1 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3 * ipa-split.c (struct split_point): Add count.
4 (consider_split): Do not compute incoming frequency; compute incoming
5 count and store it to split_point.
6 (split_function): Set count of the call to split part correctly.
8 2017-11-13 Carl Love <cel@us.ibm.com>
10 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
12 2017-11-13 Tom Tromey <tom@tromey.com>
14 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
16 2017-11-13 Carl Love <cel@us.ibm.com>
18 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
19 Add support for builtins:
20 unsigned int vec_first_{,miss}_match_{,or_eos}index,
21 vector {un,}signed {char,int,short},
22 vector {un,}signed {char,int,short}) arguments.
23 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
24 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
25 Add BU_P9V_AV_2 expansions for the builtins.
26 * config/rs6000/altivec.h (vec_first_match_index,
27 vec_first_mismatch_index, vec_first_match_or_eos_index,
28 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
29 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
30 new extern declaration.
31 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
32 * config/rs6000/vsx.md (first_match_index_<mode>,
33 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
34 first_mismatch_or_eos_index_<mode>): Add define expand.
35 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
36 * doc/extend.texi: Update the built-in documenation file for the new
39 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
41 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
42 into the min/max operations for _Float<N> and _Float<N>X types.
44 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
47 * dwarf2out.c (do_eh_frame): New static variable.
48 (dwarf2out_begin_prologue): Set it.
49 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
51 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
53 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
55 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
57 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
60 * debug.h (dwarf2out_do_eh_frame): Declare.
61 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
62 (dwarf2out_do_frame): Use it.
63 (dwarf2out_do_cfi_asm): Likewise.
64 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
65 (dwarf2out_assembly_start): Likewise.
66 (dwarf2out_begin_prologue): Fix comment.
67 * toplev.c (compile_file): Always call dwarf2out_frame_finish
68 if the target needs either debug or unwind DWARF2 info.
69 * lto-opts.c (lto_write_options): Do not save -fexceptions,
70 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
71 -ftrapping-math, -ftrapv and -fwrapv.
73 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
75 * cgraph.c (cgraph_edge::sreal_frequency): New function.
76 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
77 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
78 (estimate_edge_size_and_time): Likewise.
79 (ipa_merge_fn_summary_after_inlining): Likewise.
80 * ipa-inline.c (cgraph_freq_base_rec): Remove.
81 (compute_uninlined_call_time): Use sreal_frequency.
82 (compute_inlined_call_time): Likewise.
83 (ipa_inline): Do not initialize cgraph_freq_base_rec.
84 * profile-count.c: Include sreal.h.
85 (profile_count::to_sreal_scale): New.
86 * profile-count.h: Forward declare sreal.
87 (profile_count::to_sreal_scale): Declare.
89 2017-11-13 Nathan Sidwell <nathan@acm.org>
91 * diagnostic.c (maybe_line_and_column): New.
92 (diagnostic_get_location_text): Use it.
93 (diagnostic_report_current_module): Likewise.
94 (test_diagnostic_get_location_text): Add tests.
96 2017-11-13 Luis Machado <luis.machado@linaro.org>
98 * doc/md.texi (Specifying processor pipeline description): Fix
99 incorrect latency for the div instruction example.
101 2017-11-13 Jakub Jelinek <jakub@redhat.com>
103 PR tree-optimization/78821
104 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
105 that bit_not_p is the same.
106 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
107 (split_group): Count precisely bit_not_p bits in each statement.
108 (invert_op): New function.
109 (imm_store_chain_info::output_merged_store): Use invert_op to
110 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
111 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
113 2017-11-13 Martin Liska <mliska@suse.cz>
115 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
116 (struct source_info): Likewise.
117 (add_branch_counts): Likewise.
118 (add_line_counts): Likewise.
119 (function_summary): Likewise.
120 (output_intermediate_line): Likewise.
121 (generate_results): Likewise.
123 2017-11-13 Martin Liska <mliska@suse.cz>
125 * gcov.c (struct block_info): Remove typedef for block_t.
126 (struct line_info): Likewise.
127 (line_info::has_block): Likewise.
128 (EXIT_BLOCK): Likewise.
131 (get_cycles_count): Likewise.
132 (process_file): Likewise.
133 (read_graph_file): Likewise.
134 (solve_flow_graph): Likewise.
135 (find_exception_blocks): Likewise.
136 (add_line_counts): Likewise.
137 (accumulate_line_info): Likewise.
138 (output_line_details): Likewise.
140 2017-11-13 Martin Liska <mliska@suse.cz>
142 * gcov.c (struct arc_info): Remove typedef for arc_t.
143 (struct line_info): Likewise.
144 (add_branch_counts): Likewise.
145 (output_branch_count): Likewise.
146 (function_info::~function_info): Likewise.
148 (output_intermediate_line): Likewise.
149 (read_graph_file): Likewise.
150 (solve_flow_graph): Likewise.
151 (find_exception_blocks): Likewise.
152 (add_line_counts): Likewise.
153 (accumulate_line_info): Likewise.
154 (output_line_details): Likewise.
155 (output_function_details): Likewise.
157 2017-11-13 Martin Liska <mliska@suse.cz>
159 * gcov.c (struct function_info): Remove typedef for function_t.
160 (struct source_info): Likewise.
161 (source_info::get_functions_at_location): Likewise.
162 (solve_flow_graph): Likewise.
163 (find_exception_blocks): Likewise.
164 (add_line_counts): Likewise.
165 (output_intermediate_file): Likewise.
166 (process_file): Likewise.
167 (generate_results): Likewise.
168 (release_structures): Likewise.
169 (read_graph_file): Likewise.
170 (read_count_file): Likewise.
171 (accumulate_line_counts): Likewise.
172 (output_lines): Likewise.
174 2017-11-13 Martin Liska <mliska@suse.cz>
176 * gcov.c (function_info::function_info): Remove num_counts
177 and add vector<gcov_type>.
178 (function_info::~function_info): Use the vector.
179 (process_file): Likewise.
180 (read_graph_file): Likewise.
181 (read_count_file): Likewise.
182 (solve_flow_graph): Likewise.
184 2017-11-13 Martin Liska <mliska@suse.cz>
186 * gcov.c (function_info::is_artificial): New function.
187 (process_file): Erase all artificial early.
188 (generate_results): Skip as all artificial are already
191 2017-11-13 Martin Liska <mliska@suse.cz>
193 * gcov.c (read_graph_file): Store to global vector of functions.
194 (read_count_file): Iterate the vector.
195 (process_file): Likewise.
196 (generate_results): Likewise.
197 (release_structures): Likewise.
199 2017-11-13 Jakub Jelinek <jakub@redhat.com>
201 PR tree-optimization/82954
202 * gimple-ssa-store-merging.c
203 (imm_store_chain_info::coalesce_immediate_stores): If
204 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
206 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
208 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
209 Upddate call to ENDIAN_LANE_N.
210 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
211 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
212 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
213 and use aarch64_endian_lane_rtx.
214 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
216 2017-11-12 Tom de Vries <tom@codesourcery.com>
218 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
220 2017-11-12 Tom de Vries <tom@codesourcery.com>
222 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
224 2017-11-12 Tom de Vries <tom@codesourcery.com>
226 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
228 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
229 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
230 * defaults.h (ASM_OUTPUT_LABELREF): Same.
232 2017-11-11 Martin Sebor <msebor@redhat.com>
235 * doc/extend.texi (attribute nonstring): Remove spurious argument.
238 * prefic.c (translate_name): Replace strncpy with memcpy to
239 avoid -Wstringop-truncation.
241 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
243 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
245 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
247 * predict.c (maybe_hot_frequency_p): Do not use cfun.
249 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
251 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
254 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
256 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
257 enable generating XXBRH if the value is in a vector register.
258 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
259 value is in a vector register.
260 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
261 register to register bswap64's instead of doing the GPR sequence
262 used on previous machines.
263 (bswapdi2_xxbrd): New insn.
264 (bswapdi2_reg): Disallow on ISA 3.0.
265 (register to register bswap64 splitter): Do not split the insn on
266 ISA 3.0 systems that use XXBRD.
268 2017-11-10 Martin Sebor <msebor@redhat.com>
271 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
273 (find_subframework_file): Same.
275 2017-11-10 Jeff Law <law@redhat.com>
277 * vr-values.h (VR_INITIALIZER): Move #define here.
278 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
279 * Makefile.in (OBJS): Add tree-evrp.o
280 * tree-vrp.h (assert_info): Move structure definition here.
281 (set_value_range_to_varying): Prototype.
282 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
283 (infer_value_range, register_edge_assert_for): Likewise.
284 (stmt_interesting_for_vrp): Likewise.
285 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
286 (set_value_range_to_varying): No longer static.
287 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
288 (infer_value_range, register_edge_assert_for): Likewise.
290 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
292 * auto-profile.c (afdo_indirect_call): Drop frequency.
293 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
294 (cgraph_node::create_edge): Drop frequency argument.
295 (cgraph_node::create_indirect_edge): Drop frequency argument.
296 (cgraph_edge::make_speculative): Drop frequency arguments.
297 (cgraph_edge::resolve_speculation): Do not update frequencies
298 (cgraph_edge::dump_edge_flags): Do not dump frequency.
299 (cgraph_node::dump): Check consistency in IPA mode.
300 (cgraph_edge::maybe_hot_p): Use IPA counter.
301 (cgraph_edge::verify_count_and_frequency): Rename to ...
302 (cgraph_edge::verify_count): ... this one; drop frequency checking.
303 (cgraph_node::verify_node): Update.
304 * cgraph.h (struct cgraph_edge): Drop frequency.
305 (cgraph_edge::frequency): New function.
306 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
308 (cgraph_edge::rebuild_edges): Likewise.
309 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
310 (duplicate_thunk_for_node): Do not pass frequency.
311 (cgraph_node::create_clone): Scale only counts.
312 (cgraph_node::create_virtual_clone): Do not pass frequency.
313 (cgraph_node::create_edge_including_clones): Do not pass frequency.
314 (cgraph_node::create_version_clone): Do not pass frequency.
315 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
316 (cgraph_node::expand_thunk): Do not pass frequency.
317 (cgraph_node::create_wrapper): Do not pass frequency.
318 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
320 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
321 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
322 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
323 (gather_caller_stats): Use frequency function.
324 (ipcp_cloning_candidate_p): Use frequency function.
325 (ipcp_propagate_stage): Use frequency function.
326 (get_info_about_necessary_edges): Use frequency function.
327 (update_profiling_info): Update only IPA profile.
328 (update_specialized_profile): Use frequency functoin.
329 (perhaps_add_new_callers): Update only IPA profile.
330 * ipa-devirt.c (ipa_devirt): Use IPA profile.
331 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
332 (dump_ipa_call_summary): Use frequency function.
333 (estimate_edge_size_and_time): Use frequency function.
334 (ipa_merge_fn_summary_after_inlining): Use frequency function.
335 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
336 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
337 (update_noncloned_counts): ... ths one; scale counts only.
338 (clone_inlined_nodes): Do not scale frequency.
339 (inline_call): Do not pass frequency.
340 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
341 (compute_inlined_call_time): Use IPA profile.
342 (want_inline_small_function_p): Use IPA profile.
343 (want_inline_self_recursive_call_p): Use IPA profile.
344 (edge_badness): Use IPA profile.
345 (lookup_recursive_calls): Use IPA profile.
346 (recursive_inlining): Do not pass frequency.
347 (resolve_noninline_speculation): Do not update frequency.
348 (inline_small_functions): Collect max of IPA profile.
349 (dump_overall_stats): Dump IPA porfile.
350 (dump_inline_stats): Dump IPA porfile.
351 (ipa_inline): Collect IPA stats.
352 * ipa-inline.h (clone_inlined_nodes): Update prototype.
353 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
354 (ipa_propagate_frequency): Use frequency function.
355 (ipa_profile): Cleanup.
356 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
357 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
358 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
359 (input_node): Do not stream frequency.
360 (input_edge): Do not stream frequency.
361 (merge_profile_summaries): Scale only IPA profiles.
362 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
363 * predict.c (drop_profile): Do not recompute frequency.
364 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
365 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
366 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
367 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
368 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
369 * tree-inline.c (copy_bb): Do not scale frequency.
370 (expand_call_inline): Do not scale frequency.
371 (tree_function_versioning): Do not scale frequency.
372 * ubsan.c (ubsan_create_edge): Do not pass frequency.
374 2017-11-10 Julia Koval <julia.koval@intel.com>
376 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
377 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
378 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
379 _mm256_maskz_gf2p8affine_epi64_epi8,
380 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
381 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
382 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
383 __builtin_ia32_vgf2p8affineqb_v32qi,
384 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
385 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
387 2017-11-10 Tamar Christina <tamar.christina@arm.com>
391 (arm_option_override): Refactor.
392 (arm_option_reconfigure_globals): New.
393 (arm_options_perform_arch_sanity_checks): New.
394 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
395 (arm_options_perform_arch_sanity_checks): Likewise
397 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
399 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
401 * rs6000/rs6000.md ("type" attr): Add qmul.
402 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
403 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
404 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
406 2017-11-10 Martin Sebor <msebor@redhat.com>
409 * builtins.c (compute_objsize): Handle arrays that
410 compute_builtin_object_size likes to fail for. Make extern.
411 * builtins.h (compute_objsize): Declare.
412 (check_strncpy_sizes): New function.
413 (expand_builtin_strncpy): Call check_strncpy_sizes.
414 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
415 -Wstringop-truncation.
416 (gimple_fold_builtin_strncat): Same.
417 * gimple.c (gimple_build_call_from_tree): Set call location.
418 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
419 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
420 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
421 (handle_builtin_strlen): Use strlen_to_stridx.
422 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
424 Use strlen_to_stridx.
425 (pass_strlen::execute): Release strlen_to_stridx.
426 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
427 (-Wstringop-truncation): Document new option.
429 2017-11-10 Martin Liska <mliska@suse.cz>
431 PR gcov-profile/82702
432 * gcov.c (main): Handle intermediate files in a different
434 (get_gcov_intermediate_filename): New function.
435 (output_gcov_file): Remove support of intermediate files.
436 (generate_results): Allocate intermediate file.
437 (release_structures): Clean-up properly fn_end.
438 (output_intermediate_file): Start iterating with line 1.
440 2017-11-10 Jakub Jelinek <jakub@redhat.com>
442 PR tree-optimization/82929
443 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
444 ops_swapped_p non-static data member.
445 (store_immediate_info::store_immediate_info): Clear it.
446 (imm_store_chain_info::coalesce_immediate_stores): If swapping
447 ops set ops_swapped_p.
448 (count_multiple_uses): Handle ops_swapped_p.
450 2017-11-10 Martin Liska <mliska@suse.cz>
452 * coverage.c (coverage_init): Stream information about
453 support of has_unexecuted_blocks.
454 * doc/gcov.texi: Document that.
455 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
456 * gcov.c (read_graph_file): Likewise.
457 (output_line_beginning): Fix a small issue with
460 2017-11-10 Bin Cheng <bin.cheng@arm.com>
462 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
463 reference of trivial component.
465 2017-11-10 Jakub Jelinek <jakub@redhat.com>
468 * gimple-ssa-store-merging.c
469 (pass_store_merging::terminate_all_aliasing_chains): For
470 gimple_store_p stmts also call refs_output_dependent_p.
472 PR rtl-optimization/82913
473 * compare-elim.c (try_merge_compare): Punt if def_insn is not
476 2017-11-09 Jeff Law <law@redhat.com>
478 * vr-values.h: New file with vr_values class.
479 * tree-vrp.c: Include vr-values.h
480 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
481 data objects into the vr_values class.
482 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
483 (get_value_range): Make it a member function within vr_values class.
484 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
485 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
486 (op_with_constant_singleton_value_range): Likewise.
487 (extract_range_for_var_from_comparison_expr): Likewise.
488 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
489 (extract_range_from_binary_expr): Likewise.
490 (extract_range_from_unary_expr): Likewise.
491 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
492 (check_for_binary_op_overflow, extract_range_basic): Likewise.
