 * common.opt: Replace space with tabular for options of <number>
37414 * config/i386/i386.opt: Show <number> value for
37415 -mlarge-data-threshold.
37416 * opts.c (print_filtered_help): Do not display number in hexadecimal
37419 2017-02-28 Martin Liska <mliska@suse.cz>
37421 * common.opt: Fix --help=option -Q for options which are of
37424 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37426 * config/i386/i386.c (print_reg): Error out for values
37427 of 8-bit size in invalid integer register.
37429 2017-02-28 Martin Sebor <msebor@redhat.com>
37431 PR tree-optimization/79691
37432 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37434 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37437 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37438 gcc_unreachable with output_operand_lossage.
37440 2017-02-28 Richard Biener <rguenther@suse.de>
37442 PR tree-optimization/79740
37443 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37445 (visit_nary_op): Insert the nary into the hashtable if we
37446 pattern-matched sth.
37447 * tree-ssa-pre.c (eliminate_insert): Robustify.
37449 2017-02-28 Richard Biener <rguenther@suse.de>
37451 PR middle-end/79731
37452 * fold-const.c (decode_field_reference): Reject out-of-bound
37455 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37457 * config/i386/i386.c: Include intl.h.
37458 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37459 instead of just cond ? "..." : "...".
37460 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37461 * coverage.c (read_counts_file): Likewise.
37462 * omp-offload.c: Include intl.h.
37463 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37464 of just cond ? "..." : "...".
37465 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37466 of just cond ? "..." : "...".
37468 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37471 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37473 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37474 'tune for' CPU name.
37475 * config/arm/arm-cpu-data.h: Regenerated.
37477 2017-02-28 Richard Biener <rguenther@suse.de>
37479 PR tree-optimization/79732
37480 * tree-inline.c (expand_call_inline): Do not shadow var.
37482 2017-02-28 Richard Biener <rguenther@suse.de>
37484 PR tree-optimization/79723
37485 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37486 address-space properly.
37488 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37490 * doc/optinfo.texi (Optimization groups): Fix option used for
37492 * doc/invoke.texi (-fopt-info): Document "omp".
37493 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37494 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37495 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37496 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37497 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37499 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37501 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37502 OPTGROUP_OMP with "omp".
37504 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37507 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37508 for arithmetic shift of unsigned V2DI.
37510 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37512 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37513 arc/linux.h headers.
37514 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37515 (LINK_SPEC): Likewise.
37516 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37517 (EXTRA_SPECS): Likewise.
37518 (STARTFILE_SPEC): Likewise.
37519 (ENDFILE_SPEC): Likewise.
37520 (LIB_SPEC): Likewise.
37521 (TARGET_SDATA_DEFAULT): Likewise.
37522 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37523 (MULTILIB_DEFAULTS): Likewise.
37524 (DWARF2_UNWIND_INFO): Likewise.
37525 * config/arc/big.h: New file.
37526 * config/arc/elf.h: Likewise.
37527 * config/arc/linux.h: Likewise.
37528 * config/arc/t-uClibc: Remove.
37530 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37532 PR tree-optimization/77536
37533 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37534 (tree_transform_and_unroll_loop): Use above function to compute the
37535 estimated niter of unrolled loop and use it when scaling profile.
37536 Also use count info rather than frequency if it's non-zero.
37537 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37538 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37539 (vect_transform_loop): Call above function.
37541 2017-02-27 Richard Biener <rguenther@suse.de>
37543 PR tree-optimization/45397
37544 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37545 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37546 (visit_nary_op): Add pattern matching for CSEing sign-changed
37547 or truncated operations with wider ones.
37549 2017-02-27 Richard Biener <rguenther@suse.de>
37551 PR tree-optimization/79690
37552 * tree-vect-stmts.c (vectorizable_store): Use vector type
37553 built from the DR with address-space.
37555 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37557 * doc/invoke.texi (Optimize Options): Refine the description
37558 of asan-use-after-return.
37560 2017-02-25 Alan Modra <amodra@gmail.com>
37562 PR rtl-optimization/79584
37563 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37564 base, not ad->base_term, the reg within base. Remove assertion
37565 that ad->base == ad->base_term. Replace gen_int_mode using
37566 bogus mode with const0_rtx.
37568 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37570 PR middle-end/79396
37571 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37572 FMA_EXPR like tcc_binary or tcc_unary.
37574 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37577 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37579 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37580 (output_loc_operands): Handle DW_OP_call_ref and
37581 DW_OP_GNU_variable_value.
37582 (struct variable_value_struct): New type.
37583 (struct variable_value_hasher): Likewise.
37584 (variable_value_hash): New variable.
37585 (string_types): Remove.
37586 (copy_loc_descr): New function.
37587 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37588 (prepend_loc_descr_to_each): New function.
37589 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37590 instead of add_loc_descr_to_each if the first argument is single
37591 location list and the second has multiple.
37592 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37593 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37594 when looking for variable value which doesn't have other location info.
37595 (loc_list_from_tree): Formatting fix.
37596 (gen_array_type_die): Simplify DW_AT_string_length handling.
37597 (adjust_string_types): Remove.
37598 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37599 string_types. Call resolve_variable_values.
37600 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37601 (resolve_addr_in_expr): Likewise. Add A argument.
37602 (copy_deref_exprloc): Remove deref argument. Adjust for the
37603 original expression being DW_OP_GNU_variable_value with optionally
37604 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37605 optionally after it.
37606 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37607 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37608 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37609 (variable_value_hasher::hash, variable_value_hasher::equal): New
37611 (resolve_variable_value_in_expr, resolve_variable_value,
37612 resolve_variable_values, note_variable_value_in_expr,
37613 note_variable_value): New functions.
37614 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37617 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37620 * opts.h (handle_generated_option): Add GENERATED_P argument.
37621 * opts-common.c (handle_option): Adjust function comment.
37622 (handle_generated_option): Add GENERATED_P argument, pass it to
37624 (control_warning_option): Pass false to handle_generated_option
37626 * opts.c (maybe_default_option): Pass true to handle_generated_option
37628 * optc-gen.awk: Likewise.
37630 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37632 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37633 a REG, look at the REG it is a SUBREG of.
37634 (splitter for cmpeqsi_t): Ditto.
37636 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37638 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37639 the special USEs with the pattern of the insn, not the insn itself.
37641 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37644 * doc/invoke.texi: Document -mload-store-pairs.
37646 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37647 Sandra Loosemore <sandra@codesourcery.com>
37649 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37650 argument isn't a CONST_INT.
37651 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37652 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37653 (nios2_validate_compare): Bypass alternate compare logic if *op2
37654 is not a CONST_INT.
37655 (ldstwm_operation_p): Return false if first_base is not a REG or
37656 if first_offset is not a CONST_INT.
37658 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37660 * config/cris/cris.md: Use correct operand in a define_peephole2.
37662 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37664 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37666 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37668 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37669 this_insn if it is an INSN or JUMP_INSN.
37670 (force_offsettable): Look at base, not at addr.
37671 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37672 on things that aren't necessarily CONST_INTs.
37674 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37676 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37677 -mfpmath=sse is the default also for x86-32 targets with SSE2
37678 instruction set when @option{-ffast-math} is enabled
37680 2017-02-24 Jeff Law <law@redhat.com>
37682 PR rtl-optimizatoin/79286
37683 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37686 2017-02-24 Richard Biener <rguenther@suse.de>
37688 PR tree-optimization/79389
37689 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37692 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37694 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37695 function comment to reflect reality.
37696 (loop_exits_before_overflow): Fix typo in function description.
37698 2017-02-24 Richard Biener <rguenther@suse.de>
37700 PR tree-optimization/79389
37701 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37702 properly that a threading opportunity exists. Detect conditional
37703 copy/constant propagation opportunities.
37705 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37707 * config/visium/visium.md (type): Add trap.
37708 (b): New mode attribute.
37709 (*btst): Rename into...
37710 (*btst<mode>): ...this and adjust.
37711 (*cbranchsi4_btst_insn): Rename into...
37712 (*cbranch<mode>4_btst_insn): ...this and adjust.
37713 (trap): New define_insn.
37715 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37717 PR tree-optimization/79389
37718 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37719 (noce_reversed_cond_code): New function.
37720 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37721 reversed_comparison_code. Formatting fix.
37722 (noce_try_store_flag): Test rev_cond != NULL in addition to
37723 reversed_comparison_code.
37724 (noce_try_store_flag_constants): Likewise.
37725 (noce_try_store_flag_mask): Likewise.
37726 (noce_try_addcc): Use rev_cond if non-NULL instead of
37727 reversed_comparison_code.
37728 (noce_try_cmove_arith): Likewise. Formatting fixes.
37729 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37730 (noce_find_if_block): Initialize rev_cond.
37731 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37732 instead of false as last argument never attempt to reverse it
37735 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37737 PR tree-optimization/79663
37738 * tree-predcom.c (combine_chains): Process refs in reverse order
37739 only for ZERO length chains, and add explaining comment.
37741 2017-02-23 Jeff Law <law@redhat.com>
37743 PR tree-optimization/79578
37744 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37745 in call to operand_equal_p.
37747 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37750 * config/i386/cpuid.h: Fix another undefined behavior.
37752 2017-02-23 Richard Biener <rguenther@suse.de>
37754 PR tree-optimization/79683
37755 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37756 vector types for data-refs.
37758 2017-02-23 Martin Liska <mliska@suse.cz>
37760 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37762 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37764 PR middle-end/79665
37765 * internal-fn.c (get_range_pos_neg): Moved to ...
37766 * tree.c (get_range_pos_neg): ... here. No longer static.
37767 * tree.h (get_range_pos_neg): New prototype.
37768 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37769 are known to be in between 0 and signed maximum inclusive, try to
37770 expand both unsigned and signed divmod and use the cheaper one from
37773 2017-02-22 Jeff Law <law@redhat.com>
37775 PR tree-optimization/79578
37776 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37777 to compare base operands.
37779 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37782 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37783 gpc_reg_operand instead of fpr_reg_operand.
37785 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37787 * config/mips/mips.c (mips_return_in_memory): Force FP
37788 vector types to be returned in memory for o32 ABI.
37790 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37792 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37793 instead of DW_TAG_member for static data member declarations and don't
37794 set no_linkage_name for static inline data members.
37795 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37798 2017-02-22 Martin Liska <mliska@suse.cz>
37800 * doc/invoke.texi: Replace inequality signs with square brackets
37803 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37806 * lra-constraints.c (simplify_operand_subreg): Handle
37807 WORD_REGISTER_OPERATIONS targets.
37809 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37812 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37813 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37814 elimination by swapping fld*.
37816 2017-02-22 Richard Biener <rguenther@suse.de>
37818 PR tree-optimization/79673
37819 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37820 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37821 irrelevant address-space qualifiers and avoiding a
37822 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37824 2017-02-22 Richard Biener <rguenther@suse.de>
37826 PR tree-optimization/79666
37827 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37828 to not symbolically negate if that may introduce undefined
37831 2017-02-22 Martin Liska <mliska@suse.cz>
37834 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37835 * data-streamer-out.c (streamer_write_gcov_count_stream):
37837 * value-prof.c (stream_out_histogram_value): Make assert more
37838 precise based on type of counter.
37840 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37843 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37844 Use nonimmediate_operand instead of memory_operand for operand 1.
37845 (float-extend standard_x87sse_constant_load splitter): Ditto.
37847 2017-02-21 Jeff Law <law@redhat.com>
37849 PR tree-optimization/79621
37850 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37851 blocks with edges to themselves.
37853 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37856 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37857 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37858 Use gimple_call_builtin_p.
37861 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37862 on temporarily removed DEBUG_INSNs.
37864 PR tree-optimization/79649
37865 * tree-loop-distribution.c (classify_partition): Give up on
37866 non-generic address space loads/stores.
37868 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37870 * doc/loop.texi (Loop manipulation): Remove nonexistent
37871 tree_ssa_loop_version from the documentation.
37872 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37874 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37877 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37878 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37879 * config/rs6000/rs6000.c: Include except.h.
37880 (rs6000_expand_split_stack_prologue): Call
37881 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37883 2017-02-21 Martin Jambor <mjambor@suse.cz>
37886 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37887 have been analyzed.
37889 2017-02-21 Martin Jambor <mjambor@suse.cz>
37891 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37892 for backward compatibility only.
37893 * doc/invoke.texi (Option Summary): Remove all references to
37894 -fipa-cp-alignment.
37896 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37900 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37902 * lra-constraints.c (curr_insn_transform): Handle
37903 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37905 2017-02-21 Martin Liska <mliska@suse.cz>
37907 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37909 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37911 PR translation/79638
37912 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37914 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37917 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37918 (arm_function_ok_for_sibcall): Return false for an indirect call by
37919 descriptor if all the argument registers are used.
37920 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37921 alignment of the function.
37923 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37925 PR tree-optimization/61441
37926 * simplify-rtx.c (simplify_const_unary_operation): For
37927 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37928 the sNaN unmodified.
37930 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37932 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37933 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37934 instead of SYSTEM_HEADER_DIR.
37936 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37937 Martin Liška <mliska@suse.cz>
37939 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37940 Fix typos and grammar, use active voice, and clarify.
37942 2017-02-20 Marek Polacek <polacek@redhat.com>
37944 PR middle-end/79537
37945 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37948 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37950 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37953 * config/i386/i386.c (ix86_expand_builtin): Handle
37954 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37955 ix86_builtins_isa[fcode].isa as a requirement of those
37956 flags and any other flag in the bitmask.
37957 (ix86_init_mmx_sse_builtins): Use 0 instead of
37958 ~OPTION_MASK_ISA_64BIT as mask.
37959 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37960 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37961 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37962 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37964 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37967 * lra-constraints.c (split_reg): Check requested split mode
37968 is supported by the register.
37970 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37972 * lra-constraints.c (simplify_operand_subreg): Remove early
37975 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37978 * lra-constraints.c (curr_insn_transform): Tighten condition
37979 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37981 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 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37990 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37992 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37994 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37997 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38000 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38002 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38005 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38007 * final.c (last_columnnum, override_columnnum): New variables.
38008 (final_start_function): Set last_columnnum, pass it to begin_prologue
38009 hook and pass 0 to dwarf2out_begin_prologue.
38010 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38011 to source_line debug hook.
38012 (notice_source_line): Compute last_columnnum and for debug_column_info
38013 return true on column changes.
38014 * debug.h (struct gcc_debug_hooks): Add column argument to
38015 source_line and begin_prologue hooks.
38016 (debug_nothing_int_charstar_int_bool): Remove prototype.
38017 (debug_nothing_int_int_charstar,
38018 debug_nothing_int_int_charstar_int_bool): New prototypes.
38019 (dwarf2out_begin_prologue): Add column argument.
38020 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38021 begin_prologue hooks.
38022 (debug_nothing_int_charstar_int_bool): Remove.
38023 (debug_nothing_int_int_charstar,
38024 debug_nothing_int_int_charstar_int_bool): New functions.
38025 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38026 through to dwarf2out_source_line.
38027 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38028 (dwarf2out_source_line): Add column argument, emit it if requested.
38029 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38031 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38032 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38033 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38034 through to dwarf2out_begin_prologue.
38035 (vmsdbgout_source_line): Add column argument, pass it through to
38036 dwarf2out_source_line.
38037 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38038 dbxout_source_line caller.
38039 (dbxout_source_line): Add column argument.
38041 * common.opt (gno-column-info, gcolumn-info): New options.
38042 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38043 (check_die): Also test for multiple DW_AT_decl_column attributes.
38044 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38045 DW_AT_decl_column if requested.
38046 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38048 (gen_variable_die): Likewise.
38049 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38050 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38053 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38054 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38055 (ix86_handle_option): Handle OPT_m3dnowa.
38056 * doc/invoke.texi (-m3dnowa): Document.
38057 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38058 -m3dnowa instead of -m3dnow -march=athlon.
38061 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38062 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38064 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38067 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38068 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38069 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38070 generator for vsx_xxpermdi_<mode>_be.
38071 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38072 force big-endian semantics.
38073 (vsx_xxpermdi_<mode>_be): New define_expand with same
38074 implementation as previous version of vsx_xxpermdi_<mode>.
38076 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38078 PR tree-optimization/79327
38079 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38080 variable, its initialization and use.