493 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
494 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
495 (compare_name_with_value, compare_names): Likewise.
496 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
497 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
498 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
499 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
500 (extract_range_from_phi_node): Likewise.
501 (simplify_truth_ops_using_ranges): Likewise.
502 (simplify_div_or_mod_using_ranges): Likewise.
503 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
504 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
505 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
506 (simplify_float_conversion_using_ranges): Likewise.
507 (simplify_internal_call_using_ranges): Likewise.
508 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
509 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
510 poitner via x_vr_values for calls into gimple folder.
511 (vrp_initialize_lattice): Make this the vr_values ctor.
512 (vrp_free_lattice): Make this the vr_values dtor.
513 (set_vr_value): New function.
514 (class vrp_prop): Add vr_values data member. Add various member
515 functions as well as member functions that delegate to vr_values.
516 (check_array_ref): Make a member function within vrp_prop class.
517 (search_for_addr_array, vrp_initialize): Likewise.
518 (vrp_finalize): Likewise. Revamp to avoid direct access to
519 vr_value, values_propagated, etc.
520 (check_array_bounds): Extract vrp_prop class instance pointer from
521 walk info structure. Use it to call member functions.
522 (check_all_array_refs): Make a member function within vrp_prop class.
523 Smuggle class instance pointer via walk info structure.
524 (x_vr_values): New local static.
525 (vrp_valueize): Use x_vr_values to get class instance.
526 (vr_valueize_1): Likewise.
527 (class vrp_folder): Add vr_values data member. Add various member
528 functions as well as member functions that delegate to vr_values.
529 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
530 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
531 class instance from vr_values. Use it to call member functions.
532 (vrp_dom_walker): Add vr_values data member.
533 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
534 instance via x_vr_values.
535 (identify_jump_threads): Accept vr_values as argument. Store
536 it into the walker structure.
537 (evrp_dom_walker): Add vr_values class data member. Add various
539 (evrp_dom_walker::try_find_new_range): Use vr_values data
540 member to access the memory allocator.
541 (evrp_dom_walker::before_dom_children): Store vr_values class
542 instance into the vrp_folder class.
543 (evrp_dom_walker::push_value_range): Rework to avoid direct
544 access to num_vr_values and vr_value.
545 (evrp_dom_walker::pop_value_range): Likewise.
546 (execute_early_vrp): Remove call to vrp_initialize_lattice.
547 Use vr_values to get to dump_all_value_ranges member function.
548 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
549 and simplify_cond_using_ranges_2 via vrp_prop class instance.
550 Pass vr_values class instance down to identify_jump_threads.
551 Remove call to vrp_free_lattice.
552 (debug_all_value_ranges): Remove.
554 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
555 (vrp_folder): Likewise.
557 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
558 Get it from the existing bitmap instead.
559 (vrp_intersect_ranges_1): Likewise.
561 2017-11-09 Jakub Jelinek <jakub@redhat.com>
563 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
565 (store_immediate_info::store_immediate_info): Add bitnotp argument,
567 (imm_store_chain_info::coalesce_immediate_stores): Break group
568 if bit_not_p is different.
569 (count_multiple_uses, split_group,
570 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
571 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
572 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
573 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
574 to store_immediate_info ctor.
576 2017-11-09 Jim Wilson <jimw@sifive.com>
578 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
579 (scan_prog_file): Likewise.
581 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
583 * bb-reorder.c (max_entry_frequency): Remove.
584 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
585 connect_traces, push_to_next_round_p): Remove prototypes.
586 (find_traces_1_round): Use counts only.
587 (push_to_next_round_p): Likewise.
588 (find_traces): Likewise.
589 (rotate_loop): Likewise.
590 (find_traces_1_round): Likewise.
591 (connect_traces): Likewise.
592 (edge_order): Likewise.
594 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
596 * config/arm/arm.c (output_return_instruction): Add comments to
597 indicate requirement for cmse_nonsecure_entry return to account
598 for the size of clearing instruction output here.
599 (thumb_exit): Likewise.
600 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
601 return in hardfloat mode.
603 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
605 * config/rs6000/rs6000.c (machine_function): Add a bool,
606 "toc_is_wrapped_separately".
607 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
608 if it wasn't explicitly set or unset, we are optimizing for speed, and
609 doing separate shrink-wrapping.
610 (rs6000_get_separate_components): Enable the TOC component if
611 saving the TOC register in the prologue.
612 (rs6000_components_for_bb): Handle the TOC component.
613 (rs6000_emit_prologue_components): Store the TOC register where needed.
614 (rs6000_set_handled_components): Mark TOC as handled, if handled.
615 (rs6000_emit_prologue): Don't save the TOC if that is already done.
617 2017-11-09 Martin Jambor <mjambor@suse.cz>
619 * ipa-param-manipulation.c: New file.
620 * ipa-param-manipulation.h: Likewise.
621 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
622 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
623 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
624 (ipa_parm_adjustment): Likewise.
625 (ipa_parm_adjustment_vec): Likewise.
626 (ipa_get_vector_of_formal_parms): Moved declaration to
627 ipa-param-manipulation.h.
628 (ipa_get_vector_of_formal_parm_types): Likewise.
629 (ipa_modify_formal_parameters): Likewise.
630 (ipa_modify_call_arguments): Likewise.
631 (ipa_combine_adjustments): Likewise.
632 (ipa_dump_param_adjustments): Likewise.
633 (ipa_modify_expr): Likewise.
634 (ipa_get_adjustment_candidate): Likewise.
635 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
636 ipa-param-manipulation.c.
637 (ipa_get_vector_of_formal_parm_types): Likewise.
638 (ipa_modify_formal_parameters): Likewise.
639 (ipa_modify_call_arguments): Likewise.
640 (ipa_modify_expr): Likewise.
641 (get_ssa_base_param): Likewise.
642 (ipa_get_adjustment_candidate): Likewise.
643 (index_in_adjustments_multiple_times_p): Likewise.
644 (ipa_combine_adjustments): Likewise.
645 (ipa_dump_param_adjustments): Likewise.
646 * tree-sra.c: Also include ipa-param-manipulation.h
647 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
650 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * doc/sourcebuild.texi (vect_masked_store): Document.
656 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
662 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
663 Alan Hayward <alan.hayward@arm.com>
664 David Sherwood <david.sherwood@arm.com>
666 * doc/sourcebuild.texi (vect_variable_length): Document.
668 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
669 Alan Hayward <alan.hayward@arm.com>
670 David Sherwood <david.sherwood@arm.com>
672 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
674 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
675 Alan Hayward <alan.hayward@arm.com>
676 David Sherwood <david.sherwood@arm.com>
678 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
680 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
681 Alan Hayward <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
685 previously undocumented selectors.
686 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
688 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
690 * doc/rtl.texi (const_vector): Say that elements can be
692 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
693 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
694 (gen_vec_duplicate): Use it instead of CONSTANT_P.
695 * optabs.c (expand_vector_broadcast): Likewise.
697 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
698 Alan Hayward <alan.hayward@arm.com>
699 David Sherwood <david.sherwood@arm.com>
701 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
702 scaled index even if the unscaled address was invalid.
703 Don't increase the complexity of using a scale in that case.
705 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
709 * doc/rtl.texi: Rewrite the subreg rules so that they partition
710 the inner register into REGMODE_NATURAL_SIZE bytes rather than
711 UNITS_PER_WORD bytes.
712 * emit-rtl.c (validate_subreg): Divide subregs into blocks
713 based on REGMODE_NATURAL_SIZE of the inner mode.
714 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
715 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
716 * expmed.c (lowpart_bit_field_p): Divide the value up into
717 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
718 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
719 whether something is likely to occupy more than one register.
721 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
724 * ipa-inline-transform.c (update_noncloned_frequencies): Use
725 profile_count::adjust_for_ipa_scaling.
726 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
727 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
729 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
731 2017-11-09 Jakub Jelinek <jakub@redhat.com>
733 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
734 (split_group): Add total_orig and total_new arguments, estimate the
735 number of statements related to the store group without store merging
736 and with store merging.
737 (imm_store_chain_info::output_merged_store): Adjust split_group
738 callers, punt if estimated number of statements with store merging
739 is not smaller than estimated number of statements without it.
741 (handled_load): Remove has_single_use checks.
742 (pass_store_merging::process_store): Likewise.
744 2017-11-09 Richard Biener <rguenther@suse.de>
746 PR tree-optimization/82902
747 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
749 2017-11-09 Martin Liska <mliska@suse.cz>
752 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
755 2017-11-09 Martin Liska <mliska@suse.cz>
757 PR tree-optimization/82669
758 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
760 2017-11-09 Martin Liska <mliska@suse.cz>
762 PR gcov-profile/48463
763 * coverage.c (coverage_begin_function): Output also end locus
764 of a function and information whether the function is
766 * gcov-dump.c (tag_function): Parse and print the information.
767 * gcov.c (INCLUDE_MAP): Add include.
768 (INCLUDE_SET): Likewise.
769 (struct line_info): Move earlier in the source file because
770 of vector<line_info> in function_info structure.
771 (line_info::line_info): Likewise.
772 (line_info::has_block): Likewise.
773 (struct source_info): Add new member index.
774 (source_info::get_functions_at_location): New function.
775 (function_info::group_line_p): New function.
776 (output_intermediate_line): New function.
777 (output_intermediate_file): Use the mentioned function.
778 (struct function_start): New.
779 (struct function_start_pair_hash): Likewise.
780 (process_file): Add code that identifies group functions.
781 Assign lines either to global or function scope.
782 (generate_results): Skip artificial functions.
783 (find_source): Assign index for each source file.
784 (read_graph_file): Read new flag artificial and end_line.
785 (add_line_counts): Assign it either to global of function scope.
786 (accumulate_line_counts): Isolate core of the function to
787 accumulate_line_info and call it for both function and global
789 (accumulate_line_info): New function.
790 (output_line_beginning): Fix GNU coding style.
791 (print_source_line): New function.
792 (output_line_details): Likewise.
793 (output_function_details): Likewise.
794 (output_lines): Iterate both source (global) scope and function
796 (struct function_line_start_cmp): New class.
797 * doc/gcov.texi: Reflect changes in documentation.
799 2017-11-09 Jakub Jelinek <jakub@redhat.com>
802 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
803 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
804 and similarly for not instead of neg.
806 2017-11-08 Andi Kleen <ak@linux.intel.com>
808 * config/i386/i386.opt: Add -mforce-indirect-call.
809 * config/i386/predicates.md: Check for flag_force_indirect_call.
810 * doc/invoke.texi: Document -mforce-indirect-call
812 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
814 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
816 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
817 config/riscv/riscv.c (predict.h): New include.
818 (riscv_slow_unaligned_access_p): No longer static.
819 (riscv_block_move_straight): Add require.
820 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
822 2017-11-08 Jakub Jelinek <jakub@redhat.com>
825 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
826 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
827 nonimmediate_operand predicate for operand 1 instead of
830 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
832 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
834 * config/aarch64/constraints.md (Uml): New constraint.
835 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
838 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
840 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
841 of two vec_duplicates into a vec_concat.
843 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
845 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
846 Simplify vec_merge of vec_duplicate and vec_concat.
847 * config/aarch64/constraints.md (Utq): New constraint.
848 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
851 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
853 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
854 Simplify vec_merge of vec_duplicate and const_vector.
855 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
857 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
858 mode iterator. Update predicate on operand 1 to
859 handle non-const_vec constants. Delete constraints.
860 (*aarch64_combinez_be<mode>): Likewise for operand 2.
862 2017-11-08 Jakub Jelinek <jakub@redhat.com>
864 PR tree-optimization/78821
865 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
867 (store_operand_info::store_operand_info): Initialize it to false.
868 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
869 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
870 store in the group, and if chain_info is non-NULL, to ignore altogether
872 (compatible_load_p): Fail if bit_not_p does not match.
873 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
874 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
875 (pass_store_merging::process_store): Adjust
876 terminate_all_aliasing_chains calls to pass NULL in all current spots,
877 call terminate_all_aliasing_chains newly when adding a store into
878 a chain with non-NULL chain_info.
880 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
882 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
884 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
886 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
888 (aarch64_layout_frame): Initialise emit_frame_chain.
889 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
890 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
892 2017-11-08 Martin Liska <mliska@suse.cz>
894 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
895 of gimple_call_internal_p.
897 2017-11-07 Tom de Vries <tom@codesourcery.com>
899 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
901 2017-11-07 Tom de Vries <tom@codesourcery.com>
903 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
906 2017-11-08 Martin Liska <mliska@suse.cz>
909 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
911 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
913 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
915 2017-11-07 Jakub Jelinek <jakub@redhat.com>
918 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
919 (*cmp<mode>_ccz_1): New insn with $k alternative.
922 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
923 RTX_COMM_COMPARE as commutative as well.
924 (ix86_binary_operator_ok): Formatting fix.
925 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
926 *<code><mode>3<mask_name><round_saeonly_name>,
927 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
928 *<s>mul<mode>3_highpart<mask_name>,
929 *vec_widen_umult_even_v16si<mask_name>,
930 *vec_widen_umult_even_v8si<mask_name>,
931 *vec_widen_umult_even_v4si<mask_name>,
932 *vec_widen_smult_even_v16si<mask_name>,
933 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
934 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
935 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
936 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
937 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
938 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
939 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
940 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
941 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
942 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
943 ix86_binary_operator_ok. Formatting fixes.
944 (*<plusminus_insn><mode>3<mask_name><round_name>,
945 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
948 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
950 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
951 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
952 eq and ne if TARGET_ISEL.
953 (cmp): New code_iterator.
954 (UNS, UNSU_, UNSIK): New code_attrs.
955 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
956 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
958 ("eq<mode>3"): ... this.
959 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
961 ("ne<mode>3"): ... this.
963 2017-11-07 Julia Koval <julia.koval@intel.com>
966 * common/config/i386/i386-common.c
967 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
968 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
969 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
970 * config/i386/i386.opt: Ditto.
971 * config/i386/i386.c (ix86_target_string): Ditto.
972 (ix86_option_override_internal): Ditto.
973 (ix86_init_mpx_builtins): Move MPX to args2.
974 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
975 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
976 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
977 __builtin_ia32_vgf2p8affineinvqb_v32qi,
978 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
979 __builtin_ia32_vgf2p8affineinvqb_v16qi,
980 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
982 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
985 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
986 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
987 (zero-extendsidi peephole2): Remove peephole.
989 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
992 * stor-layout.c: Include attribs.h.
993 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
994 explicit lookup of "aligned" attribute.
996 2017-11-07 Andrew Waterman <andrew@sifive.com>
998 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
999 (riscv_expand_block_move): Likewise.
1000 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1002 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1003 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1004 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1005 (riscv_adjust_block_mem): Likewise.
1006 (riscv_block_move_loop): Likewise.
1007 (riscv_expand_block_move): Likewise.
1008 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1010 2017-11-07 Michael Clark <michaeljclark@mac.com>
1012 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1013 (MUSL_DYNAMIC_LINKER): Likewise.
1015 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1017 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1018 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1021 2017-11-07 Richard Biener <rguenther@suse.de>
1023 * match.pd: Fix build.
1025 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1026 Jackson Woodruff <jackson.woodruff@arm.com>
1028 PR tree-optimization/71026
1029 * match.pd: Canonicalize negate in division.
1031 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1034 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1036 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1038 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1039 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1041 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1043 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1044 non-scalar integral types.
1045 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1046 (-(A-B), -(~A)): New transformations.
1048 2017-11-07 Tom de Vries <tom@codesourcery.com>
1050 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1051 semicolon after "do {} while (0)".
1052 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1053 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1054 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1055 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1056 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1058 2017-11-07 Tom de Vries <tom@codesourcery.com>
1060 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1061 after "do {} while (0)".
1062 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1063 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1064 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1065 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1066 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1068 2017-11-07 Tom de Vries <tom@codesourcery.com>
1071 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1073 (arm_rtx_costs_internal): Add missing semicolon after
1074 HANDLE_NARROW_SHIFT_ARITH call.
1076 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1078 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1079 disable isel if it was not set explicitly.