38082 2017-02-17 Julia Koval <julia.koval@intel.com>
38084 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38085 (OPTION_MASK_ISA_PKU_UNSET): New.
38086 (ix86_handle_option): Handle -mrdpid.
38087 * config/i386/cpuid.h (bit_RDPID): New.
38088 * config/i386/driver-i386.c (host_detect_local_cpu):
38089 Detect RDPID feature.
38090 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38091 * config/i386/i386-c.c (ix86_target_macros_internal):
38093 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38094 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38095 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38096 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38097 * config/i386/i386.md (define_insn "rdpid"): New.
38098 * config/i386/i386.opt Add -mrdpid.
38099 * config/i386/immintrin.h (_rdpid_u32): New.
38101 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38103 PR rtl-optimization/79541
38104 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38105 instead of transforming it into USE.
38107 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38109 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38110 If HONOR_SNANS (SFmode) force the input to a register.
38111 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38112 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38113 an frsp or similar insn.
38115 2017-02-17 Martin Liska <mliska@suse.cz>
38117 PR rtl-optimization/79577
38118 * params.def (selsched-max-sched-times): Increase minimum to 1.
38120 2017-02-17 Martin Liska <mliska@suse.cz>
38122 PR rtl-optimization/79574
38123 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38125 2017-02-17 Martin Liska <mliska@suse.cz>
38127 PR tree-optimization/79529
38128 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38129 ssa_defined_default_def_p to handle cases which are implicitly
38131 * tree-ssa.c (ssa_defined_default_def_p): New function.
38132 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38133 which are implicitly defined.
38134 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38136 2017-02-17 Richard Biener <rguenther@suse.de>
38138 PR middle-end/79576
38139 * params.def (max-ssa-name-query-depth): Limit to 10.
38141 2017-02-17 Richard Biener <rguenther@suse.de>
38143 PR tree-optimization/79552
38144 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38147 2017-02-17 Richard Biener <rguenther@suse.de>
38150 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38152 2017-02-17 Marek Polacek <polacek@redhat.com>
38154 PR middle-end/79536
38155 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38156 (fold_negate_expr): New wrapper.
38158 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38160 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38161 Correct terminology and de-emphasize pre-standard behavior.
38163 2017-02-16 Alan Modra <amodra@gmail.com>
38165 PR rtl-optimization/79286
38166 * ira.c (def_dominates_uses): New function.
38167 (update_equiv_regs): Don't create an equivalence for insns that
38168 may trap where the register def does not dominate the use.
38170 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38172 PR rtl-optimization/78127
38173 * lra.c (lra): Call lra_eliminate before finish the loop after
38176 2017-02-16 Richard Biener <rguenther@suse.de>
38178 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38180 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38181 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38182 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38183 (isl_val_int_from_wi): New function.
38184 (extract_affine_gmp): Rename to ...
38185 (extract_affine_wi): ... this, take a widest_int.
38186 (extract_affine_int): Just wrap extract_affine_wi.
38187 (add_param_constraints): Use isl_val_int_from_wi.
38188 (add_loop_constraints): Likewise, and extract_affine_wi.
38190 2017-02-15 Jeff Law <law@redhat.com>
38192 PR middle-end/79521
38193 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38194 ira_init_register_move_cost_if_necessary.
38196 2017-02-15 Martin Sebor <msebor@redhat.com>
38198 PR middle-end/32003
38199 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38200 removed in a prior commit.
38202 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38204 PR tree-optimization/79347
38205 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38206 counters during peeling.
38208 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38210 * Makefile.in (site.exp): Remove "set ISLVER".
38212 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38215 * real.c (real_from_integer): Call real_convert even for decimal.
38217 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38220 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38222 2017-02-14 Andrew Pinski <apinski@cavium.com>
38224 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38225 cores and change the partno/implementer to be correct.
38226 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38227 the 'B" as the implementer.
38228 * config/aarch64/aarch64-tune.md: Regenerate.
38230 2017-02-14 Carl Love <cel@us.ibm.com>
38232 * config/rs6000/rs6000.c: Add case statement entry to make the
38233 xvcvuxdsp built-in argument unsigned.
38234 * config/rs6000/vsx.md: Fix the source and return operand types so they
38235 match the instruction definitions from the ISA document. Fix typo
38236 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38239 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38242 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38243 member early_clobber_alts.
38244 * lra-lives.c (reg_early_clobber_p): New.
38245 (process_bb_lives): Use it.
38246 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38247 (debug_operand_data): Initialize early_clobber_alts.
38248 (setup_operand_alternative): Set up early_clobber_alts.
38249 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38250 alternatives to new_insn_reg.
38251 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38253 (lra_update_insn_regno_info): Pass the new arg.
38255 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38257 PR middle-end/79505
38258 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38259 (new_oacc_loop_raw): Don't clear already cleared fields.
38262 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38263 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38264 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38266 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38269 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38271 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38274 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38275 the extra instruction to the right place to store 128-bit constant
38278 2017-02-14 Martin Sebor <msebor@redhat.com>
38280 PR middle-end/79448
38281 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38282 warning for strings of unknown length.
38284 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38286 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38288 2017-02-14 Jeff Law <law@redhat.com>
38291 * ira-costs.c (scan_one_insn): Initialize register move costs
38292 for pseudos seen in USE/CLOBBER insns.
38294 PR tree-optimization/79095
38295 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38296 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38297 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38298 if the operands are known to be not equal, then the resulting range
38300 (intersect_ranges): If the new range is ~[0,0] and the old range is
38301 wide, then prefer ~[0,0].
38302 * tree-vrp.c (overflow_comparison_p_1): New function.
38303 (overflow_comparison_p): New function.
38304 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38305 if NAME is used in an overflow test.
38306 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38307 overflow check that can be expressed as an equality test, then adjust
38308 ops to be that equality test.
38310 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38312 * config/s390/s390-builtin-types.def: Remove flags argument.
38313 * config/s390/s390.c (s390_init_builtins): Likewise.
38315 2017-02-14 Martin Liska <mliska@suse.cz>
38317 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38318 vector. Fix trailing white spaces.
38320 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38322 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38325 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38327 PR rtl-optimization/68664
38328 * config/arm/arm.c (arm_sched_can_speculate_insn):
38329 New function. Declare prototype.
38330 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38332 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38334 PR rtl-optimization/68664
38335 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38337 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38339 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38341 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38342 max skip bytes for function, loop and jump.
38344 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38346 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38347 ABS_EXPR for gimple dump.
38349 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38352 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38354 PR tree-optimization/79408
38355 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38356 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38357 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38358 also if rhs1 is INTEGER_CST.
38360 2017-02-14 Richard Biener <rguenther@suse.de>
38362 PR middle-end/79432
38363 * tree-into-ssa.c (insert_phi_nodes): When the function can
38364 have abnormal edges rewrite SSA names with broken use-def
38365 dominance out of SSA and register them for PHI insertion.
38367 2017-02-13 Martin Sebor <msebor@redhat.com>
38369 PR middle-end/79496
38370 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38371 clearing info.nowrite flag when snprintf size argument is a range.
38373 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38375 * cprop.c (cprop_jump): Add missing space in string literal.
38376 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38377 (get_constraint_for_component_ref): Likewise.
38378 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38379 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38380 * lra-constraints.c (process_alt_operands): Likewise.
38381 * ipa-inline.c (inline_small_functions): Likewise.
38382 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38383 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38384 * trans-mem.c (diagnose_tm_1_op): Likewise.
38385 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38386 (grid_parallel_clauses_gridifiable): Likewise.
38388 * config/nvptx/mkoffload.c (process): Add space in between
38391 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38392 "MOD4_SSE_REGS" and "ALL_REGS".
38394 * spellcheck.c (test_data): Add , in between "foo" and "food".
38396 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38399 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38400 boundary crossing check and subsequent code generation agree.
38402 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38404 * config/aarch64/aarch64.c (has_memory_op): Delete.
38405 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38408 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38410 PR rtl-optimization/79388
38411 PR rtl-optimization/79450
38412 * combine.c (distribute_notes): When removing TEM_INSN for which
38413 corresponding dest has last value recorded, invalidate that last
38416 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38418 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38419 of explicit '@'. Add missing assembly comment marker on branch costs
38422 2017-02-13 Nathan Sidwell <nathan@acm.org>
38424 * gengtype-lex.l (<in_struct>): Add '/'.
38426 2017-02-13 Martin Liska <mliska@suse.cz>
38429 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38431 2017-02-13 Richard Biener <rguenther@suse.de>
38433 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38435 * configure: Re-generate.
38436 * config.in: Likewise.
38437 * graphite-dependences.c: Simplify as if
38438 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38439 * graphite-isl-ast-to-gimple.c: Likewise.
38440 * graphite-optimize-isl.c: Likewise.
38441 * graphite-poly.c: Likewise.
38442 * graphite-sese-to-poly.c: Likewise.
38443 * graphite.h: Likewise.
38444 * toplev.c: Include isl/version.h and use isl_version () for
38445 printing the ISL version.
38446 * doc/install.texi: Update ISL requirement.
38448 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38450 * doc/standards.texi (Standards): Update reference to
38453 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38455 * doc/extend.texi (Named Address Spaces): sourceware.org now
38457 * doc/install.texi (Binaries): Ditto.
38460 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38462 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38463 standard terminology "stringize"/"stringizing" throughout.
38464 * doc/cppinternals.texi: Likewise.
38466 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38468 * doc/extend.texi: Fix some spelling mistakes and typos.
38469 * doc/invoke.texi: Likewise.
38471 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38474 * params.def (inline-min-speedup) Change from 10 to 8.
38476 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38478 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38481 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38484 * ipa-inline-analysis.c (get_minimal_bb): New function.
38485 (record_modified): Use it.
38486 (remap_edge_change_prob): Handle also ancestor functions.
38488 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38490 * doc/contrib.texi (Contributors): Remove broken link into
38491 the Mauve CVS repository.
38493 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38495 PR middle-end/79454
38496 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38497 result computation whenever lhs doesn't have vector mode, not
38498 just when it has BLKmode.
38500 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38502 * doc/makefile.texi (profiledbootstrap): Refer to the
38503 installation instructions only in textual form.
38505 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38508 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38510 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38512 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38513 (Specific): Update mingw-w64 reference.
38515 (Specific): Remove broken link to Renesas RX processor.
38517 2017-02-10 Richard Biener <rguenther@suse.de>
38519 * toplev.c (process_options): Do not mention obsolete graphite
38520 options when printing sorry message about missing graphite support.
38521 Mention -floop-nest-optimize.
38523 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38525 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38526 (vtst_p16): Likewise.
38527 (vtstq_p8): Likewise.
38528 (vtstq_p16): Likewise.
38530 (vtstq_p64): Likewise.
38531 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38532 (vset_lane_p64): New.
38533 (vsetq_lane_p64): New.
38535 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38537 PR tree-optimization/79411
38538 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38539 stmt operands are SSA_NAMEs used in abnormal phis.
38540 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38543 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38546 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38549 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38551 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38553 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38555 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38559 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38560 not arbitrary TREE_CONSTANT.
38563 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38564 "omp declare target link" attribute unless is_global_var.
38565 * omp-offload.c (find_link_var_op): Likewise.
38567 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38568 Chung-Lin Tang <cltang@codesourcery.com>
38570 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38572 (gimplify_adjust_omp_clauses): Don't delete TILE.
38573 (gimplify_omp_for): Deal with TILE.
38574 * internal-fn.c (expand_GOACC_TILE): New function.
38575 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38577 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38578 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38580 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38581 avoid DIV for outermost collapse var.
38582 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38583 Remove out of date comments, fix whitespace.
38584 * omp-general.c (omp_extract_for_data): Deal with tiling.
38585 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38586 adjust OLF_DIM_BASE value.
38587 (struct omp_for_data): Add tiling field.
38588 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38589 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38590 for auto loops. Remove default auto determining, moved to
38591 oacc_loop_fixed_partitions.
38592 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38593 stmts, add e_mask field.
38594 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38595 (oacc_thread_numbers): Use oacc_dim_call.
38596 (oacc_xform_tile): New.
38597 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38598 (finish_oacc_loop): Adjust for ifns vector.
38599 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38600 add case for GOACC_TILE fns.
38601 (oacc_loop_xform_loop): Delete.
38602 (oacc_loop_process): Iterate over call list directly, and add
38603 handling for GOACC_TILE fns.
38604 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38606 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38607 vector partitioning to outer loops. Assign 2 partitions to loops
38608 when available. Add TILE handling.
38609 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38610 (execite_oacc_device_lower): Process GOACC_TILE fns,
38611 ignore unknown specs.
38612 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38613 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38614 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38616 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38618 * configure.ac (ACX_BUGURL): Update.
38619 * configure: Regenerate.
38621 2017-02-09 Richard Biener <rguenther@suse.de>
38623 PR tree-optimization/69823
38624 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38625 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38627 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38629 * config/arc/arc-c.def: Add __NPS400__ definition.
38630 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38631 (TARGET_NPS400): Define.
38633 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38635 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38637 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38638 pointer, arch_info.
38639 (arc_cpu_types): Fill the arch_info field with a pointer into the
38640 arc_arch_types table.
38641 (arc_selected_cpu): Declare.
38642 * config/arc/arc.c (arc_selected_cpu): Make global.
38643 (arc_selected_arch): Delete.
38644 (arc_base_cpu): Delete.
38645 (arc_override_options): Remove references to deleted variables,
38646 update access to arch information.
38647 (ARC_OPT): Update access to arch information.
38648 (ARC_OPTX): Likewise.
38649 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38650 (TARGET_ARC600): Update access to arch information.
38651 (TARGET_ARC601): Likewise.
38652 (TARGET_ARC700): Likewise.
38653 (TARGET_EM): Likewise.
38654 (TARGET_HS): Likewise.
38655 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38658 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38661 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38662 condition/operands for integer GE/LE/GEU/LEU operations.
38664 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38666 PR translation/79397
38667 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38670 2017-02-08 Martin Jambor <mjambor@suse.cz>
38673 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38674 whether allocation happened.
38675 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38676 nothing was allocated.
38678 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38680 PR tree-optimization/79408
38681 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38682 constant, but SSA_NAME with a known integer range, use the minimum
38683 of that range instead of op1 to determine if modulo can be replaced
38684 with its first operand.
38686 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38688 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38690 2017-02-08 Richard Biener <rguenther@suse.de>
38692 PR tree-optimization/71824
38693 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38694 Check all loops contained in the merged region.
38696 2017-02-07 Andrew Pinski <apinski@cavium.com>
38698 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38700 2017-02-07 Andrew Pinski <apinski@cavium.com>
38702 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38703 (thunderxt88): Likewise.
38704 (thunderxt81): Disable LSE and change v8.1 to v8.
38705 (thunderxt83): Likewise.
38707 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38708 Richard Biener <rguenther@suse.de>
38710 PR middle-end/79399
38711 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38712 type from int to size_t.
38713 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38715 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38717 PR rtl-optimization/79386
38718 * cprop.c (bypass_conditional_jumps): Initialize
38719 bypass_last_basic_block already before splitting bbs after
38720 unconditional traps...
38721 (bypass_conditional_jumps): ... rather than here.
38724 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38725 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38726 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38727 fix -masm=intel patterns.
38729 2017-02-07 Richard Biener <rguenther@suse.de>
38731 PR tree-optimization/79256
38732 PR middle-end/79278
38733 * builtins.c (get_object_alignment_2): Use min_align_of_type
38734 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38735 and ADJUST_FIELD_ALIGN.
38737 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38739 * doc/tm.texi: Regenerate.
38740 * stor-layout.c (layout_decl): Adjust.
38741 (update_alignment_for_field): Likewise.
38742 (place_field): Likewise.
38743 (min_align_of_type): Likewise.
38744 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38745 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38746 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38747 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38748 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38749 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38750 * config/i386/i386.c (x86_field_alignment): Likewise.
38751 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38752 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38753 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38754 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38755 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38756 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38760 2017-01-30 Richard Biener <rguenther@suse.de>
38762 PR tree-optimization/79256
38763 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38764 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38767 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38769 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38770 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38771 builtins to SImode and emit a zero-extend, if necessary.
38773 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38775 * docs/invoke.texi (RISC-V Options): Alphabetize.