1081 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1083 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1085 (add_type_attribute) likewise.
1087 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1089 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1090 of struct ix86_frame.
1091 (ix86_initial_elimination_offset): Likewise.
1092 (ix86_expand_split_stack_prologue): Likewise.
1094 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1096 * tree-vrp.h (enum value_range_type): Update stale comment.
1098 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1099 Alan Hayward <alan.hayward@arm.com>
1100 David Sherwood <david.sherwood@arm.com>
1102 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1103 (aarch64_expand_vec_perm_const): Take the number of units too.
1104 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1105 (aarch64_expand_vec_perm_const): Likewise.
1106 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1107 (vec_perm<mode>): Update accordingly.
1109 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1110 Alan Hayward <alan.hayward@arm.com>
1111 David Sherwood <david.sherwood@arm.com>
1113 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1114 Take the number of units too.
1115 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1116 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1117 but check for a vector mode before rather than after the call.
1118 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1119 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1120 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1121 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1122 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1123 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1124 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1125 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1126 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1127 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1128 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1129 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1130 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1131 (aarch64_sqdmull2_n<mode>): Update accordingly.
1133 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1134 Alan Hayward <alan.hayward@arm.com>
1135 David Sherwood <david.sherwood@arm.com>
1137 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1138 the number of units too.
1139 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1140 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1141 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1142 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1143 (vec_store_lanesxi<mode>): Update accordingly.
1145 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1146 Alan Hayward <alan.hayward@arm.com>
1147 David Sherwood <david.sherwood@arm.com>
1149 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1150 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1151 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1152 of units rather than the mode.
1153 * config/aarch64/iterators.md (nunits): New mode attribute.
1154 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1155 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1156 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1157 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1158 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1159 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1160 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1161 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1162 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1163 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1164 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1165 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1166 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1167 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1168 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1169 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1170 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1171 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1172 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1173 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1174 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1175 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1176 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1177 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1178 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1179 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1180 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1181 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1182 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1183 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1184 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1185 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1186 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1187 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1188 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1189 (aarch64_simd_vec_setv2di): Likewise.
1191 2017-11-06 Carl Love <cel@us.ibm.com>
1193 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1195 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1196 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1198 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1200 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1201 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1202 Add power 8 macro expansions.
1203 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1204 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1205 power 8 instructions. (VSX_XXBR): Add iterator.
1207 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1209 * config/arm/arm.md (predicable_short_it): Change default to "no",
1210 improve documentation, remove uses that are identical to the default.
1211 (enabled_for_depr_it): Rename to enabled_for_short_it.
1212 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1213 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1214 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1215 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1216 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1218 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1221 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1222 float128 helper macros, which are no longer used after deleting
1223 the old 'q' built-in functions, and moving the round to odd
1224 built-in functions to being special built-in functions.
1225 (BU_FLOAT128_2): Likewise.
1226 (BU_FLOAT128_1_HW): Likewise.
1227 (BU_FLOAT128_2_HW): Likewise.
1228 (BU_FLOAT128_3_HW): Likewise.
1229 (FABSQ): Delete old 'q' built-in functions.
1230 (COPYSIGNQ): Likewise.
1231 (SQRTF128_ODD): Move round to odd built-in functions to be
1232 special built-in functions, so that we can handle
1233 -mabi=ieeelongdouble.
1234 (TRUNCF128_ODD): Likewise.
1235 (ADDF128_ODD): Likewise.
1236 (SUBF128_ODD): Likewise.
1237 (MULF128_ODD): Likewise.
1238 (DIVF128_ODD): Likewise.
1239 (FMAF128_ODD): Likewise.
1240 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1241 built-in names to 'f128'.
1242 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1243 old 'q' built-in functions, as the machine independent code for
1244 'f128' built-in functions handles this.
1245 (rs6000_expand_builtin): Add expansion for float128 round to odd
1246 functions, keying off on -mabi=ieeelongdouble of whether to use
1247 the KFmode or TFmode variant.
1248 (rs6000_init_builtins): Initialize the _Float128 round to odd
1250 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1251 _Float128 'q' built-in functions are now mapped into the new
1252 'f128' built-in functions.
1254 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1256 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1257 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1259 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1262 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1263 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1264 * toplev.c: Include "ipa-fnsummary.h".
1265 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1267 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1269 PR tree-optimization/82838
1270 * gimple-ssa-store-merging.c
1271 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1272 on a separate gimple_seq which is then appended to seq.
1274 2017-11-06 Jeff Law <law@redhat.com>
1277 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1278 (get_probe_interval): New functions.
1279 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1280 (ix86_adjust_stack_and_probe): Likewise.
1281 (output_adjust_stack_and_probe): Likewise.
1282 (ix86_emit_probe_stack_range): Likewise.
1283 (ix86_expand_prologue): Likewise.
1285 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1287 PR tree-optimization/82816
1288 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1289 if the modes of the two types are the same.
1290 (convert_plusminus_to_widen): Likewise.
1292 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1294 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1296 (usadv16qi): New define_expand.
1297 (usadv8hi): New define_expand.
1299 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1302 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1304 (inline_transform): Likewise.
1305 * predict.c (counts_to_freqs): Remove useless conditional.
1306 * profile-count.h (profile_count::apply_scale): Move sanity check.
1307 * tree-inline.c (copy_bb): Always scale.
1308 (copy_cfg_body): Likewise.
1310 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1313 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1315 (*cmp_ite0): Add enabled_for_depr_it attribute.
1316 (*cmp_ite1): Likewise.
1318 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1320 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1323 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1325 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1326 return the value by pointer.
1327 (extract_range_from_multiplicative_op_1): Update accordingly.
1328 Return as soon as an operation fails.
1330 2017-11-05 Tom de Vries <tom@codesourcery.com>
1333 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1334 (DEF_SANITIZER_BUILTIN): ... here.
1335 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1336 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1338 2017-11-05 Tom de Vries <tom@codesourcery.com>
1341 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1343 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1344 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1345 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1347 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1348 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1349 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1351 2017-11-05 Tom de Vries <tom@codesourcery.com>
1354 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1357 2017-11-04 Michael Clark <michaeljclark@mac.com>
1359 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1360 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1362 (*addsi3_extended): Likewise.
1363 (*addsi3_extended2): Likewise.
1364 (<optab>si3): Likewise.
1365 (<optab>di3): Likewise.
1366 (<optab><mode>3): Likewise.
1367 (<*optabe>si3_internal): Likewise.
1368 (zero_extendqi<SUPERQI:mode>2): Likewise.
1369 (*add<mode>hi3): Likewise.
1370 (*xor<mode>hi3): Likewise.
1371 (<optab>di3): Likewise.
1372 (*<optab>si3_extend): Likewise.
1373 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1374 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1375 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1377 2017-11-04 Andrew Waterman <andrew@sifive.com>
1379 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1380 TARGET_STRICT_ALIGN based upon -mtune argument.
1382 2017-11-04 Andrew Waterman <andrew@sifive.com>
1384 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1386 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1388 config/i386/i386.c (choose_basereg): Use optional scratch
1389 register and add assertion.
1390 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1391 needed, and don't allocate stack.
1392 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1393 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1394 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1396 2017-11-03 Jeff Law <law@redhat.com>
1398 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1399 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1400 to probe at the start of a noreturn function.
1402 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1404 PR tree-optimization/78821
1405 * gimple-ssa-store-merging.c: Update the file comment.
1406 (MAX_STORE_ALIAS_CHECKS): Define.
1407 (struct store_operand_info): New type.
1408 (store_operand_info::store_operand_info): New constructor.
1409 (struct store_immediate_info): Add rhs_code and ops data members.
1410 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1411 and op1r arguments to the ctor, initialize corresponding data members.
1412 (struct merged_store_group): Add load_align_base and load_align
1414 (merged_store_group::merged_store_group): Initialize them.
1415 (merged_store_group::do_merge): Update them.
1416 (merged_store_group::apply_stores): Pick the constant for
1417 encode_tree_to_bitpos from one of the two operands, or skip
1418 encode_tree_to_bitpos if neither operand is a constant.
1419 (class pass_store_merging): Add process_store method decl. Remove
1420 bool argument from terminate_all_aliasing_chains method decl.
1421 (pass_store_merging::terminate_all_aliasing_chains): Remove
1422 var_offset_p argument and corresponding handling.
1423 (stmts_may_clobber_ref_p): New function.
1424 (compatible_load_p): New function.
1425 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1426 if there is overlap and rhs_code is not INTEGER_CST. For
1427 non-overlapping stores terminate group if rhs is not mergeable.
1428 (get_alias_type_for_stmts): Change first argument from
1429 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1430 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1431 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1433 (get_location_for_stmts): Change first argument from
1434 auto_vec<gimple *> & to vec<gimple *> &.
1435 (struct split_store): Remove orig_stmts data member, add orig_stores.
1436 (split_store::split_store): Create orig_stores rather than orig_stmts.
1437 (find_constituent_stmts): Renamed to ...
1438 (find_constituent_stores): ... this. Change second argument from
1439 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1440 to info structures rather than the statements.
1441 (split_group): Rename ALLOW_UNALIGNED argument to
1442 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1443 it. Adjust find_constituent_stores caller.
1444 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1445 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1446 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1447 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1448 (mem_valid_for_store_merging): New function.
1449 (handled_load): New function.
1450 (pass_store_merging::process_store): New method.
1451 (pass_store_merging::execute): Use process_store method. Adjust
1452 terminate_all_aliasing_chains caller.
1454 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1456 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1457 Return true for more constants, symbols and label references.
1458 (aarch64_valid_floating_const): Remove unused function.
1460 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1463 * config/aarch64/aarch64.c (aarch64_layout_frame):
1464 Undo forcing of LR at bottom of frame.
1466 2017-11-03 Jeff Law <law@redhat.com>
1469 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1470 for int_registers_saved.
1472 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1473 extracted from tree-ssa-dom.c.
1474 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1475 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1476 (record_equivalences_from_incoming_edge): Add additional argument
1477 to single_pred_edge_ignoring_loop_edges call.
1478 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1479 (uncprop_dom_walker::before_dom_children): Add additional argument
1480 to single_pred_edge_ignoring_loop_edges call.
1481 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1482 single_pred_edge_ignoring_loop_edges rather than open coding.
1483 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1485 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1487 * match.pd (-(-A)): Rewrite.
1489 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1491 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1492 (rs6000_emit_int_cmove): New declaration.
1493 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1494 (rs6000_emit_sISEL): Delete.
1495 (rs6000_emit_int_cmove): Make non-static.
1496 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1497 instead of rs6000_emit_sISEL.
1499 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1501 * asan.c (create_cond_insert_point): Maintain profile.
1502 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1504 * basic-block.h (struct basic_block_def): Remove frequency.
1505 (EDGE_FREQUENCY): Use to_frequency
1506 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1508 (find_traces): Update to use to_frequency.
1509 (find_traces_1_round): Likewise; use only IPA counts.
1510 (bb_to_key): Likewise.
1511 (connect_traces): Use IPA counts only.
1512 (copy_bb_p): Update to use to_frequency.
1513 (fix_up_crossing_landing_pad): Likewise.
1514 (sanitize_hot_paths): Likewise.
1515 * bt-load.c (basic_block_freq): Likewise.
1516 * cfg.c (init_flow): Set count_max to uninitialized.
1517 (check_bb_profile): Remove frequencies; check counts.
1518 (dump_bb_info): Do not dump frequencies.
1519 (update_bb_profile_for_threading): Update counts only.
1520 (scale_bbs_frequencies_int): Likewise.
1521 (MAX_SAFE_MULTIPLIER): Remove.
1522 (scale_bbs_frequencies_gcov_type): Update counts only.
1523 (scale_bbs_frequencies_profile_count): Update counts only.
1524 (scale_bbs_frequencies): Update counts only.
1525 * cfg.h (struct control_flow_graph): Add count-max.
1526 (update_bb_profile_for_threading): Update prototype.
1527 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1528 (find_many_sub_basic_blocks): Likewise.
1529 * cfgcleanup.c (try_forward_edges): Likewise.
1530 (try_crossjump_to_edge): Likewise.
1531 * cfgexpand.c (expand_gimple_cond): Likewise.
1532 (expand_gimple_tailcall): Likewise.
1533 (construct_init_block): Likewise.
1534 (construct_exit_block): Likewise.
1535 * cfghooks.c (verify_flow_info): Check consistency of counts.
1536 (dump_bb_for_graph): Do not dump frequencies.
1537 (split_block_1): Do not update frequencies.
1538 (split_edge): Do not update frequencies.
1539 (make_forwarder_block): Do not update frequencies.
1540 (duplicate_block): Do not update frequencies.
1541 (account_profile_record): Do not update frequencies.
1542 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1543 for global heuristics.
1544 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1545 (expected_loop_iterations_unbounded): Use counts only.
1546 * cfgloopmanip.c (scale_loop_profile): Simplify.
1547 (create_empty_loop_on_edge): Simplify
1549 (duplicate_loop_to_header_edge): Simplify
1550 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1551 (update_br_prob_note): Take care of removing note when profile
1553 (relink_block_chain): Do not dump frequency.
1554 (rtl_account_profile_record): Use to_frequency.
1555 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1556 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1557 (cgraph_update_edges_for_call_stmt_node): Likewise.
1558 (cgraph_edge::verify_count_and_frequency): Update.
1559 (cgraph_node::verify_node): Temporarily disable frequency verification.
1560 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1561 to_cgraph_frequency.
1562 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1563 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1565 (cgraph_node::expand_thunk): Update profile.
1566 * except.c (dw2_build_landing_pads): Do not update frequency.
1567 * final.c (compute_alignments): Use to_frequency.
1568 (dump_basic_block_info): Do not dump frequency.
1569 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1570 (dump_gimple_bb_header): Do not dump frequency.
1571 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1573 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1574 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1575 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1576 (init_before_recovery): Do not update frequency.
1577 (sched_create_recovery_edges): Do not update frequency.
1578 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1579 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1580 (ipa_cp_c_finalize): Set max_count to uninitialized.
1581 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1582 (param_change_prob): Use counts.
1583 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1585 * ipa-split.c (consider_split): Use to_frequency.
1586 (split_function): Use to_frequency.
1587 * ira-build.c (loop_compare_func): Likewise.
1588 (mark_loops_for_removal): Likewise.
1589 (mark_all_loops_for_removal): Likewise.
1590 * loop-doloop.c (doloop_modify): Do not update frequency.
1591 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1593 * lto-streamer-in.c (input_function): Update count_max.
1594 * omp-expand.c (expand_omp_taskreg): Update count_max.
1595 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1596 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1597 (maybe_hot_count_p): Use ipa counts only.
1598 (maybe_hot_bb_p): Simplify.
1599 (maybe_hot_edge_p): Simplify.
1600 (probably_never_executed): Do not take frequency argument.
1601 (probably_never_executed_bb_p): Do not pass frequency.
1602 (probably_never_executed_edge_p): Likewise.
1603 (combine_predictions_for_bb): Check that profile is nonzero.
1604 (propagate_freq): Do not set frequency.
1605 (drop_profile): Simplify.
1606 (counts_to_freqs): Simplify.
1607 (expensive_function_p): Use to_frequency.
1608 (propagate_unlikely_bbs_forward): Simplify.
1609 (determine_unlikely_bbs): Simplify.
1610 (estimate_bb_frequencies): Add hack to silence graphite issues.
1611 (compute_function_frequency): Use ipa counts.
1612 (pass_profile::execute): Update.
1613 (rebuild_frequencies): Use counts only.
1614 (force_edge_cold): Use counts only.
1615 * profile-count.c (profile_count::dump): Dump new count types.
1616 (profile_count::differs_from_p): Check compatiblity.
1617 (profile_count::to_frequency): New function.
1618 (profile_count::to_cgraph_frequency): New function.
1619 * profile-count.h (struct function): Declare.
1620 (enum profile_quality): Add profile_guessed_local and
1621 profile_guessed_global0.
1622 (class profile_proability): Decrease number of bits to 29;
1623 update from_reg_br_prob_note and to_reg_br_prob_note.
1624 (class profile_count: Update comment; decrease number of bits
1625 to 61. Check compatibility.
1626 (profile_count::compatible_p): New private member function.
1627 (profile_count::ipa_p): New member function.
1628 (profile_count::operator<): Handle global zero correctly.
1629 (profile_count::operator>): Handle global zero correctly.
1630 (profile_count::operator<=): Handle global zero correctly.
1631 (profile_count::operator>=): Handle global zero correctly.
1632 (profile_count::nonzero_p): New member function.
1633 (profile_count::force_nonzero): New member function.
1634 (profile_count::max): New member function.
1635 (profile_count::apply_scale): Handle IPA scalling.
1636 (profile_count::guessed_local): New member function.
1637 (profile_count::global0): New member function.
1638 (profile_count::ipa): New member function.
1639 (profile_count::to_frequency): Declare.
1640 (profile_count::to_cgraph_frequency): Declare.
1641 * profile.c (OVERLAP_BASE): Delete.
1642 (compute_frequency_overlap): Delete.
1643 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1644 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1645 * sched-ebb.c (rank): Use counts only.
1646 * shrink-wrap.c (handle_simple_exit): Use counts only.
1647 (try_shrink_wrapping): Use counts only.
1648 (place_prologue_for_one_component): Use counts only.
1649 * tracer.c (find_best_predecessor): Use to_frequency.
1650 (find_trace): Use to_frequency.
1651 (tail_duplicate): Use to_frequency.
1652 * trans-mem.c (expand_transaction): Do not update frequency.
1653 * tree-call-cdce.c: Do not update frequency.
1654 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1655 (gimple_merge_blocks): Likewise.
1656 (gimple_split_edge): Likewise.
1657 (gimple_duplicate_sese_region): Likewise.
1658 (gimple_duplicate_sese_tail): Likewise.
1659 (move_sese_region_to_fn): Likewise.
1660 (gimple_account_profile_record): Likewise.
1661 (insert_cond_bb): Likewise.
1662 * tree-complex.c (expand_complex_div_wide): Likewise.
1663 * tree-eh.c (lower_resx): Update profile.
1664 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1666 (initialize_cfun): Do not initialize frequencies
1667 (freqs_to_counts): Delete.
1668 (copy_cfg_body): Ignore count parameter.
1669 (copy_body): Update.
1670 (expand_call_inline): Update count_max.
1671 (optimize_inline_calls): Update count_max.
1672 (tree_function_versioning): Update count_max.
1673 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1674 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1676 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1677 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1678 (try_peel_loop): Likewise.
1679 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1681 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1682 (tree_transform_and_unroll_loop): Do not use frequencies
1683 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1684 Use reliable prediction only.
1685 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1686 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1687 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1688 probability scaling.
1689 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1690 not update frequency
1691 (any_remaining_duplicated_blocks): Likewise.
1692 (update_profile): Likewise.
1693 (estimated_freqs_path): Delete.
1694 (freqs_to_counts_path): Delete.
1695 (clear_counts_path): Delete.
1696 (ssa_fix_duplicate_block_edges): Likewise.
1697 (duplicate_thread_path): Likewise.
1698 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1699 * tree-tailcall.c (decrease_profile): Do not update frequency.
1700 (eliminate_tail_call): Likewise.
1701 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1702 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1703 (optimize_mask_stores): Likewise.
1704 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1705 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1706 (ubsan_expand_ptr_ifn): Update profile.
1707 * value-prof.c (gimple_ic): Simplify.
1708 * value-prof.h (gimple_ic): Update prototype.
1709 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1710 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1712 (want_inline_self_recursive_call_p): Likewise.
1713 (resolve_noninline_speculation): Only cummulate defined counts.
1714 (inline_small_functions): Use nonzero_p.
1715 (ipa_inline): Do not access freed node.
1717 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1719 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1720 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1722 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1724 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1725 non-legitimate address.
1727 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1729 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1730 (*lt0_<mode>di, *lt0_<mode>si): New.
1732 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1734 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1735 TARGET_PAIRED_FLOAT.
1737 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1738 Jim Wilson <jim.wilson@linaro.org>
1740 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1741 * config/aarch64/aarch64-tune.md: Regenerated.
1742 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1743 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1745 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1747 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1748 (arc_expand_prologue): Restore blink for millicode.
1749 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1751 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1754 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1755 gen_vec_duplicate after forcing the scalar into a register.
1757 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1759 * combine (try_combine): Print the insns input to try_combine to the
1762 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1765 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1766 Remove second argument from aarch64_process_target_attr call.
1767 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1769 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1771 (aarch64_handle_attr_arch): Remove second argument.
1772 (aarch64_handle_attr_cpu): Ditto.
1773 (aarch64_handle_attr_tune): Ditto.
1774 (aarch64_handle_attr_isa_flags): Ditto.
1775 (aarch64_process_one_target_attr): Ditto.
1776 (aarch64_process_target_attr): Ditto.
1777 (aarch64_option_valid_attribute_p): Remove second argument.
1778 on aarch64_process_target_attr call.
1780 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1782 * diagnostic.c: Include "selftest-diagnostic.h".
1783 (selftest::assert_location_text): New function.
1784 (selftest::test_diagnostic_get_location_text): New function.
1785 (selftest::diagnostic_c_tests): Call it.
1787 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1789 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1790 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1791 (class selftest::test_diagnostic_context): Move to...
1792 * selftest-diagnostic.c: New file.
1793 * selftest-diagnostic.h: New file.
1795 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1797 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1798 offset range for FT32B.
1799 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1800 * config/ft32/ft32.md: Add TARGET_NOPM.
1801 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1802 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1804 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1806 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1808 2017-11-02 Jeff Law <law@redhat.com>
1810 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1811 virtual keyword on FINAL OVERRIDE members.
1813 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1814 virtuals together. Add virtual destructor.
1815 (substitute_and_fold_engine): Similarly.
1817 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1819 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1821 2017-11-02 Richard Biener <rguenther@suse.de>
1823 PR tree-optimization/82795
1824 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1826 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1828 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1830 (gcc_GAS_CHECK_FEATURE): Remove.
1831 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1832 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1833 * configure: Regenerate.
1835 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1837 * config/arc/arc.c (hwloop_optimize): Account for empty
1840 2017-11-02 Richard Biener <rguenther@suse.de>
1843 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1844 Truncate ARRAY_REF index and element size.
1846 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1848 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1850 2017-11-01 Jeff Law <law@redhat.com>
1852 * tree-ssa-ccp.c (ccp_folder): New class derived from
1853 substitute_and_fold_engine.
1854 (ccp_folder::get_value): New member function.
1855 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1856 (ccp_fold_stmt): Remove prototype.
1857 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1858 * tree-ssa-copy.c (copy_folder): New class derived from
1859 substitute_and_fold_engine.
1860 (copy_folder::get_value): Renamed from get_value.
1861 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1862 * tree-vrp.c (vrp_folder): New class derived from
1863 substitute_and_fold_engine.
1864 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1865 (vrp_folder::get_value): New member function.
1866 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1867 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1868 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1869 provide a class interface to folder/substitute routines.
1870 (ssa_prop_fold_stmt_fn): Remove typedef.
1871 (ssa_prop_get_value_fn): Likewise.
1872 (subsitute_and_fold): Remove prototype.
1873 (replace_uses_in): Likewise.
1874 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1875 Renamed from replace_uses_in. Call the virtual member function
1876 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1877 (substitute_and_fold_dom_walker): Remove initialization of
1878 data member entries for calbacks. Add substitute_and_fold_engine
1879 member and initialize it.
1880 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1881 member functions for get_value, replace_phi_args_in c
1882 replace_uses_in, and fold_stmt calls.
1883 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1884 substitute_and_fold. Remove assert. Update ctor call.
1886 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1887 (ssa_prop_visit_phi_fn): Likewise.
1888 (class ssa_propagation_engine): New class to provide an interface
1890 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1892 (ssa_prop_visit_phi): Likewise.
1893 (ssa_propagation_engine::simulate_stmt): Moved into class.
1894 Call visit_phi/visit_stmt from the class rather than via
1895 file scoped static variables.
1896 (ssa_propagation_engine::simulate_block): Moved into class.
1897 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1898 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1899 set file scoped statics for the visit_stmt/visit_phi callbacks.
1900 * tree-complex.c (complex_propagate): New class derived from
1901 ssa_propagation_engine.
1902 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1903 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1904 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1906 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1907 ssa_propagation_engine.
1908 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1909 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1910 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1911 * tree-ssa-copy.c (copy_prop): New class derived from
1912 ssa_propagation_engine.
1913 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1914 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1915 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1916 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1917 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1918 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1919 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1921 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1923 PR rtl-optimization/82778
1924 PR rtl-optimization/82597
1925 * compare-elim.c (struct comparison): Add in_a_setter field.
1926 (find_comparison_dom_walker::before_dom_children): Remove killed
1927 bitmap and df_simulate_find_defs call, instead walk the defs.
1928 Compute last_setter and initialize in_a_setter. Merge definitions
1929 with first initialization for a few variables.
1930 (try_validate_parallel): Use insn_invalid_p instead of
1931 recog_memoized. Return insn rather than just the pattern.
1932 (try_merge_compare): Fix up comment. Don't uselessly test if
1933 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1935 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1938 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1939 Alan Hayward <alan.hayward@arm.com>
1940 David Sherwood <david.sherwood@arm.com>
1942 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1943 aarch64_hard_regno_nregs to get the number of registers
1946 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1947 Alan Hayward <alan.hayward@arm.com>
1948 David Sherwood <david.sherwood@arm.com>
1950 * config/aarch64/constraints.md (Upl): Rename to...
1952 * config/aarch64/aarch64.md
1953 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1956 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1957 Alan Hayward <alan.hayward@arm.com>
1958 David Sherwood <david.sherwood@arm.com>
1960 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1961 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1964 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1965 Alan Hayward <alan.hayward@arm.com>
1966 David Sherwood <david.sherwood@arm.com>
1968 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1969 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1970 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1972 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1973 of a single element.
1974 * config/aarch64/iterators.md: Add a comment above the permute
1975 unspecs to say that they are generated directly by
1976 aarch64_expand_vec_perm_const.
1977 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1979 2017-11-01 Nathan Sidwell <nathan@acm.org>
1981 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1983 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1985 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1986 and medany code models, and describe what they do.
1988 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1990 Revert accidental duplicate:
1992 * combine.c (can_change_dest_mode): Reject changes in
1993 REGMODE_NATURAL_SIZE.
1995 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1997 PR rtl-optimization/64682
1998 PR rtl-optimization/69567
1999 PR rtl-optimization/69737
2000 PR rtl-optimization/82683
2001 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2002 register mentioned in the note, drop the note, unless it came from I3,
2003 in which case it should go to I3 again.
2005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2007 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2008 and return false if not.
2009 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2011 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2013 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2014 range is known to be empty.
2016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2017 Alan Hayward <alan.hayward@arm.com>
2018 David Sherwood <david.sherwood@arm.com>
2020 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2021 and CONST_VECTOR_NUNITS instead of computing the number of units from
2022 the byte sizes of the vector and element.
2023 (simplify_binary_operation_1): Likewise.
2024 (simplify_const_binary_operation): Likewise.
2025 (simplify_ternary_operation): Likewise.
2027 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2028 Alan Hayward <alan.hayward@arm.com>
2029 David Sherwood <david.sherwood@arm.com>
2031 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2032 (int_mem_offset): ...this new function.
2033 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2034 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2035 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2038 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2039 Alan Hayward <alan.hayward@arm.com>
2040 David Sherwood <david.sherwood@arm.com>
2042 * lower-subreg.c (interesting_mode_p): New function.
2043 (compute_costs, find_decomposable_subregs, decompose_register)
2044 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2045 (resolve_clobber, dump_choices): Use it.
2047 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2048 Alan Hayward <alan.hayward@arm.com>
2049 David Sherwood <david.sherwood@arm.com>
2051 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2054 Alan Hayward <alan.hayward@arm.com>
2055 David Sherwood <david.sherwood@arm.com>
2057 * alias.c (find_base_value, find_base_term): Only process integer
2058 truncations. Check the precision rather than the size.
2060 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2061 Alan Hayward <alan.hayward@arm.com>
2062 David Sherwood <david.sherwood@arm.com>
2064 * machmode.h (is_narrower_int_mode): New function
2065 * optabs.c (expand_float, expand_fix): Use it.
2066 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2068 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2069 Alan Hayward <alan.hayward@arm.com>
2070 David Sherwood <david.sherwood@arm.com>
2072 * rtl.h (narrower_subreg_mode): New function.
2073 * ira-color.c (update_costs_from_allocno): Use it.
2075 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2076 Alan Hayward <alan.hayward@arm.com>
2077 David Sherwood <david.sherwood@arm.com>
2079 * optabs-query.h (convert_optab_p): New function, split out from...
2080 (convert_optab_handler): ...here.
2081 (widening_optab_handler): Delete.
2082 (find_widening_optab_handler): Remove permit_non_widening parameter.
2083 (find_widening_optab_handler_and_mode): Likewise. Provide an
2084 override that operates on mode class wrappers.
2085 * optabs-query.c (widening_optab_handler): Delete.
2086 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2087 parameter. Assert that the two modes are the same class and that
2088 the "from" mode is narrower than the "to" mode. Use
2089 convert_optab_handler instead of widening_optab_handler.
2090 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2091 instead of widening_optab_handler.
2092 * expr.c (expand_expr_real_2): Update calls to
2093 find_widening_optab_handler.
2094 * optabs.c (expand_widen_pattern_expr): Likewise.
2095 (expand_binop_directly): Take the insn_code as a parameter.
2096 (expand_binop): Only call find_widening_optab_handler for
2097 conversion optabs; use optab_handler otherwise. Update calls
2098 to find_widening_optab_handler and expand_binop_directly.
2099 Use convert_optab_handler instead of widening_optab_handler.
2100 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2101 find_widening_optab_handler and use scalar_mode rather than
2103 (convert_plusminus_to_widen): Likewise.
2105 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * machmode.h (fixed_size_mode): New class.
2110 * rtl.h (get_pool_mode): Return fixed_size_mode.
2111 * gengtype.c (main): Add fixed_size_mode.
2112 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2113 (get_raw_arg_mode): Likewise.
2114 * doc/tm.texi: Regenerate.
2115 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2116 * targhooks.c (default_get_reg_raw_mode): Likewise.
2117 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2118 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2119 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2120 (msp430_get_raw_result_mode): Likewise.
2121 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2122 * dbxout.c (dbxout_parms): Require fixed-size modes.
2123 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2124 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2125 * omp-low.c (lower_oacc_reductions): Likewise.
2126 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2127 (simplify_subreg): Update accordingly.
2128 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2129 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2130 that aren't fixed-size.
2131 (get_pool_mode): Return a fixed_size_mode.
2132 (output_constant_pool_2): Take a fixed_size_mode.
2134 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * doc/rtl.texi (vec_series): Document.
2139 (const): Say that the operand can be a vec_series.
2140 * rtl.def (VEC_SERIES): New rtx code.
2141 * rtl.h (const_vec_series_p_1): Declare.
2142 (const_vec_series_p): New function.
2143 * emit-rtl.h (gen_const_vec_series): Declare.
2144 (gen_vec_series): Likewise.
2145 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2146 (gen_vec_series): Likewise.
2147 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2148 * simplify-rtx.c (simplify_unary_operation): Handle negations
2150 (simplify_binary_operation_series): New function.
2151 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2152 (test_vector_ops_series): New function.
2153 (test_vector_ops): Call it.
2154 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2155 gen_const_vec_series.
2156 (altivec_lvsr): Likewise.
2157 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2159 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2160 Alan Hayward <alan.hayward@arm.com>
2161 David Sherwood <david.sherwood@arm.com>
2163 * doc/rtl.texi (const): Update description of address constants.
2164 Say that vector constants are allowed too.
2165 * common.md (E, F): Use CONSTANT_P instead of checking for
2167 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2168 checking for CONST_VECTOR.