38777 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38779 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38782 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38784 * config/riscv/riscv.c: New file.
38785 * common/config/riscv/riscv-common.c: Likewise.
38786 * config.gcc: Likewise.
38787 * config/riscv/constraints.md: Likewise.
38788 * config/riscv/elf.h: Likewise.
38789 * config/riscv/generic.md: Likewise.
38790 * config/riscv/linux.h: Likewise.
38791 * config/riscv/multilib-generator: Likewise.
38792 * config/riscv/peephole.md: Likewise.
38793 * config/riscv/pic.md: Likewise.
38794 * config/riscv/predicates.md: Likewise.
38795 * config/riscv/riscv-builtins.c: Likewise.
38796 * config/riscv/riscv-c.c: Likewise.
38797 * config/riscv/riscv-ftypes.def: Likewise.
38798 * config/riscv/riscv-modes.def: Likewise.
38799 * config/riscv/riscv-opts.h: Likewise.
38800 * config/riscv/riscv-protos.h: Likewise.
38801 * config/riscv/riscv.h: Likewise.
38802 * config/riscv/riscv.md: Likewise.
38803 * config/riscv/riscv.opt: Likewise.
38804 * config/riscv/sync.md: Likewise.
38805 * config/riscv/t-elf-multilib: Likewise.
38806 * config/riscv/t-linux: Likewise.
38807 * config/riscv/t-linux-multilib: Likewise.
38808 * config/riscv/t-riscv: Likewise.
38809 * configure.ac: Likewise.
38810 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38811 Waterman as RISC-V maintainers.
38812 * doc/install.texi: Add RISC-V entries.
38813 * doc/invoke.texi: Add RISC-V options section.
38814 * doc/md.texi: Add RISC-V constraints section.
38815 * configure: Regenerated.
38817 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38820 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38821 false values to be constant vectors with all 0 or all 1 bits set.
38822 (vcondu<mode><mode>): Likewise.
38823 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38825 (fpmask_comparison_operator): Update comment.
38826 (vecint_comparison_operator): New predicate.
38827 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38828 vector conditionals when the true and false values are constant
38829 vectors with all 0 bits or all 1 bits set.
38831 2017-02-06 Martin Sebor <msebor@redhat.com>
38833 PR tree-optimization/79376
38834 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38836 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38838 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38839 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38840 to simplify split condition.
38842 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38844 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38845 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38848 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38850 PR rtl-optimization/68664
38851 * target.def (can_speculate_insn): New hook.
38852 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38853 * doc/tm.texi: Regenerate.
38854 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38855 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38856 (rs6000_sched_can_speculate_insn): New function.
38858 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38860 PR tree-optimization/79284
38861 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38862 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38863 vectorizable_mask_load_store, vectorizable_operation,
38864 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38865 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38866 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38867 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38868 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38869 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38870 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38871 is_gimple_assign (stmt). Replace another such test with
38872 is_gimple_assign (stmt).
38874 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38877 * config/avr/avr.c (rtl-iter.h): Include it.
38878 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38879 (avr_legitimate_combined_insn): ...and implementation.
38881 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38883 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38884 * config/s390/s390.c (s390_const_operand_ok)
38885 (s390_canonicalize_comparison, s390_extract_part)
38886 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38887 (s390_contiguous_bitmask_p, s390_rtx_costs)
38888 (legitimize_pic_address): Likewise.
38889 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38890 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38891 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38892 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38893 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38895 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38897 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38898 REGNO($0) == REGNO($1).
38900 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38902 * config/s390/linux.h(SIZE_TYPE): Add comment.
38904 2017-02-06 Julian Brown <julian@codesourcery.com>
38905 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38906 Virendra Pathak <virendra.pathak@broadcom.com>
38908 * config/aarch64/aarch64-cores.def: Change the scheduler
38910 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38911 * config/aarch64/thunderx2t99.md: New file.
38913 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38915 * doc/standards.texi (Go Language): Update link to language
38918 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38920 * tree-eh.c (lower_resx): Sanitize profile.
38921 (cleanup_empty_eh_move_lp): Likewise.
38923 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38926 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38928 * cfgloopmanip.h (loop_version): Update prototype.
38929 * modulo-sched.c (sms_schedule): Update call of loop_version.
38930 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38931 * tree-parloops.c (gen_parallel_loop): Likewise.
38932 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38933 * tree-ssa-loop-split.c (split_loop): Likewise.
38934 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38935 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38937 2017-02-05 Martin Liska <mliska@suse.cz>
38940 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38942 (print_operand_address): Initialize a struct to zero.
38944 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38946 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38947 garbage collector only in textual form.
38949 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38951 * doc/extend.texi (x86 specific memory model extensions for
38952 transactional memory): Simplify a phrase.
38954 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38957 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38958 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38959 (atomic_storedi_1): Likewise.
38961 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38963 PR tree-optimization/79338
38964 * tree-parloops.c (gather_scalar_reductions): Don't call
38965 vect_analyze_loop_form for loop->inner before destroying loop's
38968 2017-02-03 Martin Sebor <msebor@redhat.com>
38970 PR tree-optimization/79327
38971 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38972 when precision has resulted in leading zeros.
38973 (format_integer): Adjust the likely counter to assume an unknown
38974 argument that may be zero is non-zero.
38976 2017-02-03 Jason Merrill <jason@redhat.com>
38979 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38980 avoid copying non-taken branch.
38982 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38984 PR tree-optimization/79340
38985 * tree-vect-loop.c (vectorizable_reduction): Release
38986 vec_defs elements after safe_splicing them into other vectors.
38989 PR tree-optimization/79327
38990 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38991 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38993 (format_integer): Use wide_int_to_tree instead of build_int_cst
38994 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38995 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38996 of shortest and longest sequence.
38998 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39000 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39001 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39003 2017-02-03 Walter Lee <walt@tilera.com>
39006 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39007 after initial stackframe link reg save.
39008 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39010 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39013 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39014 wu for stxssp alternative.
39016 2017-02-03 Martin Sebor <msebor@redhat.com>
39018 PR tree-optimization/79352
39019 * gimple-fold.c (get_range_strlen): Add argument.
39020 (get_range_strlen): Change return type to bool.
39021 (get_maxval_strlen): Pass in a dummy argument.
39022 * gimple-fold.h (get_range_strlen): Change return type to bool.
39023 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39024 * tree.h (array_at_struct_end_p): Add argument.
39025 * tree.c (array_at_struct_end_p): Handle it.
39027 2017-02-03 Martin Liska <mliska@suse.cz>
39030 * multiple_target.c (create_dispatcher_calls): Redirect edge
39031 from a caller of a dispatcher.
39032 (expand_target_clones): Make the clones local.
39033 (ipa_target_clone): Do both target clones and resolvers.
39034 (ipa_dispatcher_calls): Remove the pass.
39035 (pass_dispatcher_calls::gate): Likewise.
39036 (make_pass_dispatcher_calls): Likewise.
39037 * passes.def (pass_target_clone): Put as very first IPA early
39040 2017-02-03 Martin Liska <mliska@suse.cz>
39042 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39043 in case of a function with ifunc attribute.
39045 2017-02-03 Martin Liska <mliska@suse.cz>
39047 * cgraph.c (cgraph_node::dump): Dump function version info.
39048 * symtab.c (symtab_node::dump_base): Add missing new line.
39050 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39052 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39053 (ifcombine_ifandif): Use it.
39055 2017-02-03 Martin Liska <mliska@suse.cz>
39057 * doc/invoke.texi: Document default value for
39058 use-after-scope-direct-emission-threshold.
39060 2017-02-03 Martin Liska <mliska@suse.cz>
39062 PR tree-optimization/79339
39063 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39064 (format_floating): Likewise.
39066 2017-02-03 Martin Liska <mliska@suse.cz>
39069 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39071 (ipa_node_params_t::remove): Likewise.
39072 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39073 initialization from removed ipa_node_params_t::insert.
39074 (ipa_node_params::~ipa_node_params): Move from removed
39075 ipa_node_params_t::release.
39076 * symbol-summary.h (symbol_summary::m_released): New member.
39077 Do not release a summary twice. Do not allow to call finalizer
39078 for types of a summary that live in GGC memory.
39080 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39082 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39085 2017-02-02 Martin Sebor <msebor@redhat.com>
39087 PR middle-end/79275
39088 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39089 (format_string): Tighten up the range of output for non-constant
39090 strings and correct the expected range for wide non-constant strings.
39092 2017-02-02 Martin Sebor <msebor@redhat.com>
39094 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39096 PR middle-end/32003
39097 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39099 (-fdump-tree-@var): Add to index and document how to come up
39100 with pass-specific option and dump file names.
39101 (-fdump-passes): Clarify where to look for output.
39103 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39105 PR middle-end/77445
39106 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39107 statistics of the analyzed path; allow threading for speed when
39108 any of BBs along the path are optimized for speed.
39110 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39112 PR middle-end/78468
39113 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39114 settings of the virtual registers.
39117 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39119 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39120 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39123 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39125 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39126 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39128 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39130 * config/s390/s390.md: Add missing comments with the expanded
39132 * config/s390/vector.md: Likewise.
39133 * config/s390/vx-builtins.md: Likewise.
39135 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39138 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39139 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39140 conditions on a single line.
39142 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39144 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39145 __S390_VX__ to __VX__.
39147 2017-02-01 Andrew Pinski <apinski@cavium.com>
39149 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39150 stmt_info to record_stmt_cost.
39151 (vect_get_known_peeling_cost): Pass stmt_info if known to
39153 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39154 cpu_vector_cost field into
39155 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39156 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39157 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39158 splitting of scalar_stmt_cost and vec_stmt_cost.
39159 (thunderx_vector_cost): Likewise.
39160 (cortexa57_vector_cost): LIkewise.
39161 (exynosm1_vector_cost): Likewise.
39162 (xgene1_vector_cost): Likewise.
39163 (thunderx2t99_vector_cost): Improve after the splitting of the two
39165 (aarch64_builtin_vectorization_cost): Update for the splitting of
39166 scalar_stmt_cost and vec_stmt_cost.
39168 2017-02-01 Torvald Riegel <triegel@redhat.com>
39169 Richard Henderson <rth@redhat.com>
39171 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39172 conditional on existance of a fast atomic load.
39173 * optabs-query.c (can_atomic_load_p): New function.
39174 * optabs-query.h (can_atomic_load_p): Declare it.
39175 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39176 no fast atomic load is available for the particular size of access.
39177 (expand_atomic_compare_and_swap): Likewise.
39178 (expand_atomic_load): Likewise.
39179 (expand_atomic_store): Likewise.
39180 (expand_atomic_fetch_op): Likewise.
39181 * testsuite/lib/target-supports.exp
39182 (check_effective_target_sync_int_128): Remove x86 because it provides
39183 no fast atomic load.
39184 (check_effective_target_sync_int_128_runtime): Likewise.
39186 2017-02-01 Richard Biener <rguenther@suse.de>
39188 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39189 (graphite_transform_loops): Provide opt-info for optimized nests.
39190 * tree-parloop.c (parallelize_loops): Provide opt-info for
39191 parallelized loops.
39193 2017-02-01 Richard Biener <rguenther@suse.de>
39195 PR middle-end/79315
39196 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39197 was not set before.
39199 2017-02-01 Richard Biener <rguenther@suse.de>
39201 PR tree-optimization/71824
39202 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39203 Verify the loops are valid in the merged SESE region.
39204 (scop_detection::can_represent_loop_1): Check analyzing the
39205 evolution of the number of iterations in the region succeeds.
39207 2017-01-31 Ian Lance Taylor <iant@golang.org>
39209 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39210 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39212 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39214 PR preprocessor/79210
39215 * input.c (get_substring_ranges_for_loc): Replace line_width
39216 assertion with error-handling.
39218 2017-01-31 Richard Biener <rguenther@suse.de>
39220 PR tree-optimization/77318
39221 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39222 (create_pw_aff_from_tree): Take loop parameter.
39223 (add_condition_to_pbb): Pass loop of the condition to
39224 create_pw_aff_from_tree.
39226 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39228 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39229 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39231 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39235 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39237 (convert_int_to_float128): Likewise.
39238 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39239 (convert_int_to_float128): Likewise.
39240 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39241 (UNSPEC_IEEE128_CONVERT): Likewise.
39242 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39243 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39244 Use local variables for IBM extended format.
39245 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39246 (fix_trunc<mode>si2_fprs): Likewise.
39247 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39248 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39249 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39250 to know that we can now have integers of all sizes in vector
39252 (fix<uns>_<mode>di2_hw): Likewise.
39253 (float<uns>_<mode>si2_hw): Likewise.
39254 (fix_<mode>si2_hw): Likewise.
39255 (fixuns_<mode>si2_hw): Likewise.
39256 (float<uns>_<mode>di2_hw): Likewise.
39257 (float_<mode>di2_hw): Likewise.
39258 (float_<mode>si2_hw): Likewise.
39259 (floatuns_<mode>di2_hw): Likewise.
39260 (floatuns_<mode>si2_hw): Likewise.
39261 (xscvqp<su>wz_<mode>): Delete, no longer used.
39262 (xscvqp<su>dz_<mode>): Likewise.
39263 (xscv<su>dqp_<mode>): Likewise.
39264 (ieee128_mfvsrd_64bit): Likewise.
39265 (ieee128_mfvsrd_32bit): Likewise.
39266 (ieee128_mfvsrwz): Likewise.
39267 (ieee128_mtvsrw): Likewise.
39268 (ieee128_mtvsrd_64bit): Likewise.
39269 (ieee128_mtvsrd_32bit): Likewise.
39271 2017-01-31 Martin Liska <mliska@suse.cz>
39274 * ipa-prop.c (ipa_free_all_node_params): Call release method
39275 instead of ~sumbol_summary to not to trigger double times
39278 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39280 PR tree-optimization/71691
39281 * bitmap.h (class auto_bitmap): New.
39282 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39283 is_maybe_undefined instead of ssa_undefined_value_p.
39285 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39287 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39288 __S390_ARCH_LEVEL__ to __ARCH__.
39290 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39292 PR tree-optimization/79267
39293 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39294 if should_remove_lhs_p is true.
39296 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39299 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39300 (add_alignment_attribute): New.
39301 (base_type_die): Add alignment attribute.
39302 (subrange_type_die): Likewise.
39303 (modified_type_die): Likewise.
39304 (gen_array_type_die): Likewise.
39305 (gen_descr_array_type_die: Likewise.
39306 (gen_enumeration_type_die): Likewise.
39307 (gen_subprogram_die): Likewise.
39308 (gen_variable_die): Likewise.
39309 (gen_field_die): Likewise.
39310 (gen_ptr_to_mbr_type_die): Likewise.
39311 (gen_struct_or_union_type_die): Likewise.
39312 (gen_subroutine_type_die): Likewise.
39313 (gen_typedef_die): Likewise.
39314 (base_type_cmp): Compare alignment attribute.
39316 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39319 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39320 (setb_unsigned) New pattern for setb with CCUNS.
39321 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39322 subfc./subfe sequence to avoid overflow problems. Generate a
39323 shorter sequence with cmpld/setb for power9.
39324 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39325 for generating subfc. instruction.
39326 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39327 now uses this instruction.
39329 2017-01-30 Ian Lance Taylor <iant@google.com>
39332 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39333 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39335 2017-01-30 Martin Sebor <msebor@redhat.com>
39337 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39338 Move constant to the right of a relational operator.
39339 (get_mpfr_format_length, format_character, format_string): Ditto.
39340 (should_warn_p, maybe_warn): Same.
39342 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39344 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39347 * asan.c (get_translation_unit_decl): Remove function.
39348 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39350 2017-01-30 Martin Liska <mliska@suse.cz>
39352 PR gcov-profile/79259
39353 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39354 -fprofile-generate.
39356 2017-01-30 Martin Liska <mliska@suse.cz>
39359 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39360 Initialize variables with NULL value.
39362 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39365 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39367 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39369 2017-01-30 Richard Biener <rguenther@suse.de>
39371 PR tree-optimization/79276
39372 * tree-vrp.c (process_assert_insertions): Properly adjust common
39373 when removing a duplicate.
39375 2017-01-30 Richard Biener <rguenther@suse.de>
39377 PR tree-optimization/79256
39378 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39379 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39381 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39383 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39386 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39387 ("*r<noxa>sbg_<mode>_sll_bitmask")
39388 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39389 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39390 Use contiguous_bitmask_nowrap_operand.
39392 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39395 * config/rs6000/altivec.h (vec_xl): Revise #define.
39396 (vec_xst): Likewise.
39398 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39400 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39402 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39404 PR rtl-optimization/79194
39405 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39406 traps before call to bypass_conditional_jumps.
39408 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39410 PR tree-optimization/71374
39411 * lra-constraints.c (check_conflict_input_operands): New.
39412 (match_reload): Use it.
39414 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39417 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39418 account to calculate conflict_set.
39420 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39422 PR rtl-optimization/78559
39423 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39424 other_insn in combine.
39426 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39428 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39429 uint16_type_node for BT_UINT16.
39431 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39433 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39434 "RTL Tests" to menu.
39435 (GIMPLE Tests): New node.
39436 (RTL Tests): New node.
39438 2017-01-27 Richard Biener <rguenther@suse.de>
39440 PR tree-optimization/79245
39441 * tree-loop-distribution.c (distribute_loop): Apply cost
39442 modeling also to detected patterns.
39444 2017-01-27 Richard Biener <rguenther@suse.de>
39446 PR tree-optimization/71433
39447 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39448 (compare_assert_loc): New function.
39449 (process_assert_insertions): Sort and optimize assert locations
39450 to remove duplicates and push down identical assertions on
39451 edges to their destination block.
39453 2017-01-27 Richard Biener <rguenther@suse.de>
39455 PR tree-optimization/79244
39456 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39457 out SSA names even if abnormal.
39459 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39461 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39462 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39463 instead of MPFR_RNDN.
39465 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39468 * arm.c (arm_option_override): Don't call build_target_option_node
39469 until after doing all option overrides.
39470 (arm_valid_target_attribute_tree): Likewise.
39472 2017-01-27 Martin Liska <mliska@suse.cz>
39474 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39475 for {cd}tors and C++ {cd}tors.
39477 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39479 * config/s390/s390.md ("*setmem_long_and")
39480 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39482 2017-01-26 Martin Sebor <msebor@redhat.com>
39484 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39487 2017-01-26 Martin Sebor <msebor@redhat.com>
39489 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39490 HAVE_DFmode before using XFmode or DFmode.
39491 (parse_directive): Avoid using the z length modifier to avoid
39492 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39494 PR middle-end/78703
39495 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39496 to accept adjustment as an array.
39497 (get_int_range): New function.
39498 (struct directive): Make width and prec arrays.
39499 (directive::set_width, directive::set_precision): Call get_int_range.
39500 (format_integer, format_floating): Handle width and precision ranges.
39501 (format_string, parse_directive): Same.
39503 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39506 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39507 comdat_type_p set, just clone them, but keep the children in the
39511 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39512 which have direct callers with -fvar-tracking-assignments enabled
39514 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39515 inside of type units.
39517 2017-01-26 Martin Sebor <msebor@redhat.com>
39519 PR middle-end/78703
39520 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39522 (struct format_result): Replace number_chars, number_chars_min,
39523 and number_chars_max with a single member of struct result_range.
39525 (format_result::operator+=): Adjust.
39526 (struct fmtresult): Remove bounded. Handle likely and unlikely
39528 (fmtresult::adjust_for_width_or_precision): New function.
39529 (fmtresult:type_max_digits): New function.
39530 (bytes_remaining): Handle likely and unlikely counters.
39531 (min_bytes_remaining): Remove.
39532 (format_percent): Simplify.
39533 (format_integer, format_floating): Set likely and unlikely counters.
39534 (get_string_length, format_character, format_string): Same.
39535 (format_plain, should_warn_p): New function.
39536 (maybe_warn): Call should_warn_p. Update diagnostic messages
39537 and handle those for all directives, including plain strings.
39538 (format_directive): Handle likely and unlikely counters.
39539 Remove unnecessary quoting from diagnostics. Add an informational
39541 (add_bytes): Remove.
39542 (pass_sprintf_length::compute_format_length): Simplify.
39543 (try_substitute_return_value): Handle likely and unlikely counters.
39545 2017-01-26 Carl Love <cel@us.ibm.com>
39547 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39548 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39550 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39553 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39554 endianess for subregs into account.
39555 * lra-constraints.c (lra_constraints): Do risky transformations
39556 always on the first iteration.
39557 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39558 last_call_used_reg_set.
39559 (process_bb_lives): Define and use last_call_used_reg_set.
39560 * lra.c (lra): Always continue after lra_constraints on the first
39563 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39565 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39567 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39569 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39571 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39572 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39573 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39574 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39575 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39576 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39577 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39578 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39579 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39581 2017-01-26 Marek Polacek <polacek@redhat.com>
39584 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39585 for the third operand.
39587 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39589 PR middle-end/79236
39590 * omp-low.c (struct omp_context): Add simt_stmt field.
39591 (scan_omp_for): Return omp_context *.
39592 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39593 context to the _simt_ SIMD stmt.
39594 (lower_omp_for): For combined SIMD with sibling _simt_
39595 SIMD, make sure to use the same decls in _looptemp_
39596 clauses as in the sibling.
39598 2017-01-26 David Sherwood <david.sherwood@arm.com>
39600 PR middle-end/79212
39601 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39604 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39607 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39608 emit fld b; fld a; if possible.
39610 * brig-builtins.def: Update copyright years.
39611 * config/arm/arm_acle_builtins.def: Update copyright years.
39613 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39616 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39617 constraint instead of o for the stxsd instruction.
39619 2017-01-25 Carl Love <cel@us.ibm.com>
39621 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39622 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39624 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39626 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39628 2017-01-25 Richard Biener <rguenther@suse.de>
39630 PR tree-optimization/69264
39631 * target.def (vector_alignment_reachable): Improve documentation.
39632 * doc/tm.texi: Regenerate.
39633 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39635 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39636 earlier changes with respect to TYPE_USER_ALIGN.
39637 (vector_alignment_reachable_p): Likewise. Improve dumping.
39639 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39642 * config/arm/arm.md (xordi3): Force constant operand into a register
39645 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39647 * doc/invoke.texi (-fstore-merging): Correct default optimization
39648 levels at which it is enabled.
39649 (-O): Move -fstore-merging from list to...
39652 2017-01-25 Richard Biener <rguenther@suse.de>
39655 * omp-expand.c: Include debug.h.
39656 (expand_omp_taskreg): Make sure to generate early debug before
39657 outlining anything from a function.
39658 (expand_omp_target): Likewise.
39659 (grid_expand_target_grid_body): Likewise.
39661 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39664 * asan.c (get_translation_unit_decl): New function.
39665 (asan_add_global): Extract modules file name from globals
39666 TRANSLATION_UNIT_DECL name.
39668 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39671 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39672 for long calls with APCS frame and VFP.
39674 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39676 * cfg.c (original_copy_tables_initialized_p): New function.
39677 * cfg.h (original_copy_tables_initialized_p): New decl.
39678 * cfgrtl.c (relink_block_chain): Guard the call to
39679 free_original_copy_tables with a call to
39680 original_copy_tables_initialized_p.
39681 * cgraph.h (symtab_node::native_rtl_p): New decl.
39682 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39683 (symtab_node::needed_p): Don't assert for early assembly output
39684 for __RTL functions.
39685 (cgraph_node::finalize_function): Set "force_output" for __RTL
39687 (cgraph_node::analyze): Bail out early for __RTL functions.
39688 (analyze_functions): Update assertion to support __RTL functions.
39689 (cgraph_node::expand): Bail out early for __RTL functions.
39690 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39692 * function.h (struct function): Update comment for field
39694 * gimple-expr.c: Include "tree-pass.h".
39695 (gimple_has_body_p): Return false for __RTL functions.
39696 * Makefile.in (OBJS): Add run-rtl-passes.o.
39697 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39699 (gcc::pass_manager::get_clean_slate): New accessor.
39700 * passes.c: Include "insn-addr.h".
39701 (should_skip_pass_p): Add logging. Update logic for running
39702 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39703 property-provider override so it is only done for gimple passes.
39705 (skip_pass): New function.
39706 (execute_one_pass): Call skip_pass when skipping passes.
39707 * read-md.c (md_reader::read_char): Support filtering
39708 the input to a subset of line numbers.
39709 (md_reader::md_reader): Initialize fields
39710 m_first_line and m_last_line.
39711 (md_reader::read_file_fragment): New function.
39712 * read-md.h (md_reader::read_file_fragment): New decl.
39713 (md_reader::m_first_line): New field.
39714 (md_reader::m_last_line): New field.
39715 * read-rtl-function.c (function_reader::create_function): Only
39716 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39717 curr_properties. Set DECL_INITIAL to a dummy block.
39718 (read_rtl_function_body_from_file_range): New function.
39719 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39721 * run-rtl-passes.c: New file.
39722 * run-rtl-passes.h: New file.
39724 2017-01-24 Jeff Law <law@redhat.com>
39726 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39729 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39731 PR tree-optimization/79159
39732 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39733 (record_nonwrapping_iv): Improve boundary using above function if no
39734 value range information.
39736 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39737 Martin Jambor <mjambor@suse.cz>
39739 * brig-builtins.def: New file.
39740 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39741 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39742 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39743 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39744 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39745 * builtin-types.def (BT_INT8): New.
39746 (BT_INT16): Likewise.
39747 (BT_UINT8): Likewise.
39748 (BT_UINT16): Likewise.
39749 (BT_FN_ULONG): Likewise.
39750 (BT_FN_UINT_INT): Likewise.
39751 (BT_FN_UINT_ULONG): Likewise.
39752 (BT_FN_UINT_LONG): Likewise.
39753 (BT_FN_UINT_PTR): Likewise.
39754 (BT_FN_ULONG_PTR): Likewise.
39755 (BT_FN_INT8_FLOAT): Likewise.
39756 (BT_FN_INT16_FLOAT): Likewise.
39757 (BT_FN_UINT32_FLOAT): Likewise.
39758 (BT_FN_UINT16_FLOAT): Likewise.
39759 (BT_FN_UINT8_FLOAT): Likewise.
39760 (BT_FN_UINT64_FLOAT): Likewise.
39761 (BT_FN_UINT16_UINT32): Likewise.
39762 (BT_FN_UINT32_UINT16): Likewise.
39763 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39764 (BT_FN_INT_PTR_INT): Likewise.
39765 (BT_FN_UINT_PTR_UINT): Likewise.
39766 (BT_FN_LONG_PTR_LONG): Likewise.
39767 (BT_FN_ULONG_PTR_ULONG): Likewise.
39768 (BT_FN_VOID_UINT64_UINT64): Likewise.
39769 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39770 (BT_FN_INT8_INT8_INT8): Likewise.
39771 (BT_FN_INT16_INT16_INT16): Likewise.
39772 (BT_FN_INT_INT_INT): Likewise.
39773 (BT_FN_UINT_FLOAT_UINT): Likewise.
39774 (BT_FN_FLOAT_UINT_UINT): Likewise.
39775 (BT_FN_ULONG_UINT_UINT): Likewise.
39776 (BT_FN_ULONG_UINT_PTR): Likewise.
39777 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39778 (BT_FN_UINT_UINT_UINT): Likewise.
39779 (BT_FN_VOID_UINT_PTR): Likewise.
39780 (BT_FN_UINT_UINT_PTR: Likewise.
39781 (BT_FN_UINT32_UINT64_PTR): Likewise.
39782 (BT_FN_INT_INT_UINT_UINT): Likewise.
39783 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39784 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39785 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39786 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39787 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39788 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39789 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39790 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39791 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39792 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39793 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39794 * doc/frontends.texi: List BRIG FE.
39795 * doc/install.texi (Testing): Add BRIG tesring requirements.
39796 * doc/invoke.texi (Overall Options): Mention BRIG.
39797 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39799 2017-01-24 Richard Biener <rguenther@suse.de>
39801 PR translation/79208
39802 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39804 2017-01-24 Martin Jambor <mjambor@suse.cz>
39807 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39808 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39809 and known_contexts.
39811 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39813 PR middle-end/79123
39814 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39815 casts from signed to unsigned really don't have a range.
39817 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39819 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39820 GMP_RNDx for compatiblity.
39822 2017-01-24 Martin Liska <mliska@suse.cz>
39825 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39826 that would prevent us to call alloca with -1 as argument.
39828 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39830 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39831 Avoid -Wformat-security warning.
39833 2017-01-23 Andrew Pinski <apinski@cavium.com>
39835 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39838 2017-01-23 Martin Sebor <msebor@redhat.com>
39840 PR middle-end/78703
39841 * gimple-ssa-sprintf.c (warn_level): New global.
39842 (format_integer): Use it here and throughout the rest of the file.
39843 Use the same switch to compute sign as base.
39844 (maybe_warn): New function.
39845 (format_directive): Factor out warnings into maybe_warn.
39846 Add debugging output. Use warn_level.
39847 (add_bytes): Use warn_level.
39848 (pass_sprintf_length::compute_format_length): Add debugging output.
39849 (try_substitute_return_value): Same.
39850 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39852 PR middle-end/78703
39853 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39854 (struct fmtresult, format_integer, format_floating): Adjust.
39855 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39856 (get_string_length, format_string,format_directive): Same.
39857 (pass_sprintf_length::compute_format_length): Same.
39858 (try_substitute_return_value): Simplify slightly.
39860 PR middle-end/78703
39861 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39862 (fmtresult::operator+=): Outlined.
39863 (struct fmtresult): Add ctors.
39864 (struct conversion_spec): Rename...
39865 (struct directive): ...to this. Add and remove data members.
39866 (directive::set_width, directive::set_precision): New functions.
39867 (format_percent): Use fmtresult ctor.
39868 (get_width_and_precision): Remove.
39869 (format_integer): Make naming changes. Avoid computing width and
39871 (format_floating): Same. Adjust indentation.
39872 (format_character, format_none): New functions.
39873 (format_string): Moved character handling to format_character.
39874 (format_directive): Remove arguments, change return type.
39875 (parse_directive): New function.
39876 (pass_sprintf_length::compute_format_length): Move directive
39877 parsing to parse_directive.
39879 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39881 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39882 (assign_assembler_name_if_needed): ... this.
39883 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39884 (assign_assembler_name_if_needed): ... this.
39885 (free_lang_data_in_cgraph): Adjust callers.
39886 * cgraphunit.c (cgraph_node::analyze): Likewise.
39887 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39890 2017-01-23 Richard Biener <rguenther@suse.de>
39892 PR tree-optimization/79088
39893 PR tree-optimization/79188
39894 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39895 resetting loop bounds after last path deletion. Reset loop
39896 bounds of the target loop, make code match the comments.
39897 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39898 Make sure loops need no fixups.
39900 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39902 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39903 exponent support with double type for first argument.
39904 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39905 type returned by __builtin_vec_extract_sig,
39906 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39907 functions from "vector int" to "vector unsigned int" or from
39908 "vector long long int" to "vector unsigned long long int".
39909 Changed type returned by __builtin_vec_extract_exp,
39910 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39911 functions from "vector int" to "vector unsigned int" or from
39912 "vector long long int" to "vector unsigned long long int".
39913 Changed return type of __builtin_vec_test_data_class,
39914 __builtin_vec_test_data_class_sp, and
39915 __builtin_vec_test_data_class_dp from "vector int" to
39916 "vector bool int" or from "vector long long int" to "vector bool
39917 long long int" and changed second argument type from "unsigned
39918 int" to "int". Added new overloaded function forms "vector float
39919 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39920 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39921 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39922 double, vector unsigned long long int)" and "vector double
39923 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39924 long int)". Changed return type of
39925 __builtin_scalar_test_data_class and
39926 __builtin_scalar_test_data_class_sp and
39927 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39928 int" and changed second argument from "unsigned int" to "int".
39929 Changed type returned by __builtin_scalar_test_neg,
39930 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39931 from "int" to "bool int". Added new overloaded function form
39932 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39933 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39934 exponent double-precision with floating point first argument.
39935 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39936 documentation of scalar_test_data_class, scalar_test_neg,
39937 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39938 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39939 vec_test_data_class built-in functions to reflect refinements in
39940 their type signatures.
39942 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39944 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39946 (aarch64_elf_asm_destructor): Likewise.