2169 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2171 * expr.c (expand_expr_real_2): Check for vector modes instead
2172 of checking for CONST_VECTOR.
2173 * rtl.h (const_vec_p): New function.
2174 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2175 (unwrap_const_vec_duplicate): Handle them here too.
2177 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2178 David Malcolm <dmalcolm@redhat.com>
2179 Alan Hayward <alan.hayward@arm.com>
2180 David Sherwood <david.sherwood@arm.com>
2182 * rtl.h (vec_duplicate_p): New function.
2183 * selftest-rtl.c (assert_rtx_eq_at): New function.
2184 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2185 (assert_rtx_eq_at): Declare.
2186 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2187 * selftest-run-tests.c (selftest::run_tests): Call it.
2188 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2189 (simplify_unary_operation_1): Recursively handle vector duplicates.
2190 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2192 (simplify_subreg): Handle subregs of vector duplicates.
2193 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2194 (selftest::simplify_rtx_c_tests): New functions.
2196 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2197 Alan Hayward <alan.hayward@arm.com>
2198 David Sherwood <david.sherwood@arm.com>
2200 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2201 (gen_vec_duplicate): Likewise.
2202 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2204 (gen_const_vector): ...here.
2205 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2206 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2207 whose elements are all equal.
2208 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2209 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2210 (simplify_relational_operation): Likewise.
2211 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2213 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2214 (aarch64_expand_vector_init): Likewise.
2215 * config/arm/arm.c (neon_vdup_constant): Likewise.
2216 (neon_expand_vector_init): Likewise.
2217 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2218 (arm_block_set_unaligned_vect): Likewise.
2219 (arm_block_set_aligned_vect): Likewise.
2220 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2221 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2222 (expand_vec_perm_even_odd_pack): Likewise.
2223 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2224 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2225 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2226 gen_const_vec_duplicate.
2227 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2228 * config/mips/mips.c (mips_gen_const_int_vector): Use
2229 gen_const_vec_duplicate.
2230 (mips_expand_vector_init): Use CONST0_RTX.
2231 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2232 (define_split): Use gen_const_vec_duplicate.
2233 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2234 (define_split): Use gen_const_vec_duplicate.
2235 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2236 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2237 * config/spu/spu.c (spu_const): Likewise.
2239 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2240 Alan Hayward <alan.hayward@arm.com>
2241 David Sherwood <david.sherwood@arm.com>
2243 * combine.c (can_change_dest_mode): Reject changes in
2244 REGMODE_NATURAL_SIZE.
2246 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2248 * configure.ac (--enable-libssp): New.
2249 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2250 trying to determine target-specific default. Adjust indentation.
2251 * configure: Regenerated.
2252 * doc/install.texi (Configuration): Expand --disable-libssp
2255 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2257 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2260 2017-10-31 Martin Jambor <mjambor@suse.cz>
2263 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2265 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2267 * auto-profile.c (autofdo_source_profile::read): Use
2268 UNKNOWN_LOCATION rather than 0.
2269 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2270 (warning_at): ...this overload.
2271 (warning_at_rich_loc_n): Rename to...
2272 (warning_n): ...this overload.
2273 (error_at_rich_loc): Rename to...
2274 (error_at): ...this overload.
2275 (pedwarn_at_rich_loc): Rename to...
2276 (pedwarn): ...this overload.
2277 (permerror_at_rich_loc): Rename to...
2278 (permerror): ...this overload.
2279 (inform_at_rich_loc): Rename to...
2280 (inform): ...this overload.
2281 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2282 (diagnostic_n_impl_richloc): Rename to...
2283 (diagnostic_n_impl): ...this rich_location *-based decl.
2284 (inform_at_rich_loc): Rename to...
2285 (inform): ...this, and add an assertion.
2286 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2287 (warning_at_rich_loc): Rename to...
2288 (warning_at): ...this, and add an assertion.
2289 (warning_at_rich_loc_n): Rename to...
2290 (warning_n): ...this, and add an assertion.
2291 (warning_n): Update location_t-based implementation for removal of
2292 location_t-based diagnostic_n_impl.
2293 (pedwarn_at_rich_loc): Rename to...
2294 (pedwarn): ...this, and add an assertion.
2295 (permerror_at_rich_loc): Rename to...
2296 (permerror): ...this, and add an assertion.
2297 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2298 (error_at_rich_loc): Rename to...
2299 (error_at): ...this, and add an assertion.
2300 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2301 (driver::do_spec_on_infiles): Likewise.
2302 * substring-locations.c (format_warning_va): Update for renaming
2303 of inform_at_rich_loc.
2305 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2307 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2308 _Float<N>X built-in functions so that the variant without the
2309 "__builtin_" prefix is only enabled for the GNU C and Objective C
2310 languages when they are in non-strict ANSI/ISO mode.
2311 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2312 * target.def (floatn_builtin_p): Add a target hook to control
2313 whether _Float<N> and _Float<N>X built-in functions without the
2314 "__builtin_" prefix are enabled, and return true for C and
2315 Objective C in the default hook. Include langhooks.h in
2317 * targhooks.h (default_floatn_builtin_p): Likewise.
2318 * targhooks.c (default_floatn_builtin_p): Likewise.
2319 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2320 floatn_builtin_p target hook.
2321 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2323 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2324 Eric Botcazou <ebotcazou@adacore.com>
2326 PR rtl-optimization/81803
2327 * lra-constraints.c (curr_insn_transform): Also reload the whole
2328 register for a strict subreg no wider than a word if this is for
2329 a WORD_REGISTER_OPERATIONS target.
2331 2017-10-31 Jason Merrill <jason@redhat.com>
2333 * gdbinit.in: Skip over inlines from timevar.h.
2335 2017-10-31 Martin Liska <mliska@suse.cz>
2337 * doc/gcov.texi: Document new option.
2338 * gcov.c (print_usage): Likewise print it.
2339 (process_args): Support the argument.
2340 (format_count): New function.
2341 (format_gcov): Use the function.
2343 2017-10-31 Martin Liska <mliska@suse.cz>
2345 * gcov.c (struct name_map): do not use typedef.
2346 Define operator== and operator<.
2347 (name_search): Remove.
2348 (name_sort): Remove.
2349 (main): Do not allocate names.
2350 (process_file): Add vertical space.
2351 (generate_results): Use std::find.
2352 (release_structures): Do not release memory.
2353 (find_source): Use std::find.
2355 2017-10-31 Martin Liska <mliska@suse.cz>
2357 * gcov.c (struct line_info): Remove it's typedef.
2358 (line_info::line_info): Add proper ctor.
2359 (line_info::has_block): Do not use a typedef.
2360 (struct source_info): Do not use typedef.
2361 (circuit): Likewise.
2362 (get_cycles_count): Likewise.
2363 (output_intermediate_file): Iterate via vector iterator.
2364 (add_line_counts): Use std::vector methods.
2365 (accumulate_line_counts): Likewise.
2366 (output_lines): Likewise.
2368 2017-10-31 Martin Liska <mliska@suse.cz>
2370 * gcov.c (struct source_info): Remove typedef.
2371 (source_info::source_info): Add proper ctor.
2372 (accumulate_line_counts): Use struct, not it's typedef.
2373 (output_gcov_file): Likewise.
2374 (output_lines): Likewise.
2375 (main): Do not allocate an array.
2376 (output_intermediate_file): Use size of vector container.
2377 (process_file): Resize the vector.
2378 (generate_results): Do not preallocate, use newly added vector
2380 (release_structures): Do not release sources.
2381 (find_source): Use vector methods.
2382 (add_line_counts): Do not use typedef.
2384 2017-10-31 Martin Liska <mliska@suse.cz>
2386 * doc/gcov.texi: Document that.
2387 * gcov.c (add_line_counts): Mark lines with a non-executed
2389 (output_line_beginning): Handle such lines.
2390 (output_lines): Pass new argument.
2391 (output_intermediate_file): Print it in intermediate format.
2393 2017-10-31 Martin Liska <mliska@suse.cz>
2395 * color-macros.h: New file.
2396 * diagnostic-color.c: Factor out color related to macros to
2398 * doc/gcov.texi: Document -k option.
2399 * gcov.c (INCLUDE_STRING): Include string.h.
2400 (print_usage): Add -k option.
2401 (process_args): Parse it.
2402 (pad_count_string): New function.
2403 (output_line_beginning): Likewise.
2404 (DEFAULT_LINE_START): New macro.
2405 (output_lines): Support color output.
2407 2017-10-31 Martin Liska <mliska@suse.cz>
2409 PR gcov-profile/82633
2410 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2411 their interaction with GCOV infrastructure.
2412 * configure.ac: Add -fkeep-{inline,static}-functions to
2414 * configure: Regenerate.
2416 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2419 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2421 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2424 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2425 into a register if it does not fit into an immediate offset field.
2427 2017-10-31 Olivier Hainque <hainque@adacore.com>
2429 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2431 2017-10-31 Julia Koval <julia.koval@intel.com>
2433 * config.gcc: Add gfniintrin.h.
2434 * config/i386/gfniintrin.h: New.
2435 * config/i386/i386-builtin-types.def
2436 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2437 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2438 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2439 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2440 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2441 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2442 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2443 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2444 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2445 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2446 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2447 * config/i386/immintrin.h: Include gfniintrin.h.
2448 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2450 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2452 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2454 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2456 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2457 (arm_ashldi3_1bit): Remove pattern.
2458 (ashrdi3): Remove shift by 1 expansion.
2459 (arm_ashrdi3_1bit): Remove pattern.
2460 (lshrdi3): Remove shift by 1 expansion.
2461 (arm_lshrdi3_1bit): Remove pattern.
2462 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2463 cost of ashldi3 by 1.
2464 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2465 (<shift>di3_neon): Likewise.
2467 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2469 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2470 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2471 and (*aarch64_simd_mov<VQ:mode>).
2472 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2473 pattern alternative.
2474 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2475 attributes to match pattern alternative.
2477 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2479 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2480 * config/rs6000/emmintrin.h: New file.
2481 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2483 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2485 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2486 * (movdi_vfp_cortexa8): Remove pattern.
2488 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2490 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2492 (Specific, alpha*-dec-osf5.1): Remove.
2493 (Specific, mips-sgi-irix5): Remove.
2494 (Specific, mips-sgi-irix6): Remove.
2496 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2499 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2500 arguments to clear_bit_region_be.
2502 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2504 * gimplify.c: Include memmodel.h.
2506 2017-10-30 Martin Jambor <mjambor@suse.cz>
2508 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2509 condition whether loop should be executed at all.
2511 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2514 gimple folding of vec_madd() intrinsics.
2515 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2516 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2517 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2519 2017-10-30 Richard Biener <rguenther@suse.de>
2521 PR tree-optimization/82762
2523 2017-10-23 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/82129
2527 2017-08-01 Richard Biener <rguenther@suse.de>
2529 PR tree-optimization/81181
2530 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2531 (compute_antic): ... end of iteration here.
2533 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2535 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2536 -std=iso9899:2017 and -std=gnu17.
2537 * doc/standards.texi (C Language): Document C17 support.
2538 * doc/cpp.texi (Overview): Mention -std=c17.
2539 (Standard Predefined Macros): Document C11 and C17 values of
2540 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2541 * doc/extend.texi (Inline): Do not list individual options for
2542 standards newer than C99.
2543 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2545 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2548 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2550 * asan.c (asan_finish_file): Align asan globals array by shadow
2553 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2556 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2557 (struct store_immediate_info): Add bitregion_start and bitregion_end
2559 (store_immediate_info::store_immediate_info): Add brs and bre
2560 arguments and initialize bitregion_{start,end} from those.
2561 (struct merged_store_group): Add bitregion_start, bitregion_end,
2562 align_base and mask fields. Drop unnecessary struct keyword from
2563 struct store_immediate_info. Add do_merge method.
2564 (clear_bit_region_be): Use memset instead of loop storing zeros.
2565 (merged_store_group::do_merge): New method.
2566 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2567 stores as long as the surrounding bitregions have no gaps.
2568 (merged_store_group::merge_overlapping): Use do_merge.
2569 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2570 is byte aligned, rather than requiring that start and width are
2571 byte aligned. Drop unnecessary struct keyword from
2572 struct store_immediate_info. Allocate and populate also mask array.
2573 Make start of the arrays relative to bitregion_start rather than
2574 start and size them according to bitregion_{end,start} difference.
2575 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2576 struct store_immediate_info.
2577 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2578 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2579 struct keyword from struct store_immediate_info.
2580 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2581 between stores as long as the surrounding bitregions have no gaps.
2583 (struct split_store): Add orig non-static data member.
2584 (split_store::split_store): Initialize orig to false.
2585 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2586 if there is exactly a single original stmt. Change stmts argument
2587 to pointer from reference, if NULL, don't push anything to it. Add
2588 first argument, use it to optimize skipping over orig stmts that
2589 are known to be before bitpos already. Simplify.
2590 (split_group): Return unsigned int count how many stores are or
2591 would be needed rather than a bool. Add allow_unaligned argument.
2592 Change split_stores argument from reference to pointer, if NULL,
2593 only do a dry run computing how many stores would be produced.
2594 Rewritten algorithm to use both alignment and misalign if
2595 !allow_unaligned and handle bitfield stores with gaps.
2596 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2597 from bitregion_start instead of start. Compute allow_unaligned
2598 here, if true, do 2 split_group dry runs to compute which one
2599 produces fewer stores and prefer aligned if equal. Punt if
2600 new count is bigger or equal than original before emitting any
2601 statements, rather than during that. Remove no longer needed
2602 new_ssa_names tracking. Replace num_stmts with
2603 split_stores.length (). Use 32-bit stack allocated entries
2604 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2605 if possible. Handle bitfields with gaps.
2606 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2607 Compute bitregion_{start,end} for the stores and construct
2608 store_immediate_info with that. Formatting fixes.
2610 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2613 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2614 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2616 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2618 * gimplify.c: Include tm_p.h.
2620 * common.opt (gcoff): Re-add as ignored option.
2621 (gcoff1, gcoff2, gcoff3): Likewise.
2623 * Makefile.in (OBJS): Delete sdbout.o.
2624 (GTFILES): Delete $(srcdir)/sdbout.c.
2625 * debug.h: Delete sdb_debug_hooks.
2626 * final.c: Delete sdbout.h include.
2627 (final_scan_insn): Delete SDB_DEBUG check.
2628 (rest_of_clean_state): Likewise.
2629 * output.h: Delete sdb_begin_function_line.
2632 * toplev.c: Delete sdbout.h include.
2633 (process_options): Delete SDB_DEBUG check.
2634 * tree-core.h (tree_type_common): Delete pointer field of
2636 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2637 TYPE_SYMTAB_POINTER.
2638 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2639 (TYPE_SYMTAB_IS_POINTER): Delete.
2640 (TYPE_SYMTAB_IS_DIE): Renumber.
2641 * xcoffout.c: Refer to former sdbout.c file.
2642 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2644 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2645 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2646 (Debugging Options): Delete -gcoff.
2647 (-gstabs): Delete SDB reference.
2649 (-gcoff@var{level}): Delete.
2650 * doc/passes.texi (Debugging information output): Delete SDB and
2651 sdbout.c references.
2652 * doc/tm.texi: Regenerate.
2653 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2654 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2656 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2657 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2658 (SDB_DEBUGGING_INFO): Delete.
2659 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2660 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2661 * target.def (output_source_filename): Delete COFF reference.
2663 * common.opt (gcoff): Delete.
2664 (gxcoff+): Update Negative chain.
2665 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2667 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2668 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2669 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2671 (expand_function_start): Change sdb reference to past tense.
2672 (expand_function_end): Change sdb reference to past tense.
2673 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2674 * opts.c (debug_type_names): Delete coff entry.
2675 (common_handle_option): Delete OPT_gcoff case.
2676 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2678 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2679 * config/cris/cris.h: Delete SDB reference in comment.
2680 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2681 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2682 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2683 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2685 (ix86_expand_prologue): Likewise.
2686 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2687 * config/ia64/ia64.h: Likewise.