39948 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39950 PR rtl-optimization/78634
39951 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39952 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39953 * ifcvt.c (noce_try_cmove): Add missing cost check.
39955 PR rtl-optimization/71724
39956 * combine.c (if_then_else_cond): Look for situations where it is
39957 beneficial to undo the work of one of the recursive calls.
39959 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39961 PR tree-optimization/70754
39962 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39963 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39964 combined stmt before it if not NULL.
39965 (combine_chains): Process refs reversely and compute dominance point
39968 2017-01-23 Martin Liska <mliska@suse.cz>
39970 PR tree-optimization/79196
39971 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39972 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39974 (strlen_optimize_stmt): Call the renamed function.
39976 2017-01-23 Michael Matz <matz@suse.de>
39978 PR tree-optimization/78384
39979 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39981 2017-01-23 Richard Biener <rguenther@suse.de>
39983 PR tree-optimization/79186
39984 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39985 both incoming edges before moving an assert.
39987 2017-01-23 Martin Jambor <mjambor@suse.cz>
39989 * ipa-prop.c (load_from_param_1): Removed.
39990 (load_from_unmodified_param): Bits from load_from_param_1 put back
39992 (load_from_param): Removed.
39993 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39994 with stmt. Reverted back to use of load_from_unmodified_param.
39996 2017-01-23 Martin Jambor <mjambor@suse.cz>
39999 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40000 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40001 field a pointer to garbage collected vector, mark lattices and
40002 ipcp_orig_node with GTY((skip)).
40003 (ipa_get_param_count): Adjust to descriptors being a pointer.
40004 (ipa_get_param): Likewise.
40005 (ipa_get_type): Likewise.
40006 (ipa_get_param_move_cost): Likewise.
40007 (ipa_set_param_used): Likewise.
40008 (ipa_get_controlled_uses): Likewise.
40009 (ipa_set_controlled_uses): Likewise.
40010 (ipa_is_param_used): Likewise.
40011 (ipa_node_params_t): Move into garbage collector. New methods insert
40013 (ipa_node_params_sum): Annotate wth GTY(()).
40014 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40016 (ipa_load_from_parm_agg): Adjust declaration.
40017 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40018 * ipa-profile.c (ipa_profile): Likewise.
40019 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40020 (ipa_populate_param_decls): Make descriptors parameter garbage
40022 (ipa_dump_param): Adjust to descriptors being a pointer.
40023 (ipa_alloc_node_params): Likewise.
40024 (ipa_initialize_node_params): Likewise.
40025 (load_from_param_1): Make descriptors parameter garbage collected.
40026 (load_from_unmodified_param): Likewise.
40027 (load_from_param): Likewise.
40028 (ipa_load_from_parm_agg): Likewise.
40029 (ipa_node_params::~ipa_node_params): Removed.
40030 (ipa_free_all_node_params): Remove call to delete operator.
40031 (ipa_node_params_t::insert): New.
40032 (ipa_node_params_t::remove): Likewise.
40033 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40034 copy known_csts and known_contexts vectors.
40035 (ipa_read_node_info): Adjust to descriptors being a pointer.
40036 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40038 (ipcp_transform_function): Make descriptors variable garbage
40041 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40043 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40044 * config/i386/avx512dqintrin.h: Ditto.
40045 * config/i386/avx512fintrin.h: Ditto.
40046 * config/i386/i386.c: Handle new builtins.
40047 * config/i386/i386-builtin.def: Add new builtins.
40048 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40049 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40051 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40052 Martin Liska <mliska@suse.cz>
40054 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40055 * asan.c (asan_expand_poison_ifn): Support stores and use
40056 appropriate ASAN report function.
40057 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40058 * internal-fn.def (ASAN_POISON_USE): Declare.
40059 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40060 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40061 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40062 ASAN_POISON calls w/o LHS.
40063 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40064 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40065 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40066 * gimplify.c (asan_poison_variables): Add attribute
40067 use_after_scope_memory to variables that really needs to live
40069 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40070 having the attribute.
40072 2017-01-23 Martin Liska <mliska@suse.cz>
40074 * asan.c (create_asan_shadow_var): New function.
40075 (asan_expand_poison_ifn): Likewise.
40076 * asan.h (asan_expand_poison_ifn): New declaration.
40077 * internal-fn.c (expand_ASAN_POISON): Likewise.
40078 * internal-fn.def (ASAN_POISON): New builtin.
40079 * sanopt.c (pass_sanopt::execute): Expand
40080 asan_expand_poison_ifn.
40081 * tree-inline.c (copy_decl_for_dup_finish): Make function
40083 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40084 * tree-ssa.c (is_asan_mark_p): New function.
40085 (execute_update_addresses_taken): Rewrite local variables
40086 (identified just by use-after-scope as addressable) into SSA.
40088 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40090 * doc/install.texi (Specific): opensource.apple.com uses https
40091 now. Remove trailing slash.
40093 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40095 * README.Portability: Remove note on an Irix compatibility issue.
40097 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40099 * gcov.c (INCLUDE_ALGORITHM): Define.
40100 (INCLUDE_VECTOR): Define.
40101 No longer include <vector> and <algorithm> directly.
40103 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40105 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40107 * doc/invoke.texi (Code Gen Options): Ditto.
40109 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40112 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40114 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40116 rtl-optimization/79125
40117 * cprop.c (local_cprop_pass): Handle cases where we make an
40118 unconditional trap.
40120 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40124 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40125 read from, for big endian.
40127 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40129 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40130 register pauth builtins for LP64 only.
40132 2017-01-20 Marek Polacek <polacek@redhat.com>
40135 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40138 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40140 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40142 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40143 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40144 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40146 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40149 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40150 in big-endian mode when they are not a single duplicated value.
40152 2017-01-20 Richard Biener <rguenther@suse.de>
40154 * BASE-VER: Bump to 7.0.1.
40156 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40158 * omp-low.c (omplow_simd_context): New struct. Use it...
40159 (lower_rec_simd_input_clauses): ...here and...
40160 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40161 references to idx, lane, max_vf, is_simt.
40163 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40165 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40168 2017-01-20 Martin Jambor <mjambor@suse.cz>
40170 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40171 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40173 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40174 (GTFILES): Rename hsa.c to hsa-common.c.
40175 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40176 * hsa-dump.c: Likewise.
40177 * hsa-gen.c: Likewise.
40178 * hsa-regalloc.c: Likewise.
40179 * ipa-hsa.c: Likewise.
40180 * omp-expand.c: Likewise.
40181 * omp-low.c: Likewise.
40182 * toplev.c: Likewise.
40184 2017-01-20 Marek Polacek <polacek@redhat.com>
40187 * doc/invoke.texi: Document -Wduplicated-branches.
40188 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40189 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40190 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40191 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40192 return 0 only when not OEP_LEXICOGRAPHIC.
40193 (fold_build_cleanup_point_expr): Use the expression
40194 location when building CLEANUP_POINT_EXPR.
40195 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40196 * tree.c (add_expr): Handle error_mark_node.
40198 2017-01-20 Martin Liska <mliska@suse.cz>
40201 * tree-profile.c (init_ic_make_global_vars): Do not call
40203 (gimple_init_gcov_profiler): Likewise.
40205 2017-01-20 Martin Liska <mliska@suse.cz>
40208 * cgraph.h (maybe_create_reference): Remove argument and
40210 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40212 * ipa-cp.c (create_specialized_node): Likewise.
40213 * symtab.c (symtab_node::maybe_create_reference): Handle
40214 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40216 2017-01-20 Martin Liska <mliska@suse.cz>
40218 * read-rtl-function.c (function_reader::create_function): Use
40219 build_decl instread of build_decl_stat.
40221 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40223 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40224 * config/i386/avx512dqintrin.h: Ditto.
40225 * config/i386/avx512fintrin.h: Ditto.
40226 * config/i386/i386-builtin-types.def: Add new types.
40227 * config/i386/i386.c: Handle new types.
40228 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40229 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40230 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40231 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40232 (__builtin_ia32_kshiftridi): New.
40233 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40235 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40239 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40240 define to rs6000_init_stack_protect_guard.
40241 (rs6000_init_stack_protect_guard): New function.
40243 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40244 Yunqiang Su <yunqiang.su@imgtec.com>
40246 * config.gcc (supported_defaults): Add madd4.
40247 (with_madd4): Add validation.
40248 (all_defaults): Add madd4.
40249 * config/mips/mips.opt (mmadd4): New option.
40250 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40252 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40254 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40255 (ISA_HAS_FUSED_MADD4): Likewise.
40256 * doc/invoke.texi (-mmadd4): Document the new option.
40257 * doc/install.texi (--with-madd4): Document the new option.
40259 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40261 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40262 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40263 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40264 (aarch64_init_pauth_hint_builtins): New.
40265 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40266 (aarch64_expand_builtin): Expand new builtins.
40268 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40270 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40271 * combine-stack-adj.c (no_unhandled_cfa): Handle
40272 REG_CFA_TOGGLE_RA_MANGLE.
40273 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40274 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40275 info for return address signing.
40276 (aarch64_expand_epilogue): Likewise.
40278 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40280 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40281 * config/aarch64/aarch64-protos.h
40282 (aarch64_return_address_signing_enabled): New declaration.
40283 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40285 (aarch64_expand_prologue): Sign return address before it's pushed onto
40287 (aarch64_expand_epilogue): Authenticate return address fetched from
40289 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40290 (aarch64_attributes): New function attributes for "sign-return-address".
40291 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40292 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40293 ("*do_return"): Generate combined instructions according to key index.
40294 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40295 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40297 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40298 * config/aarch64/aarch64.opt (msign-return-address=): New.
40299 * doc/extend.texi (AArch64 Function Attributes): Documents
40300 "sign-return-address=".
40301 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40303 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40305 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40306 overall option summary.
40308 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40310 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40311 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40312 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40313 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40315 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40317 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40318 -mpower9-minmax by default for -mcpu=power9.
40319 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40320 128-bit floating point.
40322 2017-01-20 Alan Modra <amodra@gmail.com>
40324 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40325 optimizing for size.
40327 2017-01-20 Alan Modra <amodra@gmail.com>
40330 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40331 for strcmp and strncmp from corresponding builtin decl.
40333 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40335 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40336 instead of i386/rtems-64.h.
40337 * config/i386/rtems-64.h: Remove.
40339 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40343 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40345 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40347 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40349 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40350 Change int to HOST_WIDE_INT.
40351 * config/aarch64/aarch64-protos.h
40352 (aarch64_simd_gen_const_vector_dup): Likewise.
40353 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40355 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40357 * langhooks-def.h (lhd_type_for_size): New decl.
40358 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40359 * langhooks.c (lhd_type_for_size): New function, taken from
40362 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40364 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40365 define_bypass for CR latency.
40366 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40367 (power9-alu2): Add define_bypass for CR latency.
40369 (power9-mul): Update insn latency.
40370 (power9-mul-compare): Update insn latency, bypass latency and remove
40373 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40375 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40377 * config/aarch64/aarch64.md
40378 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40379 aarch64_nopcrelative_literal_loads.
40380 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40382 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40384 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40385 TARGET_LOONGSON_3A.
40386 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40388 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40391 * config.gcc (supported_defaults): Add lxc1-sxc1.
40392 (with_lxc1_sxc1): Add validation.
40393 (all_defaults): Add lxc1-sxc1.
40394 * config/mips/mips.opt (mlxc1-sxc1): New option.
40395 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40397 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40398 __mips_no_lxc1_sxc1.
40399 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40400 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40401 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40403 2017-01-19 Richard Biener <rguenther@suse.de>
40405 PR tree-optimization/72488
40406 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40407 sure to restore SSA info.
40408 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40410 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40412 PR rtl-optimization/79121
40413 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40414 of the inner type when shifting an extended value.
40416 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40419 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40420 interposable aliases.
40422 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40425 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40426 Use the evmergelohi instruction.
40427 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40428 (mov_sitf_e500_subreg8_2_be): Likewise.
40429 (mov_sitf_e500_subreg12_2_le): Likewise.
40430 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40431 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40432 (mov_sitf_e500_subreg8_2_le): Likewise.
40433 (mov_sitf_e500_subreg12_2_be): Likewise.
40435 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40437 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40438 attribute from vecsimple to vecperm.
40439 (altivec_vbpermq2): Likewise.
40441 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40444 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40446 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40447 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40448 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40449 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40450 case where N arg is SIZE_MAX.
40451 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40452 (cmpstrsi): Add pattern.
40454 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40457 __builtin_vec_revb builtins.
40458 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40459 built-in functions to support generation of the ISA 3.0 XXBR<x>
40460 vector byte reverse instructions.
40461 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40462 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40463 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40464 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40465 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40466 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40467 (P9V_BUILTIN_VEC_REVB): Likewise.
40468 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40469 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40470 (p9_xxbrq_v16qi): Likewise.
40471 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40472 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40473 (p9_xxbrh_v8hi): Likewise.
40474 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40475 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40476 vec_revb built-in functions.
40478 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40480 PR rtl-optimization/78952
40481 * config/i386/i386.md (any_extract): New code iterator.
40482 (*insvqi_2): Use any_extract for source operand.
40483 (*insvqi_3): Use any_shiftrt for source operand.
40485 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40487 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40489 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40491 2017-01-18 Matthias Klose <doko@ubuntu.com>
40493 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40495 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40497 * config/rs6000/altivec.h (vec_bperm): Change #define.
40498 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40499 (altivec_vbpermq2): New define_insn.
40500 (altivec_vbpermd): Likewise.
40501 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40502 function interface.
40503 (VBPERMD): Likewise.
40504 (VBPERM): New polymorphic function interface.
40505 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40506 Add entries for P9V_BUILTIN_VEC_VBPERM.
40507 * doc/extend.texi: Add interfaces for vec_bperm.
40509 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40511 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40512 first letter of error messages.
40513 (s390_resolve_overloaded_builtin): Likewise.
40514 * config/s390/s390.c (s390_expand_builtin): Likewise.
40515 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40516 (s390_valid_target_attribute_inner_p): Likewise.
40517 * config/s390/s390.md ("tabort"): Likewise.
40519 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40521 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40522 (ISA_AVOID_DIV_HILO): New macro.
40523 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40524 (ISA_HAS_DDIV): Likewise.
40526 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40528 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40530 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40532 * doc/invoke.texi (fabi-version): Spelling fix.
40534 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40537 * doc/invoke.texi (fabi-version): Mention mangling fix for
40540 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40543 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40545 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40548 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40551 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40553 (stack_protect_test): Ditto.
40554 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40555 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40557 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40558 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40559 -mstack-protector-guard-offset=.
40560 (RS/6000 and PowerPC Options): Ditto.
40562 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40564 * config/i386/i386.h (MASK_CLASS_P): New define.
40565 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40566 there are no registers from different register sets also when
40567 mask registers are used. Update function comment.
40568 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40569 to (*k/*r) and (*k/*km) alternatives.
40571 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40573 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40574 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40575 (EH_RETURN_HANDLER_RTX): New define.
40576 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40577 Force frame pointer in EH return functions.
40578 (aarch64_expand_epilogue): Add barrier for eh_return.
40579 (aarch64_final_eh_return_addr): Remove.
40580 (aarch64_eh_return_handler_rtx): New function.
40581 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40583 (aarch64_eh_return_handler_rtx): New prototype.
40585 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40587 * config/rs6000/altivec.h (vec_rlmi): New #define.
40588 (vec_vrlnm): Likewise.
40589 (vec_rlnm): Likewise.
40590 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40591 (UNSPEC_VRLNM): Likewise.
40592 (VIlong): New mode iterator.
40593 (altivec_vrl<VI_char>mi): New define_insn.
40594 (altivec_vrl<VI_char>nm): Likewise.
40595 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40597 (VRLDNM): Likewise.
40598 (RLNM): New polymorphic function entry.
40599 (VRLWMI): New monomorphic function entry.
40600 (VRLDMI): Likewise.
40601 (RLMI): New polymorphic function entry.
40602 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40603 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40604 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40607 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40610 * dwarf2out.c (field_byte_offset): Restore the
40611 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40612 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40613 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40616 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40619 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40621 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40624 * dwarf2out.c (add_data_member_location_attribute): For constant
40625 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40626 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40627 DW_AT_byte_size attributes.
40629 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40631 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40632 after forcing to constant memory when the code model is medium.
40634 2017-01-17 Julia Koval <julia.koval@intel.com>
40637 * config/i386/avx512fintrin.h
40638 (_mm512_i32gather_ps): Change __addr type to void const*.
40639 (_mm512_mask_i32gather_ps): Ditto.
40640 (_mm512_i32gather_pd): Ditto.
40641 (_mm512_mask_i32gather_pd): Ditto.
40642 (_mm512_i64gather_ps): Ditto.
40643 (_mm512_mask_i64gather_ps): Ditto.
40644 (_mm512_i64gather_pd): Ditto.
40645 (_mm512_mask_i64gather_pd): Ditto.
40646 (_mm512_i32gather_epi32): Ditto.
40647 (_mm512_mask_i32gather_epi32): Ditto.
40648 (_mm512_i32gather_epi64): Ditto.
40649 (_mm512_mask_i32gather_epi64): Ditto.
40650 (_mm512_i64gather_epi32): Ditto.
40651 (_mm512_mask_i64gather_epi32): Ditto.
40652 (_mm512_i64gather_epi64): Ditto.
40653 (_mm512_mask_i64gather_epi64): Ditto.
40654 (_mm512_i32scatter_ps): Change __addr type to void*.
40655 (_mm512_mask_i32scatter_ps): Ditto.
40656 (_mm512_i32scatter_pd): Ditto.
40657 (_mm512_mask_i32scatter_pd): Ditto.
40658 (_mm512_i64scatter_ps): Ditto.
40659 (_mm512_mask_i64scatter_ps): Ditto.
40660 (_mm512_i64scatter_pd): Ditto.
40661 (_mm512_mask_i64scatter_pd): Ditto.
40662 (_mm512_i32scatter_epi32): Ditto.
40663 (_mm512_mask_i32scatter_epi32): Ditto.
40664 (_mm512_i32scatter_epi64): Ditto.
40665 (_mm512_mask_i32scatter_epi64): Ditto.
40666 (_mm512_i64scatter_epi32): Ditto.
40667 (_mm512_mask_i64scatter_epi32): Ditto.
40668 (_mm512_i64scatter_epi64): Ditto.
40669 (_mm512_mask_i64scatter_epi64): Ditto.
40670 * config/i386/avx512pfintrin.h
40671 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40672 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40673 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40674 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40675 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40676 (_mm512_prefetch_i32scatter_ps): Ditto.
40677 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40678 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40679 (_mm512_prefetch_i64scatter_pd): Ditto.
40680 (_mm512_prefetch_i64scatter_ps): Ditto.
40681 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40682 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40683 * config/i386/avx512vlintrin.h
40684 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40685 (_mm_mmask_i32gather_ps): Ditto.
40686 (_mm256_mmask_i32gather_pd): Ditto.
40687 (_mm_mmask_i32gather_pd): Ditto.
40688 (_mm256_mmask_i64gather_ps): Ditto.
40689 (_mm_mmask_i64gather_ps): Ditto.
40690 (_mm256_mmask_i64gather_pd): Ditto.
40691 (_mm_mmask_i64gather_pd): Ditto.
40692 (_mm256_mmask_i32gather_epi32): Ditto.
40693 (_mm_mmask_i32gather_epi32): Ditto.
40694 (_mm256_mmask_i32gather_epi64): Ditto.
40695 (_mm_mmask_i32gather_epi64): Ditto.
40696 (_mm256_mmask_i64gather_epi32): Ditto.
40697 (_mm_mmask_i64gather_epi32): Ditto.
40698 (_mm256_mmask_i64gather_epi64): Ditto.
40699 (_mm_mmask_i64gather_epi64): Ditto.
40700 (_mm256_i32scatter_ps): Change __addr type to void*.
40701 (_mm256_mask_i32scatter_ps): Ditto.
40702 (_mm_i32scatter_ps): Ditto.
40703 (_mm_mask_i32scatter_ps): Ditto.
40704 (_mm256_i32scatter_pd): Ditto.
40705 (_mm256_mask_i32scatter_pd): Ditto.
40706 (_mm_i32scatter_pd): Ditto.
40707 (_mm_mask_i32scatter_pd): Ditto.
40708 (_mm256_i64scatter_ps): Ditto.
40709 (_mm256_mask_i64scatter_ps): Ditto.
40710 (_mm_i64scatter_ps): Ditto.
40711 (_mm_mask_i64scatter_ps): Ditto.
40712 (_mm256_i64scatter_pd): Ditto.
40713 (_mm256_mask_i64scatter_pd): Ditto.
40714 (_mm_i64scatter_pd): Ditto.
40715 (_mm_mask_i64scatter_pd): Ditto.
40716 (_mm256_i32scatter_epi32): Ditto.
40717 (_mm256_mask_i32scatter_epi32): Ditto.
40718 (_mm_i32scatter_epi32): Ditto.
40719 (_mm_mask_i32scatter_epi32): Ditto.
40720 (_mm256_i32scatter_epi64): Ditto.
40721 (_mm256_mask_i32scatter_epi64): Ditto.
40722 (_mm_i32scatter_epi64): Ditto.
40723 (_mm_mask_i32scatter_epi64): Ditto.
40724 (_mm256_i64scatter_epi32): Ditto.
40725 (_mm256_mask_i64scatter_epi32): Ditto.
40726 (_mm_i64scatter_epi32): Ditto.
40727 (_mm_mask_i64scatter_epi32): Ditto.
40728 (_mm256_i64scatter_epi64): Ditto.
40729 (_mm256_mask_i64scatter_epi64): Ditto.
40730 (_mm_i64scatter_epi64): Ditto.
40731 (_mm_mask_i64scatter_epi64): Ditto.
40732 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40733 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40734 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40735 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40736 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40737 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40738 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40739 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40740 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40741 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40742 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40743 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40744 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40745 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40746 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40747 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40748 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40749 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40750 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40751 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40752 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40753 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40754 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40755 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40756 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40757 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40758 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40759 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40760 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40761 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40762 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40763 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40764 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40765 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40766 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40767 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40768 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40769 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40770 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40771 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40772 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40773 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40774 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40775 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40776 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40777 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40778 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40779 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40780 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40781 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40782 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40783 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40784 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40785 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40786 definitions accordingly.
40788 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40789 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40792 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40795 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40798 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40799 hard regs for inner regno.
40801 2017-01-17 Martin Liska <mliska@suse.cz>
40804 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40805 assumption and add comment.
40807 2017-01-17 Nathan Sidwell <nathan@acm.org>
40809 * ipa-visibility.c (localize_node): New function, broken out of ...
40810 (function_and_variable_visibility): ... here. Call it.
40812 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40814 PR middle-end/77445
40815 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40816 correctly set frequency of oudgoing edge.
40817 (duplicate_thread_path): Fix profile updating.
40819 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40822 * configure.ac: Add GCC_BASE_VER.
40823 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40824 version from BASE-VER file.
40825 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40826 (gcc.o): Depend on $(BASEVER).
40827 * common.opt (dumpfullversion): New option.
40828 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40829 * doc/invoke.texi: Document -dumpfullversion.
40830 * doc/install.texi: Document --with-gcc-major-version-only.
40831 * configure: Regenerated.
40833 2017-01-17 Richard Biener <rguenther@suse.de>
40835 PR tree-optimization/71433
40836 * tree-vrp.c (register_new_assert_for): Merge same asserts
40837 on all incoming edges.
40838 (process_assert_insertions_for): Handle insertions at the
40841 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40843 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40844 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40846 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40849 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40852 2017-01-17 Alan Modra <amodra@gmail.com>
40855 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40856 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40857 symbolic stack limit when pic.
40859 2017-01-16 Martin Sebor <msebor@redhat.com>
40861 PR tree-optimization/78608
40862 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40864 2017-01-16 Jeff Law <law@redhat.com>
40867 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40868 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40869 for several include directories that may be relative to sysroot.
40870 * config/i386/x-mingw32 (gplus_includedir): Define.
40871 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40872 (native_system_includedir): Likewise.
40873 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40874 override if TARGET_SYSTEM_ROOT is defined.
40875 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40877 PR tree-optimization/79090
40878 PR tree-optimization/33562
40879 PR tree-optimization/61912
40880 PR tree-optimization/77485
40881 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40882 and computed trims into the dump file.
40884 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40886 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40888 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40891 * gimplify.c (gimplify_init_constructor): If want_value and
40892 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40896 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40897 sequence. Formatting fixes.
40898 (doloop_optimize): Formatting fixes.
40901 * gcc.c (debug_level_greater_than_spec_func): New function.
40902 (static_spec_functions): Add debug-level-gt spec function.
40903 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40905 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40906 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40907 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40908 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40909 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40910 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40912 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40914 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40915 QImode fixups to general and mask registers only.
40917 2017-01-16 Carl Love <cel@us.ibm.com>
40919 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40920 for built-in functions
40921 vector signed char vec_nabs (vector signed char)
40922 vector signed short vec_nabs (vector signed short)
40923 vector signed int vec_nabs (vector signed int)
40924 vector signed long long vec_nabs (vector signed long long)
40925 vector float vec_nabs (vector float)
40926 vector double vec_nabs (vector double)
40927 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40929 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40930 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40931 * doc/extend.texi: Update the documentation file for the new built-in
40934 2017-01-16 Martin Sebor <msebor@redhat.com>
40936 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40939 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40941 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40942 UNSPEC_VSX__XXSPLTD to require special splat handling.
40944 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40947 * system.h: Poison strndup.
40949 2017-01-16 Alan Modra <amodra@gmail.com>
40952 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40955 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40957 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40959 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40961 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40962 call recog here. Assert that INSN_CODE (insn) is non-negative.
40964 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40967 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40969 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40970 in the currently scheduled RTL fragment.
40972 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40974 PR rtl-optimization/78751
40975 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40978 2017-01-14 Jeff Law <law@redhat.com>
40980 PR tree-optimization/79090
40981 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40982 variable length stores.
40983 (compute_trims): Delete dead assignment to *trim_tail.
40984 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40987 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40989 PR rtl-optimization/78626
40990 PR rtl-optimization/78727
40991 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40992 of a block, and split such blocks after everything else is finished.
40994 2017-01-14 Alan Modra <amodra@gmail.com>
40997 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40998 target legitimate_combined_insn.
40999 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41000 (rs6000_legitimate_combined_insn): New function.
41001 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41003 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41004 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41005 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41007 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41009 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41011 2017-01-13 Jeff Law <law@redhat.com>
41013 PR tree-optimization/33562
41014 PR tree-optimization/61912
41015 PR tree-optimization/77485
41016 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41018 (delete_dead_assignment): Likewise.
41019 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41020 statement to delete_dead_call and delete_dead_assignment.
41022 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41025 * substring-locations.c (format_warning_va): Strengthen case 1 so
41026 that both endpoints of the substring must be within the format
41027 range for just the substring to be printed.
41029 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41031 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41032 * config/i386/i386.c (ix86_target_string): Add missing options
41033 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41034 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41035 flags_other and ix86_target_other to flags2_other. Display unknown
41037 (ix86_valid_target_attribute_inner_p): Add missing options and
41038 reorder options by implied ISAs, as in ix86_target_string.
41040 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41042 * hash-table.h (hash_table::too_empty_p): New function.
41043 (hash_table::expand): Use it.
41044 (hash_table::traverse): Likewise.
41045 (hash_table::empty_slot): Use sizeof (value_type) instead of
41046 sizeof (PTR) to convert bytes to elements. Shrink the table
41047 if the current size is excessive for the current number of
41050 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41052 * ira-costs.c (record_reg_classes): Break from the inner loop
41053 early once alt_fail is known to be true. Update outer loop
41054 handling accordingly.
41056 2017-01-13 Jeff Law <law@redhat.com>
41058 * tree-ssa-dse.c (decrement_count): New function.
41059 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41060 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41061 when we know the partially dead statement is a mem* function.
41063 PR tree-optimization/61912
41064 PR tree-optimization/77485
41065 * tree-ssa-dse.c: Include expr.h.
41066 (maybe_trim_constructor_store): New function.
41067 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41069 PR tree-optimization/33562
41070 PR tree-optimization/61912
41071 PR tree-optimization/77485
41072 * doc/invoke.texi: Document new dse-max-object-size param.
41073 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41074 * tree-ssa-dse.c: Include params.h.
41075 (dse_store_status): New enum.
41076 (initialize_ao_ref_for_dse): New, partially extracted from
41078 (valid_ao_ref_for_dse, normalize_ref): New.
41079 (setup_live_bytes_from_ref, compute_trims): Likewise.
41080 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41081 (maybe_trim_partially_dead_store): Likewise.
41082 (maybe_trim_complex_store): Likewise.
41083 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41084 Track what bytes live from the original store. Return tri-state
41085 for dead, partially dead or live.
41086 (dse_dom_walker): Add constructor, destructor and new private members.
41087 (delete_dead_call, delete_dead_assignment): New extracted from
41089 (dse_optimize_stmt): Make a member of dse_dom_walker.
41090 Use initialize_ao_ref_for_dse.
41092 PR tree-optimization/33562
41093 PR tree-optimization/61912
41094 PR tree-optimization/77485
41095 * sbitmap.h (bitmap_count_bits): Prototype.
41096 (bitmap_clear_range, bitmap_set_range): Likewise.
41097 * sbitmap.c (bitmap_clear_range): New function.
41098 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41100 2017-01-13 Martin Liska <mliska@suse.cz>
41103 * function.c (set_cfun): Add new argument force.
41104 * function.h (set_cfun): Likewise.
41105 * ipa-inline-transform.c (inline_call): Use the function when
41106 strict alising from is dropped for function we inline to.
41108 2017-01-13 Richard Biener <rguenther@suse.de>
41110 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41111 for dumping GIMPLE INTEGER_CSTs.
41113 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41115 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41116 to 201112L since C++17.
41118 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41121 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41122 if -fsanitize=kernel-address is present.
41124 2017-01-13 Richard Biener <rguenther@suse.de>
41126 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41127 as _Literal ( type ) number in case usual suffixes do not
41128 preserve all information.
41130 2017-01-13 Richard Biener <rguenther@suse.de>
41132 PR tree-optimization/77283
41133 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41134 and ssa-iterators.h.
41135 (is_feasible_trace): Implement a cost model based on joiner
41138 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41141 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41142 char or short to __float128/_Float128 directly.
41144 2017-01-12 Martin Sebor <msebor@redhat.com>
41146 to -Wformat-overflow.
41147 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41148 (min_bytes_remaining): Same.
41149 (get_string_length): Same.
41150 (format_string): Same.
41151 (format_directive): Same.
41153 (pass_sprintf_length::handle_gimple_call): Same.
41155 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41157 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41158 info.nowrite calls with no lhs that can't throw. Return bool
41159 whether gsi_remove has been called or not.
41160 (pass_sprintf_length::handle_gimple_call): Return bool whether
41161 try_substitute_return_value called gsi_remove. Formatting fix.
41162 (pass_sprintf_length::execute): Don't use gsi_remove if
41163 handle_gimple_call returned true.
41166 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41167 be removed due to side-effects, don't remove following barrier nor
41168 turn the successor edge into fallthru edge.
41170 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41173 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41174 element-reversing loads and stores as not swappable.
41176 2017-01-12 Nathan Sidwell <nathan@acm.org>
41177 Nicolai Stange <nicstange@gmail.com>
41179 * combine.c (try_combine): Don't ignore result of overlap checking
41180 loop. Combine overlap & asm check into single loop.
41182 2017-01-12 Richard Biener <rguenther@suse.de>
41184 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41185 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41187 2017-01-12 Richard Biener <rguenther@suse.de>
41189 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41190 and TS_TARGET_OPTION directly derive from TS_BASE.
41191 * tree-core.h (tree_optimization_option): Derive from tree_base.
41192 (tree_target_option): Likewise.
41194 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41196 * config/i386/i386.c (memory_address_length): Increase len
41197 only when rip_relative_addr_p returns false.
41199 2017-01-11 Julia Koval <julia.koval@intel.com>
41201 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41202 (OPTION_MASK_ISA_SGX_SET): New.
41203 (ix86_handle_option): Handle OPT_msgx.
41204 * config.gcc: Added sgxintrin.h.