2688 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2689 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2691 * config/mmix/mmix.h: Likewise.
2692 * config/nds32/nds32.c: Likewise.
2693 * config/stormy/storym16.h: Likewise.
2694 * config/visium/visium.h: Likewise.
2695 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2697 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2699 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2700 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2701 FRAME_POINTER_REGNUM point at high end of local var area.
2703 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2705 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2706 Move comment around. Do not reset best_edge for a copiable
2707 destination if the copy would cause a partition change.
2708 (better_edge_p): Remove redundant check.
2710 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2712 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2714 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2716 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2717 for math functions that have _Float<N> and _Float<N>X variants.
2718 (mathfn_built_in_2): Add support for math functions that have
2719 _Float<N> and _Float<N>X variants.
2720 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2721 (expand_builtin_mathfn_ternary): Add support for fma with
2722 _Float<N> and _Float<N>X variants.
2723 (expand_builtin): Likewise.
2724 (fold_builtin_3): Likewise.
2725 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2726 create math function _Float<N> and _Float<N>X variants as external
2728 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2729 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2730 the __builtin_ prefix and if not strict ansi, without the prefix.
2731 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2732 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2733 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2734 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2735 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2736 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2737 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2738 function signatures for fma _Float<N> and _Float<N>X variants.
2739 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2740 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2741 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2742 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2743 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2744 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2745 * gencfn-macros.c (print_case_cfn): Add support for math functions
2746 that have _Float<N> and _Float<N>X variants.
2747 (print_define_operator_list): Likewise.
2748 (fltfn_suffixes): Likewise.
2750 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2751 for math functions that have _Float<N> and _Float<N>X variants.
2752 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2753 and _Float<N>X variants.
2754 (COPYSIGN): Likewise.
2757 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2758 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2760 (integer_valued_read_call_p): Likewise.
2761 * fold-const-call.c (fold_const_call_ss): Likewise.
2762 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2763 _Float<N> and _Float<N>X variants.
2764 (fold_const_call_ssss): Add support for fma _Float<N> and
2765 _Float<N>X variants.
2766 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2767 support for copysign and fma _Float<N> and _Float<N>X variants.
2768 (backprop::process_builtin_call_use): Likewise.
2769 * tree-call-cdce.c (can_test_argument_range); Add support for
2770 sqrt _Float<N> and _Float<N>X variants.
2771 (edom_only_function): Likewise.
2772 (get_no_error_domain): Likewise.
2773 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2774 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2775 copysign _Float<N> and _Float<N>X variants.
2776 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2777 handled by machine independent code.
2778 (FMAF128): Likewise.
2779 * doc/cpp.texi (Common Predefined Macros): Document defining
2780 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2781 fma _Float<N> and _Float<N>X variants.
2783 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2786 * config/i386/i386-modes.def (CCFPU): Remove definition.
2787 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2788 (ix86_cc_modes_compatible): Ditto.
2789 (ix86_expand_carry_flag_compare): Ditto.
2790 (ix86_expand_int_movcc): Ditto.
2791 (ix86_expand_int_addcc): Ditto.
2792 (ix86_reverse_condition): Ditto.
2793 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2794 Return true/false for unordered/ordered fp comparisons.
2795 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2796 (ix86_prepare_fp_compare_args): Update for rename.
2797 (ix86_expand_fp_compare): Update for rename. Generate unordered
2798 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2799 (ix86_expand_sse_compare_and_jump): Ditto.
2800 * config/i386/predicates.md (fcmov_comparison_operator):
2801 Remove CCFPU mode handling.
2802 (ix86_comparison_operator): Ditto.
2803 (ix86_carry_flag_operator): Ditto.
2804 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2805 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2806 (*cmpu<mode>_cc_i387): Ditto.
2807 (FPCMP): Remove mode iterator.
2808 (unord): Remove mode attribute.
2809 (unord_subst): New define_subst transformation
2810 (unord): New define_subst attribute.
2812 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2813 (*cmpi<unord>xf_i387): Ditto.
2814 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2815 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2816 using unord_subst transformation.
2817 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2818 (round_saeonly): Also handle CCFP mode.
2819 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2820 Remove UNSPEC_SAHF unspec handling.
2822 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2824 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2826 2017-10-27 Jeff Law <law@redhat.com>
2828 * gimple-ssa-sprintf.c: Include domwalk.h.
2829 (class sprintf_dom_walker): New class, derived from dom_walker.
2830 (sprintf_dom_walker::before_dom_children): New function.
2831 (struct call_info): Moved into sprintf_dom_walker class
2832 (compute_formath_length, handle_gimple_call): Likewise.
2833 (sprintf_length::execute): Call the dominator walker rather
2834 than walking the statements.
2836 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2837 gimple statement locations.
2838 (check_array_bounds): Corresponding changes. Get the statement's
2839 location directly from wi->stmt.
2841 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2844 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2846 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2848 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2849 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2851 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2854 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2855 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2856 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2857 maybe_get_pool_constant.
2858 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2861 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2863 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2864 2.26 caveat. Update gas and gld versions.
2865 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2868 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2870 * cgraph.h (set_malloc_flag): Declare.
2871 * cgraph.c (set_malloc_flag_1): New function.
2872 (set_malloc_flag): Likewise.
2873 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2874 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2876 (read_ipa_call_summary): Add support for reading is_return_callee.
2877 (write_ipa_call_summary): Stream is_return_callee.
2878 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2879 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2880 ipa-prop.h, ipa-fnsummary.h.
2881 (pure_const_names): Change to static.
2882 (malloc_state_e): Define.
2883 (malloc_state_names): Define.
2884 (funct_state_d): Add field malloc_state.
2885 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2886 (check_retval_uses): New function.
2887 (malloc_candidate_p): Likewise.
2888 (analyze_function): Add support for malloc attribute.
2889 (pure_const_write_summary): Stream malloc_state.
2890 (pure_const_read_summary): Add support for reading malloc_state.
2891 (dump_malloc_lattice): New function.
2892 (propagate_malloc): New function.
2893 (warn_function_malloc): New function.
2894 (ipa_pure_const::execute): Call propagate_malloc and
2895 ipa_free_fn_summary.
2896 (pass_local_pure_const::execute): Add support for malloc attribute.
2897 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2898 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2900 2017-10-27 Martin Liska <mliska@suse.cz>
2902 PR gcov-profile/82457
2903 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2904 for fork-like functions to be properly instrumented.
2906 2017-10-27 Richard Biener <rguenther@suse.de>
2909 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2910 info when we redirected EH.
2912 2017-10-26 Michael Collison <michael.collison@arm.com>
2914 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2916 (<optab>_trunchf<GPI:mode>2: New pattern.
2917 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2918 * config/aarch64/iterators.md (wv): New mode attribute.
2919 (vf, VF): New mode attributes.
2920 (vgp, VGP): New mode attributes.
2921 (s): Update attribute with SImode and DImode prefixes.
2923 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2925 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2926 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2927 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2928 (nios2_option_overide): Initialize it. Don't allow R0-relative
2929 addressing with PIC.
2930 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2931 (nios2_symbolic_constant_p): Likewise.
2932 (nios2_legitimate_address_p): Likewise.
2933 (nios2_r0rel_section_name_p): New.
2934 (nios2_symbol_ref_in_r0rel_data_p): New.
2935 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2936 (r0rel_constant_p): New.
2937 (nios2_print_operand_address): Handle r0rel_constant_p.
2938 (nios2_cdx_narrow_form_p): Likewise.
2939 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2940 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2941 (Nios II Options): Document -mr0rel-sec.
2943 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2945 * config/nios2/nios2.c: Include xregex.h.
2946 (nios2_gprel_sec_regex): New.
2947 (nios2_option_overide): Initialize it. Don't allow GP-relative
2948 addressing with PIC.
2949 (nios2_small_section_name_p): Check for regex match.
2950 * config/nios2/nios2.opt (mgprel-sec=): New option.
2951 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2952 (Nios II Options): Document -mgprel-sec.
2954 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2956 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2958 2017-10-26 Tom de Vries <tom@codesourcery.com>
2960 PR tree-optimization/82707
2961 * gimple.c (gimple_copy): Fix unsharing of
2962 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2964 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2966 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2967 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2968 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2969 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2970 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2971 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2972 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2973 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2974 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2975 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2976 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2977 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2978 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2979 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2980 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2981 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2982 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2984 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2986 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2988 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2989 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2990 warning to rs6000.c. Remove the Undocumented flag, since it has
2992 (-mabi=ibmlongdouble): Likewise.
2993 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2994 already set, set the default format for long double.
2995 (rs6000_debug_reg_global): Print whether long double is IBM or
2997 (rs6000_option_override_internal): Rework setting long double
2998 format. Only warn if the user is changing the long double default
2999 and they did not use -Wno-psabi.
3000 * doc/invoke.texi (PowerPC options): Update the documentation for
3001 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3003 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3004 Alan Hayward <alan.hayward@arm.com>
3005 David Sherwood <david.sherwood@arm.com>
3007 * rtl.h (wider_subreg_mode): New function.
3008 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3009 rather than an unsigned int *.
3010 * ira-color.c (regno_max_ref_width): Replace with...
3011 (regno_max_ref_mode): ...this new variable.
3012 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3013 Use wider_subreg_mode.
3014 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3015 rather than an unsigned int *.
3016 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3017 (process_alt_operands): Likewise.
3018 (invariant_p): Likewise.
3019 * lra-spills.c (assign_mem_slot): Likewise.
3020 (add_pseudo_to_slot): Likewise.
3021 * lra.c (collect_non_operand_hard_regs): Likewise.
3022 (add_regs_to_insn_regno_info): Likewise.
3023 * reload1.c (regno_max_ref_width): Replace with...
3024 (regno_max_ref_mode): ...this new variable.
3025 (reload): Update accordingly. Update call to
3026 ira_sort_regnos_for_alter_reg.
3027 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3028 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3029 (scan_paradoxical_subregs): Likewise.
3031 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3033 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3034 (aarch64_frame): Add emit_frame_chain boolean.
3035 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3036 Move eh_return case to aarch64_layout_frame.
3037 (aarch64_layout_frame): Initialize emit_frame_chain.
3038 (aarch64_expand_prologue): Use emit_frame_chain.
3040 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3042 * config/aarch64/aarch64.c (aarch64_layout_frame):
3043 Ensure LR is always stored at the bottom of the callee-saves.
3044 Remove rarely used frame layout which saves callee-saves at top of
3045 frame, so the store of LR can be used as a valid probe in all cases.
3047 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3049 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3050 Improve unaligned TImode/TFmode base/offset split.
3052 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3053 Alan Hayward <alan.hayward@arm.com>
3054 David Sherwood <david.sherwood@arm.com>
3056 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3057 * combine.c (find_single_use_1): Likewise.
3058 (expand_field_assignment): Likewise.
3059 (move_deaths): Likewise.
3060 * lra-constraints.c (simplify_operand_subreg): Likewise.
3061 (curr_insn_transform): Likewise.
3062 * lra.c (collect_non_operand_hard_regs): Likewise.
3063 (add_regs_to_insn_regno_info): Likewise.
3064 * rtlanal.c (reg_referenced_p): Likewise.
3065 (covers_regno_no_parallel_p): Likewise.
3067 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3069 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3070 Don't print any bits outside the precision of the value.
3071 * wide-int.cc (test_printing): Add some new tests.
3073 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3075 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3076 supports -xbrace_comment option.
3077 * configure: Regenerate.
3078 * config.in: Regenerate.
3079 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3080 (ASM_CPU_SPEC): Use it.
3082 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3084 * target.def (static_rtx_alignment): New hook.
3085 * targhooks.h (default_static_rtx_alignment): Declare.
3086 * targhooks.c (default_static_rtx_alignment): New function.
3087 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3088 * doc/tm.texi: Regenerate.
3089 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3090 instead of targetm.constant_alignment. Remove call to
3092 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3093 (cris_preferred_mininum_alignment): New function, split out from...
3094 (cris_constant_alignment): ...here.
3095 (cris_static_rtx_alignment): New function.
3096 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3098 (ix86_constant_alignment): ...here.
3099 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3100 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3101 (mmix_static_rtx_alignment): New function.
3102 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3103 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3105 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3108 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3109 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3111 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3113 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3114 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3115 mark as a sign-extending load.
3116 (local_pic_load_u): Define.
3118 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3121 * fold-const.c (operand_equal_for_comparison_p): Also return true
3122 if ARG0 is a simple variant of ARG1 with narrower precision.
3123 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3125 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3127 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3129 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3130 scatter_static, scatter_per_elt.
3131 * x86-tune-costs.h: Add new cost entries.
3133 2017-10-25 Richard Biener <rguenther@suse.de>
3135 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3136 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3137 class pass_fre): Move to ...
3138 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3139 class pass_fre): ... here and adjust for statistics.
3141 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3144 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3145 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3146 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3147 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3150 2017-10-25 Richard Biener <rguenther@suse.de>
3152 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3153 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3154 eliminate_push_avail, eliminate_insert): Move inside...
3155 (class eliminate_dom_walker): ... this class in preparation
3157 (fini_eliminate): Remove by merging with ...
3158 (eliminate): ... this function. Adjust for class changes.
3159 (pass_pre::execute): Remove fini_eliminate call.
3160 (pass_fre::execute): Likewise.
3162 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3165 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3166 (VPERMI2, VPERMI2I): New mode iterators.
3167 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3168 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3170 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3171 mode iterator. Remove 3 old define_insn patterns.
3172 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3173 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3174 VPERMI2 mode iterator, remove the other two expanders.
3175 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3176 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3177 instructions, remove the other two patterns.
3178 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3179 mode iterator, remove the other two patterns.
3180 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3181 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3182 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3183 and adjust argument order accordingly.
3184 (ix86_expand_vec_perm): Adjust caller.
3185 (expand_vec_perm_1): Likewise.
3186 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3187 (expand_vec_perm_vpermt2_vpshub2): ... this.
3188 (ix86_expand_vec_perm_const_1): Adjust caller.
3189 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3192 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3193 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3194 (vec_shl_<mode>): Remove unused expander.
3195 (avx512bw_<shift_insn><mode>3): New define_insn.
3196 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3197 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3199 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3202 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3205 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3207 PR rtl-optimization/82396
3208 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3209 (autopref_multipass_init): Simplify initialization.
3210 (autopref_rank_data): Simplify sort order.
3211 * gcc/sched-int.h (autopref_multipass_data_): Remove
3212 multi_mem_insn_p, min_offset and max_offset.
3214 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3217 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3218 Check special value of flag_omit_frame_pointer.
3219 (aarch64_can_eliminate): Likewise.
3220 (aarch64_override_options_after_change_1): Simplify handling of
3221 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3223 2017-10-24 Richard Biener <rguenther@suse.de>
3225 PR tree-optimization/82697
3226 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3227 zero for conditional load and unconditional store.
3229 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3231 * doc/install.texi: Document bootstrap-cet.
3233 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3236 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3237 ENDBR instruction at function entrance if function is only
3240 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3243 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3244 patterns to better describe from which operation the CF is computed.
3245 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3246 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3247 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3248 is 0, use _0 suffixed expanders instead of emitting a comparison
3251 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3253 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3254 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3256 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3259 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3260 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3261 * loop-iv.c (iv_get_reaching_def): Likewise.
3262 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3263 variable is promoted and the partition contains undefined values.
3265 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3267 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3269 (nios2_address_cost): Define.
3270 (nios2_legitimize_address): Recognize (exp + constant) directly.
3271 (TARGET_ADDRESS_COST): Define.
3273 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3275 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3276 (nios2_symbolic_memory_operand_p): Declare.
3277 (nios2_split_large_constant): Declare.
3278 (nios2_split_symbolic_memory_operand): Declare.
3279 * config/nios2/nios2.c: Adjust includes.
3280 (nios2_symbolic_constant_allowed): New.
3281 (nios2_symbolic_constant_p): New.
3282 (nios2_plus_symbolic_constant_p): New.
3283 (nios2_valid_addr_expr_p): Recognize addresses involving
3285 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3286 (nios2_symbolic_memory_operand_p): New.
3287 (nios2_large_constant_p): New.
3288 (nios2_split_large_constant): New.
3289 (nios2_split_plus_large_constant): New.
3290 (nios2_split_symbolic_memory_operand): New.
3291 (nios2_legitimize_address): Code refactoring. Handle addresses
3292 involving symbolic constants.
3293 (nios2_emit_move_sequence): Likewise.
3294 (nios2_print_operand): Improve error output.
3295 (nios2_print_operand_address): Handle LO_SUM.
3296 (nios2_cdx_narrow_form_p): Likewise.
3297 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3298 operands involving symbolic constants.
3299 (movhi_internal, movsi_internal): Likewise.
3300 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3301 (extendhisi2, extendqi<mode>2): Likewise.
3303 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3305 * tree-pass.h (PROP_rtl_split_insns): Define.
3306 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3308 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3310 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3312 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3315 * target.def (const_not_ok_for_debug_p): Default to
3316 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3317 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3318 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3319 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3320 which targetm.const_not_ok_for_debug_p returned true.
3321 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3323 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3325 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3326 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3327 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3329 (ix86_const_not_ok_for_debug_p): New function.
3330 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3331 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3333 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3336 * system.h: Conditionally include "unique-ptr.h" if
3337 INCLUDE_UNIQUE_PTR is defined.
3338 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3339 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3341 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3343 * config/rl78/rl78.md: New define_expand "subdi3".
3345 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3348 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3349 DF_REF_INSN if DF_REF_INSN_INFO is false.
3351 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3353 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3354 xmm_move instead of sse_move.
3355 (sse_store_index): New function.
3356 (ix86_register_move_cost): Be more sensible about mismatch stall;
3357 model AVX moves correctly; make difference between sse->integer and
3359 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3360 moves; make difference between SSE and AVX.
3361 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3362 and zmm_move. Increase size of sse load and store tables;
3363 add unaligned load and store tables; add ssemmx_to_integer.
3364 * x86-tune-costs.h: Update all entries according to real
3365 move latencies from Agner Fog's manual and chip documentation.
3367 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3370 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3371 * config/i386/constraints.md (Wf): New constraint.
3372 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3373 (cmp<dwi>_doubleword): Removed.
3374 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3375 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3376 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3377 expand with cmp<dwi>_doubleword. For LTU and GEU use
3378 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3380 * common.opt (gcolumn-info): Enable by default.
3381 * doc/invoke.texi (gcolumn-info): Document new default.
3383 2017-10-23 Richard Biener <rguenther@suse.de>
3385 PR tree-optimization/82672
3386 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3387 Fold the stmt if we propagated into it.
3389 2017-10-23 Richard Biener <rguenther@suse.de>
3391 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3392 (bitmap_remove_expr_from_set): ... this. All callers call this
3393 for non-constant values.
3394 (bitmap_set_subtract): Rename to...
3395 (bitmap_set_subtract_expressions): ... this. Adjust and
3397 (bitmap_set_contains_value): Remove superfluous check.
3398 (bitmap_set_replace_value): Inline into single caller ...
3399 (bitmap_value_replace_in_set): ... here and simplify.
3400 (dependent_clean): Merge into ...
3401 (clean): ... this using an overload. Adjust.
3402 (prune_clobbered_mems): Adjust.
3403 (compute_antic_aux): Likewise.
3404 (compute_partial_antic_aux): Likewise.
3406 2017-10-23 Richard Biener <rguenther@suse.de>
3408 PR tree-optimization/82129
3410 2017-08-01 Richard Biener <rguenther@suse.de>
3412 PR tree-optimization/81181
3413 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3414 (compute_antic): ... end of iteration here.
3416 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3418 * target.def (starting_frame_offset): New hook.
3419 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3420 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3421 * doc/tm.texi.in: Regenerate.
3422 * hooks.h (hook_hwi_void_0): Declare.
3423 * hooks.c (hook_hwi_void_0): New function.
3424 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3425 STARTING_FRAME_OFFSET.
3426 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3427 * reload1.c (reload): Likewise.
3428 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3429 instead of STARTING_FRAME_OFFSET.
3430 * function.c (try_fit_stack_local): Likewise.
3431 (assign_stack_local_1): Likewise
3432 (instantiate_virtual_regs): Likewise.
3433 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3434 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3435 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3436 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3437 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3438 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3439 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3440 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3441 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3442 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3443 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3444 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3445 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3446 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3447 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3448 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3449 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3450 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3451 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3452 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3453 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3454 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3455 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3456 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3457 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3458 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3459 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3460 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3461 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3462 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3463 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3464 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3465 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3466 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3467 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3468 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3469 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3470 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3471 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3472 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3473 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3474 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3475 return a HOST_WIDE_INT.
3476 (avr_builtin_setjmp_frame_value): Use it instead of
3477 STARTING_FRAME_OFFSET.
3478 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3479 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3480 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3482 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3483 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3484 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3485 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3486 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3487 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3488 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3489 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3490 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3491 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3492 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3493 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3495 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3496 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3497 * config/mips/mips.c (mips_compute_frame_info): Refer to
3498 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3499 (mips_starting_frame_offset): New function.
3500 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3501 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3502 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3503 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3504 and return a HOST_WIDE_INT.
3505 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3506 (mmix_initial_elimination_offset): Refer to
3507 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3508 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3509 * config/pa/pa.c (pa_starting_frame_offset): New function.
3510 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3511 (pa_expand_prologue): Likewise.
3512 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3513 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3514 !FRAME_GROWS_DOWNWARD handling to...
3515 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3516 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3517 !FRAME_GROWS_DOWNWARD handling to...
3518 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3519 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3520 !FRAME_GROWS_DOWNWARD handling to...
3521 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3522 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3524 (rs6000_starting_frame_offset): New function.
3525 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3526 !FRAME_GROWS_DOWNWARD handling to...
3527 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3528 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3529 !FRAME_GROWS_DOWNWARD handling to...
3530 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3531 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3532 !FRAME_GROWS_DOWNWARD handling to...
3533 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3534 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3535 (rs6000_starting_frame_offset): New function.
3536 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3537 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3538 * config/vax/vax.c (vax_starting_frame_offset): New function.
3539 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3540 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3541 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3542 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3543 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3544 * system.h (STARTING_FRAME_OFFSET): Poison.
3546 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3548 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3549 SCALAR_TYPE_MODE instead of TYPE_MODE.
3551 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3552 Alan Hayward <alan.hayward@arm.com>
3553 David Sherwood <david.sherwood@arm.com>
3555 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3557 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3558 Alan Hayward <alan.hayward@arm.com>
3559 David Sherwood <david.sherwood@arm.com>
3561 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3563 2017-10-23 Richard Biener <rguenther@suse.de>
3565 PR tree-optimization/82129
3566 * tree-ssa-pre.c (bitmap_set_and): Remove.
3567 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3568 canonicalizing expressions in the set to those with lowest
3569 ID rather than taking that from the first edge.
3571 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3573 * combine.c (rtx_equal_for_field_assignment_p): Use
3574 byte_lowpart_offset.
3576 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3577 Alan Hayward <alan.hayward@arm.com>
3578 David Sherwood <david.sherwood@arm.com>
3580 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3581 to a SUBREG_PROMOTED_VAR.
3583 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3584 Alan Hayward <alan.hayward@arm.com>
3585 David Sherwood <david.sherwood@arm.com>
3587 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3588 (expand_debug_source_expr): Likewise.
3589 * combine.c (combine_simplify_rtx): Likewise.
3590 * cse.c (fold_rtx): Likewise.
3591 * optabs.c (expand_float): Likewise.
3592 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3593 (simplify_binary_operation_1): Likewise.
3595 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3596 Alan Hayward <alan.hayward@arm.com>
3597 David Sherwood <david.sherwood@arm.com>
3599 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3600 (record_promoted_value): Likewise.
3601 * expr.c (expand_expr_real_2): Likewise.
3602 * ree.c (update_reg_equal_equiv_notes): Likewise.
3603 (combine_set_extension): Likewise.
3604 * rtlanal.c (low_bitmask_len): Likewise.
3605 * simplify-rtx.c (neg_const_int): Likewise.
3606 (simplify_binary_operation_1): Likewise.
3608 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3613 * regcprop.c (maybe_mode_change): Likewise.
3614 * reload1.c (alter_reg): Likewise.
3616 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3618 * inchash.h (inchash::hash::add_wide_int): New function.
3619 * lto-streamer-out.c (hash_tree): Use it.
3621 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3623 * inchash.h (inchash::hash::add_wide_int): Rename to...
3624 (inchash::hash::add_hwi): ...this.
3625 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3626 (polymorphic_call_target_hasher::hash): Likewise.
3627 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3628 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3629 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3630 * lto-streamer-out.c (hash_tree): Likewise.
3631 * optc-save-gen.awk: Likewise.
3632 * tree.c (add_expr): Likewise.
3634 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3637 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3638 for ordered inequality comparisons even with TARGET_IEEE_FP.
3640 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3643 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3644 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3645 Expand with cmp<dwi>_doubleword.
3647 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3649 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3650 List CET intrinsics.
3651 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3652 specific to -fcf-protection option.
3654 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3656 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3657 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3658 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3659 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3660 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3661 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3662 (extra_objs): Add cet.o for Linux/x86 targets.
3663 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3664 * config/i386/cet.c: New file.
3665 * config/i386/cetintrin.h: Likewise.
3666 * config/i386/t-cet: Likewise.
3667 * config/i386/cpuid.h (bit_SHSTK): New.
3668 (bit_IBT): Likewise.
3669 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3670 pass IBT and SHSTK bits.
3671 * config/i386/i386-builtin-types.def
3672 (VOID_FTYPE_UNSIGNED_PVOID): New.
3673 (VOID_FTYPE_UINT64_PVOID): Likewise.
3674 * config/i386/i386-builtin.def: Add CET intrinsics.
3675 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3676 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3677 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3678 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3680 * config/i386/i386.c (rest_of_insert_endbranch): New.
3681 (pass_data_insert_endbranch): Likewise.
3682 (pass_insert_endbranch): Likewise.
3683 (make_pass_insert_endbranch): Likewise.
3684 (ix86_notrack_prefixed_insn_p): Likewise.
3685 (ix86_target_string): Add -mibt, -mshstk flags.
3686 (ix86_option_override_internal): Add flag_cf_protection
3688 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3689 (ix86_print_operand): Add 'notrack' prefix output.
3690 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3691 (ix86_expand_builtin): Expand CET intrinsics.
3692 (x86_output_mi_thunk): Add 'endbranch' instruction.
3693 * config/i386/i386.h (TARGET_IBT): New.
3694 (TARGET_IBT_P): Likewise.
3695 (TARGET_SHSTK): Likewise.
3696 (TARGET_SHSTK_P): Likewise.
3697 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3698 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3699 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3700 (builtin_setjmp_setup): New pattern.
3701 (builtin_longjmp): Likewise.
3702 (rdssp<mode>): Likewise.
3703 (incssp<mode>): Likewise.
3704 (saveprevssp): Likewise.
3705 (rstorssp): Likewise.
3706 (wrss<mode>): Likewise.
3707 (wruss<mode>): Likewise.
3708 (setssbsy): Likewise.
3709 (clrssbsy): Likewise.
3710 (nop_endbr): Likewise.
3711 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3713 * config/i386/immintrin.h: Include <cetintrin.h>.
3714 * config/i386/linux-common.h
3715 (file_end_indicate_exec_stack_and_cet): New prototype.
3716 (TARGET_ASM_FILE_END): New.
3718 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3720 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3721 latencies instead of having separate table; make difference between
3722 integer and float costs.
3723 * i386.h (processor_costs): Remove scalar_stmt_cost,
3724 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3725 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3727 * x86-tune-costs.h: Remove entries which has been removed in
3728 procesor_costs from all tables; make cond_taken_branch_cost
3729 and cond_not_taken_branch_cost COST_N_INSNS based.
3731 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3733 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3735 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3737 * config/i386/i386.md (isa): Remove fma_avx512f.
3738 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3739 <avx512>_fmadd_<mode>_mask3<round_name>,
3740 <avx512>_fmsub_<mode>_mask<round_name>,
3741 <avx512>_fmsub_<mode>_mask3<round_name>,
3742 <avx512>_fnmadd_<mode>_mask<round_name>,
3743 <avx512>_fnmadd_<mode>_mask3<round_name>,
3744 <avx512>_fnmsub_<mode>_mask<round_name>,
3745 <avx512>_fnmsub_<mode>_mask3<round_name>,
3746 <avx512>_fmaddsub_<mode>_mask<round_name>,
3747 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3748 <avx512>_fmsubadd_<mode>_mask<round_name>,
3749 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3750 (*vec_widen_umult_even_v16si<mask_name>,
3751 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3752 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3754 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3756 * extend.texi: Add 'nocf_check' documentation.
3757 * gimple.texi: Add second parameter to
3758 gimple_build_call_from_tree.
3759 * invoke.texi: Add -fcf-protection documentation.
3760 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3762 2017-10-20 Richard Biener <rguenther@suse.de>
3764 PR tree-optimization/82473
3765 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3766 the largest input type.
3768 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3770 * c-attribs.c (handle_nocf_check_attribute): New function.
3771 (c_common_attribute_table): Add 'nocf_check' handling.
3772 * gimple-parser.c: Add second argument NULL to
3773 gimple_build_call_from_tree.
3774 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3775 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3777 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3778 * common.opt: Add fcf-protection flag.
3779 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3780 * flag-types.h: Add enum cf_protection_level.
3781 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3782 Add 'nocf_check' attribute propagation to gimple call.
3783 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3784 (gimple_build_call_from_tree): Update prototype.
3785 (gimple_call_nocf_check_p): New function.
3786 (gimple_call_set_nocf_check): Likewise.
3787 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3788 * ipa-icf.c: Add nocf_check attribute in statement hash.
3789 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3790 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3791 * toplev.c (process_options): Add flag_cf_protection handling.
3793 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3795 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3797 2017-10-20 Richard Biener <rguenther@suse.de>
3799 PR tree-optimization/82603
3800 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3801 remove false predicated stores.
3803 2017-10-20 Richard Biener <rguenther@suse.de>
3805 * graphite-isl-ast-to-gimple.c
3806 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3807 Remove return value and simplify, dump copied stmt after lhs
3809 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3810 Reduce dump verbosity.
3811 (gsi_insert_earliest): Likewise.
3812 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3813 * graphite.c (print_global_statistics): Adjust dumping.
3814 (print_graphite_scop_statistics): Likewise.
3815 (print_graphite_statistics): Do not dump loops here.
3816 (graphite_transform_loops): But here.
3818 2017-10-20 Nicolas Roche <roche@adacore.com>
3820 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3821 * configure: Regenerate.
3823 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3826 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3827 functions when optimizing replace GIMPLE_RETURN stmts with
3828 calls to __builtin_unreachable ().
3831 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3832 for -fsanitize=thread link of executables.
3833 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3834 link of executables.
3837 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3839 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3840 define_insns for logical vector shifts to use VI248_AVX512BW
3841 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3842 condition, useless isa and prefix attributes. Change the first
3843 2 of these define_insns to ...
3844 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3845 define_insn for avx512vl.
3846 (<shift_insn><mode>3): ... and this, new define_insn without
3847 masking for non-avx512vl.
3850 * config/i386/sse.md (*andnot<mode>3,
3851 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3852 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3853 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3854 not applied use empty suffix even for TARGET_AVX512VL.
3855 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3856 is applied, supply evex,evex or evex,evex,evex instead of just
3859 2017-10-20 Julia Koval <julia.koval@intel.com>
3861 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3862 (OPTION_MASK_ISA_GFNI_UNSET): New.
3863 (ix86_handle_option): Handle OPT_mgfni.
3864 * config/i386/cpuid.h (bit_GFNI): New.
3865 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3866 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3867 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3868 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3869 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3870 * config/i386/i386.opt: Add mgfni.
3872 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3874 * config/msp430/msp430.c (msp430_option_override): Disable
3875 -fdelete-null-pointer-checks.