41205 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41206 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41207 * config/i386/i386.c (ix86_target_string): Add -msgx.
41209 (ix86_option_override_internal): Handle new options.
41210 (ix86_valid_target_attribute_inner_p): Add sgx.
41211 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41212 * config/i386/i386.opt: Add msgx.
41213 * config/i386/sgxintrin.h: New file.
41214 * config/i386/x86intrin.h: Add sgxintrin.h.
41216 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41219 * fold-const.c (maybe_nonzero_address): Return 1 for function
41221 (tree_single_nonzero_warnv_p): Don't handle function local objects
41225 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41228 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41231 * opts.c: Include "spellcheck.h"
41232 (struct string_fragment): New struct.
41233 (struct edit_distance_traits<const string_fragment &>): New
41235 (get_closest_sanitizer_option): New function.
41236 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41238 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41240 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41242 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41243 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41244 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41245 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41246 for initial die_offset if dwarf_split_debug_info.
41247 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41248 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41250 (output_skeleton_debug_sections): Formatting fix. Use
41251 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41252 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41254 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41256 * config/arm/cortex-a53.md: Add bypasses for
41257 cortex_a53_r2f_cvt.
41258 (cortex_a53_r2f): Only use for transfers.
41259 (cortex_a53_f2r): Likewise.
41260 (cortex_a53_r2f_cvt): Add reservation for conversions.
41261 (cortex_a53_f2r_cvt): Likewise.
41263 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41265 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41266 to all inlined functions, change static to extern.
41268 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41271 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41273 (arm_assemble_integer): Likewise.
41275 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41277 * config.gcc: Use new awk script to check CPU, FPU and architecture
41278 parameters for --with-... options.
41279 * config/arm/parsecpu.awk: New file
41280 * config/arm/arm-cpus.in: New file.
41281 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41283 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41285 * config/arm/t-arm: Update dependency rules.
41286 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41287 of processing .def files.
41288 * config/arm/genopt.sh: Deleted.
41289 * config/arm/gentune.sh: Deleted.
41290 * config/arm/arm-cores.def: Deleted.
41291 * config/arm/arm-arches.def: Deleted.
41292 * config/arm/arm-fpus.def: Deleted.
41293 * config/arm/arm-tune.md: Regenerated.
41294 * config/arm/arm-tables.opt: Regenerated.
41295 * config/arm/arm-cpu.h: New generated file.
41296 * config/arm/arm-cpu-data.h: New generated file.
41297 * config/arm/arm-cpu-cdata.h: New generated file.
41299 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41302 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41304 (input_varpool_node): Unpack dynamically_initialized bit.
41306 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41308 PR rtl-optimization/79032
41309 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41310 the alignment of the adjusted memory reference against that of MODE,
41311 instead of the alignment of the original memory reference.
41313 2017-01-11 Martin Jambor <mjambor@suse.cz>
41315 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41317 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41318 decorated functions.
41320 2017-01-11 Richard Biener <rguenther@suse.de>
41322 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41323 set range/nonnull info for PHI results. Do not set it on
41324 stmts marked for removal.
41326 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41328 * expr.c (store_field): In the bitfield case, fetch the return value
41329 from the registers before applying a single big-endian adjustment.
41330 Always do a final load for a BLKmode value not larger than a word.
41332 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41335 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41336 that we correctly handle column numbers greater than
41337 LINE_MAP_MAX_COLUMN_NUMBER.
41339 2017-01-10 Martin Sebor <msebor@redhat.com>
41341 PR middle-end/78245
41342 * gimple-ssa-sprintf.c (get_destination_size): Call
41343 {init,fini}object_sizes.
41344 * tree-object-size.c (addr_object_size): Adjust.
41345 (pass_through_call): Adjust.
41346 (pass_object_sizes::execute): Adjust.
41347 * tree-object-size.h (fini_object_sizes): Declare.
41349 2017-01-10 Martin Sebor <msebor@redhat.com>
41351 PR tree-optimization/78775
41352 * builtins.c (get_size_range): Move...
41353 * calls.c: ...to here.
41354 (alloc_max_size): Accept zero argument.
41355 (operand_signed_p): Remove.
41356 (maybe_warn_alloc_args_overflow): Call get_size_range.
41357 * calls.h (get_size_range): Declare.
41359 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41361 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41362 from TI's devices.csv file as of September 2016.
41363 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41365 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41367 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41368 * doc/invoke.texi: Likewise.
41369 * doc/md.texi: Likewise.
41370 * doc/objc.texi: Likewise.
41372 2017-01-10 Joshua Conner <joshconner@google.com>
41374 * config/arm/fuchsia-elf.h: New file.
41375 * config/fuchsia.h: New file.
41376 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41377 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41379 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41381 2016-01-10 Richard Biener <rguenther@suse.de>
41383 PR tree-optimization/79034
41384 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41385 Propagate out degenerate PHIs in the joiner.
41387 2017-01-10 Martin Liska <mliska@suse.cz>
41389 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41390 (sort_congruence_classes_by_decl_uid): Likewise.
41391 (sort_congruence_class_groups_by_decl_uid): Likewise.
41392 (sem_item_optimizer::merge_classes): Sort class, groups in these
41393 classes and members in the groups by DECL_UID of declarations.
41394 This would make merge operations stable.
41396 2017-01-10 Martin Liska <mliska@suse.cz>
41398 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41399 usage of m_classes_vec.
41400 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41401 (sem_item_optimizer::get_group_by_hash): Likewise.
41402 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41403 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41404 (sem_item_optimizer::verify_classes): Likewise.
41405 (sem_item_optimizer::process_cong_reduction): Likewise.
41406 (sem_item_optimizer::dump_cong_classes): Likewise.
41407 (sem_item_optimizer::merge_classes): Likewise.
41408 * ipa-icf.h (congruence_class_hash): Rename from
41409 congruence_class_group_hash. Remove declaration of m_classes_vec.
41411 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41413 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41414 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41415 * config.gcc: Add avx512vpopcntdqintrin.h.
41416 * config/i386/avx512vpopcntdqintrin.h: New.
41417 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41418 * config/i386/i386-builtin-types.def: Add new types.
41419 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41420 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41421 __builtin_ia32_vpopcountq_v8di_mask): New.
41422 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41423 __AVX512VPOPCNTDQ__.
41424 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41425 (PTA_AVX512VPOPCNTDQ): Define.
41426 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41427 TARGET_AVX512VPOPCNTDQ_P): Define.
41428 * config/i386/i386.opt: Add mavx512vpopcntdq.
41429 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41430 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41432 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41434 PR middle-end/77484
41435 * predict.def (PRED_CALL): Set to 67.
41437 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41439 * expr.c (store_field): In the bitfield case, if the value comes from
41440 a function call and is of an aggregate type returned in registers, do
41441 not modify the field mode; extract the value in all cases if the mode
41442 is BLKmode and the size is not larger than a word.
41444 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41447 * config/i386/cpuid.h: Fix undefined behavior.
41449 2017-01-04 Jeff Law <law@redhat.com>
41451 PR tree-optimization/79007
41452 PR tree-optimization/67955
41453 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41454 conservative for pt.null when flag_non_call_exceptions is on.
41456 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41458 PR translation/79019
41459 PR translation/79020
41460 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41461 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41462 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41464 * config/avr/avr.opt (maccumulate-args): Likewise.
41465 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41466 * common.opt (freport-bug): Likewise.
41467 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41468 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41469 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41470 translatable string.
41471 * config/i386/i386.c (function_value_32): Likewise.
41472 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41473 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41475 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41476 * common/config/msp430/msp430-common.c (msp430_handle_option):
41478 * symtab.c (symtab_node::verify_base): Likewise.
41479 * opts.c (set_debug_level): Likewise.
41480 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41481 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41482 missing whitespace to translatable strings.
41483 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41484 * config/sh/superh.h: Likewise.
41485 * config/i386/xopintrin.h: Likewise.
41486 * config/i386/znver1.md: Likewise.
41487 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41488 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41489 * double-int.h (struct double_int): Likewise.
41490 * double-int.c (div_and_round_double): Likewise.
41491 * wide-int.cc: Likewise.
41492 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41493 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41494 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41495 (crossjumps_occurred): ... this.
41496 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41499 PR tree-optimization/78899
41500 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41501 returning bool return struct loop *, NULL for failure and the new
41503 (versionable_outer_loop_p): Don't version outer loop if it has
41504 dont_vectorized bit set.
41505 (tree_if_conversion): When versioning outer loop, ensure
41506 tree_if_conversion is performed also on the inner loop of the
41507 non-vectorizable outer loop copy.
41508 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41509 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41510 prevent vectorization of it.
41511 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41512 the outer loop vectorization of the non-scalar version is attempted
41513 before vectorization of the inner loop in scalar version. If
41514 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41515 vectorization of its inner loop.
41516 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41517 has 2 inner loops, rename also on edges from bb whose single pred
41518 is outer_loop->header. Fix typo in function comment.
41520 2017-01-09 Martin Sebor <msebor@redhat.com>
41523 * asan.c (asan_emit_stack_protection): Increase local buffer size
41524 to avoid snprintf truncation warning.
41526 2017-01-09 Andrew Pinski <apinski@cavium.com>
41528 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41529 to reference thunderx2t99 for the tuning structure
41530 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41532 (thunderx2t99_extra_costs): This.
41533 * config/aarch64/aarch64-tune.md: Regenerate.
41534 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41535 (vulcan_addrcost_table): This.
41536 (vulcan_regmove_cost): Rename to ...
41537 (thunderx2t99_regmove_cost): This.
41538 (vulcan_vector_cost): Rename to ...
41539 (thunderx2t99_vector_cost): this.
41540 (vulcan_branch_cost): Rename to ...
41541 (thunderx2t99_branch_cost): This.
41542 (vulcan_tunings): Rename to ...
41543 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41544 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41546 2017-01-09 Martin Jambor <mjambor@suse.cz>
41550 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41551 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41552 (propagate_vr_accross_jump_function): Use the above function for all
41553 value range computations for pass-through jump functions and type
41554 converasion from explicit value range values.
41555 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41556 parameters from TYPE_ARG_TYPES.
41557 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41558 (ipa_write_node_info): Stream type of the actual argument.
41559 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41561 2017-01-09 Martin Liska <mliska@suse.cz>
41564 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41565 (lookup_compiler): Do not show error message with have_E.
41567 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41569 PR tree-optimization/78938
41570 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41571 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41572 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41573 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41576 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41578 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41581 2017-01-09 Richard Biener <rguenther@suse.de>
41583 PR tree-optimization/78997
41584 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41585 name condition properly.
41587 2017-01-09 Richard Biener <rguenther@suse.de>
41590 * dwarf2out.c (is_cxx): New overload with context.
41591 (is_naming_typedef_decl): Use it.
41593 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41595 * invoke.texi (Option Summary): Correct spacing in option lists
41596 and add line breaks to fix over-long lines.
41598 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41600 PR middle-end/17660
41602 * extend.texi (Common Variable Attributes): Add xref to GCC
41603 Internals manual to explain mode attribute keywords.
41605 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41608 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41609 and Preprocessor Options.
41610 (Options for Linking): Document -pthread here....
41611 (RS/6000 and PowerPC Options): ...not here.
41612 (Solaris 2 Options): ...or here.
41613 * doc/cppopts.texi: Document -pthread.
41615 2017-01-08 Martin Sebor <msebor@redhat.com>
41617 PR middle-end/77708
41618 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41619 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41620 New member functions.
41621 (format_directive): Used them.
41623 (pass_sprintf_length::handle_gimple_call): Same.
41624 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41625 to avoid truncation for any argument.
41626 (extract_affine_mul): Same.
41627 * tree.c (get_file_function_name): Same.
41629 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41631 PR middle-end/77484
41632 * predict.def (PRED_INDIR_CALL): Set to 86.
41634 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41636 PR preprocessor/54124
41637 * doc/cppopts.texi: Reformat -d subtable to list the full name
41638 of the options. Add cross-reference to the docs for the general
41639 compiler -d options.
41640 * doc/invoke.texi (Developer Options): Add cross-reference to the
41641 preprocessor-specific -d option documentation.
41643 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41645 PR preprocessor/13498
41646 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41647 redudant material, and reflect new command-line options.
41648 (System Headers): Likewise.
41650 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41652 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41653 -isystem, and -idirafter. Copy-edit.
41654 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41655 default for -ftrack-macro-expansion. Delete obsolete and
41656 badly-formatted implementation details about -fdebug-cpp output.
41657 * doc/cppwarnopts.texi: Copy-edit.
41659 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41662 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41663 that the transition from a max line width >= 1<<10 to narrower
41664 lines works correctly.
41666 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41668 * doc/options.texi (PerFunction): New.
41669 * opt-functions.awk (switch_flags): Map both Optimization and
41670 PerFunction to CL_OPTIMIZATION.
41671 * opth-gen.awk: Test for PerFunction flag along with
41673 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41674 it only when the latter is present. Skip those that don't in
41675 the hash function generator.
41676 * common.opt (fvar-tracking): Mark as PerFunction instead of
41678 (fvar-tracking-assignments): Likewise.
41679 (fvar-tracking-assignments-toggle): Likewise.
41680 (fvar-tracking-uninit): Likewise.
41682 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41684 PR translation/79018
41685 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41688 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41691 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41692 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41693 TARGET_LONG_JUMP_TABLE_OFFSETS.
41694 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41695 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41696 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41697 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41698 * config/m68k/m68k.md (tablejump expander): Likewise.
41699 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41700 TARGET_LONG_JUMP_TABLE_OFFSETS.
41701 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41702 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41704 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41705 David Holsgrove <david.holsgrove@xilinx.com>
41707 * common/config/microblaze/microblaze-common.c
41708 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41709 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41711 * config/microblaze/microblaze.c (microblaze_must_save_register)
41712 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41714 (microblaze_eh_return): New function.
41715 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41716 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41717 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41718 * config/microblaze/microblaze.md (eh_return): New pattern.
41720 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41722 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41723 GCC_DIAGNOSTIC_STRINGIFY): Define.
41725 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41727 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41729 * config/arm/arm.md (<mcrr>): New.
41731 * config/arm/arm.c (arm_arch5te): New.
41732 (arm_option_override): Set arm_arch5te.
41733 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41735 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41736 (arm_mcrr_qualifiers): ... this. New.
41737 (MRRC_QUALIFIERS): Define to...
41738 (arm_mrrc_qualifiers): ... this. New.
41739 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41741 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41742 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41743 (MRRCI, mrrc, MRRC): New.
41744 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41745 VUNSPEC_MRRC2): New.
41747 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41749 * config/arm/arm.md (<mcr>): New.
41751 * config/arm/arm.c (arm_coproc_builtin_available): Add
41752 support for mcr, mrc, mcr2 and mrc2.
41753 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41754 (arm_mcr_qualifiers): ... this. New.
41755 (MRC_QUALIFIERS): Define to ...
41756 (arm_mrc_qualifiers): ... this. New.
41757 (MCR_QUALIFIERS): Define to ...
41758 (arm_mcr_qualifiers): ... this. New.
41759 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41761 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41762 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41763 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41764 VUNSPEC_MRC2): New.
41766 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41768 * config/arm/arm.md (*ldc): New.
41772 * config/arm/arm.c (arm_coproc_builtin_available): Add
41773 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41774 (arm_coproc_ldc_stc_legitimate_address): New.
41775 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41776 'qualifier_const_pointer'.
41777 (LDC_QUALIFIERS): Define to...
41778 (arm_ldc_qualifiers): ... this. New.
41779 (STC_QUALIFIERS): Define to...
41780 (arm_stc_qualifiers): ... this. New.
41781 * config/arm/arm-protos.h
41782 (arm_coproc_ldc_stc_legitimate_address): New.
41783 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41784 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41785 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41786 stc2, stcl, stc2l): New.
41787 * config/arm/constraints.md (Uz): New.
41788 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41789 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41790 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41791 VUNSPEC_STC2L): New.
41793 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41795 * config/arm/arm.md (<cdp>): New.
41796 * config/arm/arm.c (neon_const_bounds): Rename this ...
41797 (arm_const_bounds): ... this.
41798 (arm_coproc_builtin_available): New.
41799 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41800 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41801 (CDP_QUALIFIERS): Define to...
41802 (arm_cdp_qualifiers): ... this. New.
41804 (arm_expand_builtin_args): Add case for 6 arguments.
41805 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41806 (arm_const_bounds): ... this.
41807 (arm_coproc_builtin_available): New.
41808 * config/arm/arm_acle.h (__arm_cdp): New.
41810 * config/arm/arm_acle_builtins.def (cdp): New.
41812 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41813 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41814 'arm_const_bounds'.
41815 * config/arm/types.md (coproc): New.
41816 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41817 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41818 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41819 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41821 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41823 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41824 (UBINOP_QUALIFIERS): New.
41826 (acle_builtin_data): New. Change comment.
41827 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41828 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41829 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41830 arm_acle_builtins.def.
41831 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41832 (arm_init_acle_builtins): New.
41833 (CRC32_BUILTIN): Remove.
41834 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41835 crc32cb, crc32ch and crc32cw.
41836 (arm_init_crc32_builtins): Remove.
41837 (arm_init_builtins): Use arm_init_acle_builtins rather
41838 than arm_init_crc32_builtins.
41839 (arm_expand_acle_builtin): New.
41840 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41841 * config/arm/arm_acle_builtins.def: New.
41843 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41845 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41846 (arm_builtin_datum): ... this.
41847 (arm_init_neon_builtin): Rename to ...
41848 (arm_init_builtin): ... this. Add a new parameters PREFIX
41849 and USE_SIG_IN_NAME.
41850 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41851 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41852 'arm_builtin_datum'.
41853 (arm_init_vfp_builtins): Likewise.
41854 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41855 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41856 (arm_expand_neon_args): Rename to ...
41857 (arm_expand_builtin_args): ... this. Rename builtin_arg
41858 enum values and differentiate between ARG_BUILTIN_MEMORY
41859 and ARG_BUILTIN_NEON_MEMORY.
41860 (arm_expand_neon_builtin_1): Rename to ...
41861 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41862 values, arm_expand_builtin_args and add bool parameter NEON.
41863 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41864 (arm_expand_vfp_builtin): Likewise.
41865 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41867 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41869 PR middle-end/77484
41870 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41871 * predict.c (tree_estimate_probability_bb): Reverse direction of
41872 polymorphic call predictor.
41874 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41876 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41877 (determine_pass_name_match): ...this new function and...
41878 (should_skip_pass_p): ...this new function.
41880 2017-01-06 Nathan Sidwell <nathan@acm.org>
41882 * ipa-visibility.c (function_and_variable_visibility): Reformat
41883 comments and long lines. Remove extrneous if.
41884 * symtab.c (symtab_node::make_decl_local): Fix code format.
41885 (symtab_node::set_section_for_node): Fix comment typo.
41887 2017-01-06 Martin Liska <mliska@suse.cz>
41890 * lra-constraints.c: Rename invariant to lra_invariant.
41891 * predict.c (set_even_probabilities): Initialize e to NULL.
41893 2017-01-05 Martin Sebor <msebor@redhat.com>
41895 PR tree-optimization/78910
41896 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41897 (format_integer): Correct off-by-one error in the handling
41898 of precision with negative numbers in signed conversions..
41900 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41902 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41904 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41906 PR tree-optimization/71016
41907 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41908 factor_out_conditional_conversion. Formatting fix.
41909 (factor_out_conditional_conversion): Add cond_stmt argument.
41910 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41911 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41914 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41916 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41917 read-rtl-function.o, and selftest-rtl.o.
41918 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41919 (selftest::aarch64_test_loading_full_dump): New function.
41920 (selftest::aarch64_run_selftests): New function.
41921 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41922 selftest::aarch64_run_selftests.
41923 * config/i386/i386.c
41924 (selftest::ix86_test_loading_dump_fragment_1): New function.
41925 (selftest::ix86_test_loading_call_insn): New function.
41926 (selftest::ix86_test_loading_full_dump): New function.
41927 (selftest::ix86_test_loading_unspec): New function.
41928 (selftest::ix86_run_selftests): Call the new functions.
41929 * emit-rtl.c (maybe_set_max_label_num): New function.
41930 * emit-rtl.h (maybe_set_max_label_num): New decl.
41931 * function.c (instantiate_decls): Guard call to
41932 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41933 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41935 * gensupport.c (gen_reader::gen_reader): Pass "false"
41936 for new "compact" param of rtx_reader.
41937 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41938 rather than an empty string for NULL strings.
41939 * read-md.c: Potentially include config.h rather than bconfig.h.
41940 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41941 (have_error): New global, copied from errors.c.
41942 (md_reader::read_name): Rename to...
41943 (md_reader::read_name_1): ...this, adding "out_loc" param,
41944 and converting "missing name or number" to returning false, rather
41946 (md_reader::read_name): Reimplement in terms of read_name_1.
41947 (md_reader::read_name_or_nil): New function.
41948 (md_reader::read_string): Handle "(nil)" by returning NULL.
41949 (md_reader::md_reader): Add new param "compact".
41950 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41951 (md_reader::read_file): New method.
41952 * read-md.h (md_reader::md_reader): Add new param "compact".
41953 (md_reader::read_file): New method.
41954 (md_reader::is_compact): New accessor.
41955 (md_reader::read_name): Convert return type from void to file_location.
41956 (md_reader::read_name_or_nil): New decl.
41957 (md_reader::read_name_1): New decl.
41958 (md_reader::m_compact): New field.
41959 (noop_reader::noop_reader): Pass "false" for new "compact" param
41961 (rtx_reader::rtx_reader): Add new "compact" param.
41962 (rtx_reader::read_rtx_operand): Make virtual and convert return
41963 type from void to rtx.
41964 (rtx_reader::read_until): New decl.
41965 (rtx_reader::handle_any_trailing_information): New virtual function.
41966 (rtx_reader::postprocess): New virtual function.
41967 (rtx_reader::finalize_string): New virtual function.
41968 (rtx_reader::m_in_call_function_usage): New field.
41969 (rtx_reader::m_reuse_rtx_by_id): New field.
41970 * read-rtl-function.c: New file.
41971 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41972 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41973 (selftest::verify_three_block_rtl_cfg): New decl.
41974 * read-rtl-function.h: New file.
41975 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41976 For host, include function.h, memmodel.h, and emit-rtl.h.
41977 (one_time_initialization): New function.
41978 (struct compact_insn_name): New struct.
41979 (compact_insn_names): New array.
41980 (find_code): Handle insn codes in compact dumps.
41981 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41982 (bind_subst_iter_and_attr): Likewise.
41983 (add_condition_to_string): Likewise.
41984 (add_condition_to_rtx): Likewise.
41985 (apply_attribute_uses): Likewise.
41986 (add_current_iterators): Likewise.
41987 (apply_iterators): Likewise.
41988 (initialize_iterators): Guard usage of apply_subst_iterator with
41989 #ifdef GENERATOR_FILE.
41990 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41991 (md_reader::read_mapping): Likewise.
41992 (add_define_attr_for_define_subst): Likewise.
41993 (add_define_subst_attr): Likewise.
41994 (read_subst_mapping): Likewise.
41995 (check_code_iterator): Likewise.
41996 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41998 (one_time_initialization): New function.
41999 (rtx_reader::read_until): New method.
42000 (read_flags): New function.
42001 (parse_reg_note_name): New function.
42002 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42003 Handle reuse_rtx ids.
42004 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42005 Add parsing support for RTL dumps, mirroring the special-cases in
42006 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42007 values, and calling handle_any_trailing_information.
42008 (rtx_reader::read_rtx_operand): Convert return type from void
42009 to rtx, returning return_rtx. Handle case 'e'. Call
42010 finalize_string on XSTR and XTMPL fields.
42011 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42012 "(nil)" values were omitted. Call the postprocess vfunc on the
42014 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42015 class ctor. Initialize m_in_call_function_usage. Call
42016 one_time_initialization.
42017 * rtl-tests.c (selftest::test_uncond_jump): Call
42018 set_new_first_and_last_insn.
42019 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42020 * selftest-rtl.c: New file.
42021 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42022 (selftest::get_insn_by_uid): New decl.
42023 * selftest-run-tests.c (selftest::run_tests): Call
42024 read_rtl_function_c_tests.
42025 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42026 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42029 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42031 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42032 operands in a special way. Assert that pos+len <= mode precision.
42034 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42036 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42037 3 argument Alias with unlimited for the negative form.
42038 (fno-vect-cost-model): Removed.
42040 2017-01-05 Martin Liska <mliska@suse.cz>
42042 * hsa-gen.c (gen_hsa_divmod): New function.
42043 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42045 2017-01-05 Martin Liska <mliska@suse.cz>
42048 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42051 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42053 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42054 small constant length operands.
42056 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42058 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42059 between loop iterations.
42061 2017-01-05 Martin Liska <mliska@suse.cz>
42064 * gimplify.c (gimplify_decl_expr): Compare to
42065 asan_poisoned_variables instread of checking flags.
42066 (gimplify_target_expr): Likewise.
42067 (gimplify_expr): Likewise.
42068 (gimplify_function_tree): Conditionally initialize
42069 asan_poisoned_variables.
42071 2017-01-04 Jeff Law <law@redhat.com>
42073 PR tree-optimizatin/78812
42074 * rtl.h (contains_mem_rtx_p): Prototype.
42075 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42076 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42077 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42078 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42079 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42081 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42083 * input.c (assert_char_at_range): Default-initialize actual_range.
42085 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42087 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42088 to match the caller.
42090 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42092 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42093 insns after final jump in test to emit dummy move.
42095 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42097 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42098 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42100 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42102 * multiple_target.c (create_dispatcher_calls): Init e_next.
42103 * tree-ssa-loop-split.c (split_loop): Init border.
42104 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42107 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42112 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42113 (altivec_register_operand): Do not return true if the operand
42114 contains a SUBREG mixing SImode and SFmode.
42115 (vsx_register_operand): Likewise.
42116 (vsx_reg_sfsubreg_ok): New predicate.
42117 (vfloat_operand): Do not return true if the operand contains a
42118 SUBREG mixing SImode and SFmode.
42119 (vint_operand): Likewise.
42120 (vlogical_operand): Likewise.
42121 (gpc_reg_operand): Likewise.
42122 (int_reg_operand): Likewise.
42123 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42124 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42125 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42127 (rs6000_emit_move_si_sf_subreg): New helper function.
42128 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42129 fixup SUBREGs involving SImode and SFmode.
42130 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42131 numbers for the new peephole2 optimization.
42132 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42133 the GLIBC math library that do AND/IOR/XOR operations on single
42134 precision floating point.
42135 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42136 target macros to say whether we need to avoid SUBREGs mixing
42138 (TARGET_ALLOW_SF_SUBREG): Likewise.
42139 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42140 (UNSPEC_SI_FROM_SF): Likewise.
42141 (iorxor): Change spacing.
42142 (and_ior_xor): New iterator for AND, IOR, and XOR.
42143 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42144 (movdi_from_sf_zero_ext): Likewise.
42145 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42146 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42147 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42148 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42149 (fms<mode>4): Likewise.
42150 (fnma<mode>4): Likewise.
42151 (fnms<mode>4): Likewise.
42152 (nfma<mode>4): Likewise.
42153 (nfms<mode>4): Likewise.
42155 2017-01-04 Marek Polacek <polacek@redhat.com>
42158 * doc/invoke.texi: Document -Wpointer-compare.
42160 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42162 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42165 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42166 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42169 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42172 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42173 documentation of the powerpc_popcntb_ok attribute.
42174 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42175 code to issue warning messages if a requested CPU configuration is
42176 not supported by the binary (assembler and loader) toolchain.
42177 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42178 made to define a built-in function that has been disabled.
42179 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42180 made to define a built-in function that has been disabled.
42181 (altivec_init_builtins): Add comment explaining why definition
42182 of the DST built-in functions is not preceded by an assertion
42183 check. Add assertions to prevent ICE if attempts are made to
42184 define an altivec predicate or an abs* built-in function that has
42186 (htm_init_builtins): Add comment explaining why definition of the
42187 htm built-in functions is not preceded by an assertion check.
42189 2017-01-04 Jeff Law <law@redhat.com>
42191 PR tree-optimizatin/67955
42192 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42193 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42194 the points-to solution does not include pt_null. Use DECL_PT_UID
42197 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42199 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42200 Use gen_int_mode instead of gen_lopwart for const_int operands.
42202 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42204 PR tree-optimization/71563
42205 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42206 out of range value - has low bits known to be zero.
42208 2017-01-04 Alan Modra <amodra@gmail.com>
42210 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42211 * configure: Regenerate.
42212 * config.in: Regenerate.
42214 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42217 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42218 a substring of the message, but strcmp with the whole message. Ifdef
42219 ENABLE_NLS, translate the message first using dgettext.
42221 2017-01-03 Jeff Law <law@redhat.com>
42223 PR tree-optimizatin/78856
42224 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42225 (mark_threaded_blocks): Remove code to truncate thread paths that
42226 cross multiple loop headers. Instead invalidate the cached loop
42227 iteration information and handle case of a thread path walking
42228 into an irreducible region.
42230 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42233 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42234 assertions. Add support for doing the signbit if the IEEE 128-bit
42235 floating point value is in a GPR.
42236 * config/rs6000/rs6000.md (Fsignbit): Delete.
42237 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42238 Update the length attribute if the value is in a GPR.
42239 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42240 the sign or zero extension instruction, since the value is always 0/1.
42241 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42244 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42245 extracting SImode to a GPR register so that we can generate a
42246 store, limit the vector to be in a traditional Altivec register
42247 for the vextuwrx instruction.
42249 2017-01-03 Ian Lance Taylor <iant@google.com>
42251 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42253 2017-01-03 Martin Sebor <msebor@redhat.com>
42255 PR tree-optimization/78696
42256 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42257 precision. Use MPFR for %f for greater fidelity. Correct handling
42259 (pass_sprintf_length::compute_format_length): Set width and precision
42260 specified by asrerisk to void_node for vararg functions.
42261 (try_substitute_return_value): Adjust dump output.
42263 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42265 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42267 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42269 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42270 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42271 -mlra/-mno-lra was passed to the compiler.
42273 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42275 PR rtl-optimization/65618
42276 * emit-rtl.c (try_split): Move initialization of "before" and
42277 "after" to just before the call to emit_insn_after_setloc.
42279 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42281 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42283 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42285 * dwarf2out.c (gen_enumeration_type_die): When
42286 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42288 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42290 PR tree-optimization/78965
42291 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42292 Change first argument from const call_info & to call_info &. For %n
42293 set info.nowrite to false.
42295 PR middle-end/78901
42296 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42297 possibly throwing calls.
42299 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42300 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42301 and fns handling, rather than in a separate case SSA_NAME.
42303 2017-01-02 Jeff Law <law@redhat.com>
42305 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42306 fixes for first_period and second_period variables.
42308 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42311 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42312 (*insvqi_1): New insn pattern.
42313 (*insvqi_1_mem_rex64): Ditto.
42314 (*insvqi_2): Ditto.
42315 (*insvqi_3): Rename from *insvqi.
42317 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42319 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42321 * doc/cfg.texi (Edges): Remove reference to Java.
42322 (Maintaining the CFG): Ditto.
42324 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42326 PR middle-end/77674
42327 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42328 transparent aliases.
42330 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42332 PR middle-end/77484
42333 * predict.def (PRED_CALL): Update hitrate.
42334 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42335 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42336 into direct/indirect/polymorphic variants.
42338 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42340 Update copyright years.
42342 * gcc.c (process_command): Update copyright notice dates.
42343 * gcov-dump.c (print_version): Ditto.
42344 * gcov.c (print_version): Ditto.
42345 * gcov-tool.c (print_version): Ditto.
42346 * gengtype.c (create_file): Ditto.
42347 * doc/cpp.texi: Bump @copying's copyright year.
42348 * doc/cppinternals.texi: Ditto.
42349 * doc/gcc.texi: Ditto.
42350 * doc/gccint.texi: Ditto.
42351 * doc/gcov.texi: Ditto.
42352 * doc/install.texi: Ditto.
42353 * doc/invoke.texi: Ditto.
42355 Copyright (C) 2017 Free Software Foundation, Inc.
42357 Copying and distribution of this file, with or without modification,
42358 are permitted in any medium without royalty provided the copyright
42359 notice and this notice are preserved.