3876 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3878 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3880 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3881 of x87 and SSE instructions.
3883 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3885 * asan.c (create_cond_insert_point): Do not update edge count.
3886 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3887 (afdo_propagate_circuit): Likewise.
3888 (afdo_calculate_branch_prob): Likewise.
3889 (afdo_annotate_cfg): Likewise.
3890 * basic-block.h (struct edge_def): Remove count.
3891 (edge_def::count): New accessor.
3892 * bb-reorder.c (rotate_loop): Update.
3893 (find_traces_1_round): Update.
3894 (connect_traces): Update.
3895 (sanitize_hot_paths): Update.
3896 * cfg.c (unchecked_make_edge): Update.
3897 (make_single_succ_edge): Update.
3898 (check_bb_profile): Update.
3899 (dump_edge_info): Update.
3900 (update_bb_profile_for_threading): Update.
3901 (scale_bbs_frequencies_int): Update.
3902 (scale_bbs_frequencies_gcov_type): Update.
3903 (scale_bbs_frequencies_profile_count): Update.
3904 (scale_bbs_frequencies): Update.
3905 * cfganal.c (connect_infinite_loops_to_exit): Update.
3906 * cfgbuild.c (compute_outgoing_frequencies): Update.
3907 (find_many_sub_basic_blocks): Update.
3908 * cfgcleanup.c (try_forward_edges): Update.
3909 (try_crossjump_to_edge): Update
3910 * cfgexpand.c (expand_gimple_cond): Update
3911 (expand_gimple_tailcall): Update
3912 (construct_exit_block): Update
3913 * cfghooks.c (verify_flow_info): Update
3914 (redirect_edge_succ_nodup): Update
3915 (split_edge): Update
3916 (make_forwarder_block): Update
3917 (duplicate_block): Update
3918 (account_profile_record): Update
3919 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3920 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3921 * cfgloopmanip.c (scale_loop_profile): Update.
3923 (lv_adjust_loop_entry_edge): Update.
3924 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3925 (force_nonfallthru_and_redirect): Update.
3926 (purge_dead_edges): Update.
3927 (rtl_flow_call_edges_add): Update.
3928 * cgraphunit.c (init_lowered_empty_function): Update.
3929 (cgraph_node::expand_thunk): Update.
3930 * gimple-pretty-print.c (dump_probability): Update.
3931 (dump_edge_probability): Update.
3932 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3933 * haifa-sched.c (sched_create_recovery_edges): Update.
3934 * hsa-gen.c (convert_switch_statements): Update.
3935 * ifcvt.c (dead_or_predicable): Update.
3936 * ipa-inline-transform.c (inline_transform): Update.
3937 * ipa-split.c (split_function): Update.
3938 * ipa-utils.c (ipa_merge_profiles): Update.
3939 * loop-doloop.c (add_test): Update.
3940 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3941 * lto-streamer-in.c (input_cfg): Update.
3942 (input_function): Update.
3943 * lto-streamer-out.c (output_cfg): Update.
3944 * modulo-sched.c (sms_schedule): Update.
3945 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3946 * predict.c (maybe_hot_edge_p): Update.
3947 (unlikely_executed_edge_p): Update.
3948 (probably_never_executed_edge_p): Update.
3949 (dump_prediction): Update.
3950 (drop_profile): Update.
3951 (propagate_unlikely_bbs_forward): Update.
3952 (determine_unlikely_bbs): Update.
3953 (force_edge_cold): Update.
3954 * profile.c (compute_branch_probabilities): Update.
3955 * reg-stack.c (better_edge): Update.
3956 * shrink-wrap.c (handle_simple_exit): Update.
3957 * tracer.c (better_p): Update.
3958 * trans-mem.c (expand_transaction): Update.
3959 (split_bb_make_tm_edge): Update.
3960 * tree-call-cdce.c: Update.
3961 * tree-cfg.c (gimple_find_sub_bbs): Update.
3962 (gimple_split_edge): Update.
3963 (gimple_duplicate_sese_region): Update.
3964 (gimple_duplicate_sese_tail): Update.
3965 (gimple_flow_call_edges_add): Update.
3966 (insert_cond_bb): Update.
3967 (execute_fixup_cfg): Update.
3968 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3969 * tree-complex.c (expand_complex_div_wide): Update.
3970 * tree-eh.c (lower_resx): Update.
3971 (unsplit_eh): Update.
3972 (cleanup_empty_eh_move_lp): Update.
3973 * tree-inline.c (copy_edges_for_bb): Update.
3974 (freqs_to_counts): Update.
3975 (copy_cfg_body): Update.
3976 * tree-ssa-dce.c (remove_dead_stmt): Update.
3977 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3978 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3979 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3980 (unloop_loops): Update.
3981 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3982 * tree-ssa-loop-split.c (connect_loops): Update.
3983 (split_loop): Update.
3984 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3985 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3986 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3987 * tree-ssa-reassoc.c (branch_fixup): Update.
3988 * tree-ssa-tail-merge.c (replace_block_by): Update.
3989 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3990 (compute_path_counts): Update.
3991 (update_profile): Update.
3992 (recompute_probabilities): Update.
3993 (update_joiner_offpath_counts): Update.
3994 (estimated_freqs_path): Update.
3995 (freqs_to_counts_path): Update.
3996 (clear_counts_path): Update.
3997 (ssa_fix_duplicate_block_edges): Update.
3998 (duplicate_thread_path): Update.
3999 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4000 (case_bit_test_cmp): Update.
4001 (collect_switch_conv_info): Update.
4002 (gen_inbound_check): Update.
4003 (do_jump_if_equal): Update.
4004 (emit_cmp_and_jump_insns): Update.
4005 * tree-tailcall.c (decrease_profile): Update.
4006 (eliminate_tail_call): Update.
4007 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4008 (vect_do_peeling): Update.
4009 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4010 * ubsan.c (ubsan_expand_null_ifn): Update.
4011 (ubsan_expand_ptr_ifn): Update.
4012 * value-prof.c (gimple_divmod_fixed_value): Update.
4013 (gimple_mod_pow2): Update.
4014 (gimple_mod_subtract): Update.
4015 (gimple_ic): Update.
4016 (gimple_stringop_fixed_value): Update.
4018 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4021 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4023 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4025 PR rtl-optimization/82395
4026 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4027 based on non_spilled_static_chain_regno_p.
4029 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4031 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4032 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4033 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4035 2017-10-19 Martin Sebor <msebor@redhat.com>
4037 PR tree-optimization/82596
4038 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4040 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4042 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4043 (get_mem_refs_of_builtin_call): Likewise.
4044 * builtins.c (expand_builtin_apply): Adjust call to
4045 allocate_dynamic_stack_space.
4046 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4047 the third argument to allocate_dynamic_stack_space, otherwise -1.
4048 (expand_builtin): Deal with all alloca variants.
4049 (is_inexpensive_builtin): Likewise.
4050 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4051 * calls.c (special_function_p): Deal with all alloca variants.
4052 (initialize_argument_information): Adjust call to
4053 allocate_dynamic_stack_space.
4054 (expand_call): Likewise.
4055 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4056 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4057 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4058 use it for the stack usage computation.
4059 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4060 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4061 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4062 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4063 (in_loop_p): Remove first argument and useless check.
4064 (pass_walloca::execute): Remove useless test and adjust call to above.
4065 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4066 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4067 (gimplify_call_expr): Deal with all alloca variants.
4068 * hsa-gen.c (gen_hsa_alloca): Likewise.
4069 (gen_hsa_insns_for_call): Likewise.
4070 * ipa-pure-const.c (special_builtin_state): Likewise.
4071 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4072 * tree-object-size.c (alloc_object_size): Likewise.
4073 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4074 (call_may_clobber_ref_p_1): Likewise.
4075 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4076 (ccp_fold_stmt): Likewise.
4077 (optimize_stack_restore): Likewise.
4078 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4079 (mark_all_reaching_defs_necessary_1): Likewise.
4080 (propagate_necessity): Likewise.
4081 (eliminate_unnecessary_stmts): Likewise.
4082 * tree.c (build_common_builtin_nodes): Build
4083 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4084 (build_alloca_call_expr): New function.
4085 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4086 (CASE_BUILT_IN_ALLOCA): Likewise.
4087 (build_alloca_call_expr): Declare.
4088 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4090 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4093 * dwarf2out.c (new_die_raw): New static inline function.
4094 (new_die): Use it to create the DIE.
4095 (add_AT_external_die_ref): Likewise.
4096 (clone_die): Likewise.
4097 (clone_as_declaration): Likewise.
4098 (dwarf2out_vms_debug_main_pointer): Likewise.
4099 (base_type_die): Likewise. Remove early return for corner cases.
4100 Do not call add_pubtype on the DIE here.
4101 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4102 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4103 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4104 native order exists for base types, attach the DIE manually and call
4105 add_pubtype on it. Do not equate a reverse order DIE to the type.
4107 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4109 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4110 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4112 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4114 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4116 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4117 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4118 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4119 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4120 * opts.c (sanitizer_opts): Add builtin.
4121 * ubsan.c (instrument_builtin): New function.
4122 (pass_ubsan::execute): Call it.
4123 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4124 * doc/invoke.texi: Document -fsanitize=builtin.
4126 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4127 builtins, store max (log2 (align), 0) into uchar field instead of
4128 align into uptr field.
4129 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4130 store uchar 0 field instead of uptr 0 field.
4131 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4132 instead of passing one address of struct with 2 locations pass
4133 two addresses of structs with 1 location each.
4134 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4135 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4136 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4137 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4138 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4139 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4140 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4141 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4143 2017-10-19 Martin Liska <mliska@suse.cz>
4146 * file-find.c (remove_prefix): Remove.
4147 * file-find.h (remove_prefix): Likewise.
4148 * gcc-ar.c: Remove smartness of lookup.
4150 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4152 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4153 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4154 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4156 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4159 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4160 (setcc + and to xor + setcc): New peephole2.
4162 2017-10-19 Tom de Vries <tom@codesourcery.com>
4164 * doc/sourcebuild.texi (Test Directives, Variants of
4165 dg-require-support): Add dg-require-stack-size.
4167 2017-10-19 Martin Liska <mliska@suse.cz>
4170 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4171 that have a large alignment.
4172 (gimplify_target_expr): Likewise.
4174 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4176 PR rtl-optimization/82602
4177 * ira.c (rtx_moveable_p): Return false for volatile asm.
4179 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4182 * config/i386/i386-modes.def (CCGZ): New CC mode.
4183 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4184 * config/i386/predicates.md (ix86_comparison_operator):
4186 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4187 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4188 with double-word subtraction.
4189 (put_condition_code): Handle CCGZmode.
4191 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4193 * wide-int.cc (debug (const wide_int &)): New.
4194 (debug (const wide_int *)): New.
4195 (debug (const widest_int &)): New.
4196 (debug (const widest_int *)): New.
4198 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4201 * lra-constraints.c (curr_insn_transform): Use non-input operand
4202 instead of output one for matched reload.
4204 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4206 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4207 (tree-ssa-loop-ivopts.h): New header file.
4208 (struct builtin_info): New fields.
4209 (classify_builtin_1): Compute and record base and offset parts for
4210 memset builtin partition by calling strip_offset.
4211 (offset_cmp, fuse_memset_builtins): New functions.
4212 (finalize_partitions): Fuse adjacent memset partitions by calling
4214 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4215 Expose the interface.
4216 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4218 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4220 PR tree-optimization/82574
4221 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4222 that data reference must be executed exactly once per iteration
4223 against the outermost loop in nest.
4224 (classify_partition): Update call to above function.
4226 2017-10-18 Richard Biener <rguenther@suse.de>
4228 PR tree-optimization/82591
4229 * graphite.c (graphite_transform_loops): Move code gen message
4231 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4232 Here. Handle scop_to_isl_ast failing.
4233 (scop_to_isl_ast): Limit the number of ISL operations.
4235 2017-10-18 Richard Biener <rguenther@suse.de>
4237 * graphite-isl-ast-to-gimple.c
4238 (translate_isl_ast_to_gimple::set_rename): Simplify.
4239 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4240 (graphite_copy_stmts_from_block): ... here.
4241 (copy_bb_and_scalar_dependences): Simplify.
4242 (add_parameters_to_ivs_params): Canonicalize.
4243 (generate_entry_out_of_ssa_copies): Simplify.
4244 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4245 by passing in ISL dimension.
4246 (parameter_index_in_region_1): Rename to ...
4247 (parameter_index_in_region): ... this.
4248 (extract_affine): Adjust assert, pass down parameter index.
4249 (add_param_constraints): Use range-info when available.
4250 (build_scop_context): Adjust.
4251 * sese.c (new_sese_info): Adjust.
4252 (free_sese_info): Likewise.
4253 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4254 Remove unused typedefs.
4255 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4257 2017-10-18 Martin Liska <mliska@suse.cz>
4259 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4261 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4263 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4264 (s390_sched_init): Do not reset s390_sched_state if we entered the
4265 current basic block via a fallthru edge and all others are unlikely.
4267 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4269 * config/s390/s390.c (NUM_SIDES): New variable.
4270 (LONGRUNNING_THRESHOLD): New variable.
4271 (LATENCY_FACTOR): New variable.
4272 (s390_sched_score): Decrease score for long-running instructions on
4274 (s390_sched_variable_issue): Perform bookkeeping for long-running
4277 2017-10-18 Richard Biener <rguenther@suse.de>
4279 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4280 Simplify with removal of the parameter rename map.
4281 (set_rename): Likewise.
4282 (should_copy_to_new_region): Likewise.
4283 (graphite_copy_stmts_from_block): Likewise.
4284 (copy_bb_and_scalar_dependences): Remove initialization of
4285 unused copied_bb_map.
4287 (copy_internal_parameters): Likewise.
4288 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4289 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4290 Use INTEGRAL_TYPE_P.
4291 (parameter_index_in_region_1): Rename to ...
4292 (assign_parameter_index_in_region): ... this. Assert we have
4293 a parameter we handle.
4294 (scan_tree_for_params): Adjust.
4295 * sese.h (parameter_rename_map_t): Remove.
4296 (struct sese_info_t): Remove unused parameter_rename_map and
4297 copied_bb_map members.
4298 * sese.c (new_sese_info): Adjust.
4299 (free_sese_info): Likewise.
4301 2017-10-18 Martin Liska <mliska@suse.cz>
4304 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4305 on an abnormal edge.
4307 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4309 * doc/invoke.texi (ffunction-sections and fdata-sections):
4312 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4314 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4315 the use statement can throw internally.
4317 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4319 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4320 any RTX present on the RHS of a SET.
4321 * compare-elim.c (try_eliminate_compare): Restore comment.
4323 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4325 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4326 may be also called on tcc_type nodes.
4327 * langhooks.c (lhd_tree_size): Likewise.
4329 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4331 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4332 format_warning_at_substring.
4333 (maybe_warn): Convert source_range * param to a location_t. Pass
4334 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4335 (format_directive): Remove code to extract source_ranges and
4336 source_range * in favor of just a location_t.
4337 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4339 * substring-locations.c (format_warning_va): Convert
4340 source_range * param to a location_t.
4341 (format_warning_at_substring): Likewise.
4342 * substring-locations.h (format_warning_va): Likewise.
4343 (format_warning_at_substring): Likewise.
4345 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4347 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4349 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4350 and scatter/gather ops.
4352 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4353 vec_gather_load and vec_scatter_store.
4354 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4355 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4356 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4357 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4358 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4359 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4361 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4363 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4364 Detect FTST instruction and handle its register pops. Only pop
4365 second operand if can_pop_second_op is true.
4366 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4367 set can_pop_second_op to false in the compare_for_stack_reg call.
4369 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4370 output_fp_compare for stack register operands.
4371 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4372 instructions here. Do not emit stack register pops here. Assert
4373 that FCOMPP pops next to top stack register. Rewrite function.
4375 2017-10-17 Nathan Sidwell <nathan@acm.org>
4378 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4379 use DECL_ASSEMBLER_NAME_RAW.
4382 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4385 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4386 Wilco Dijkstra <wilco.dijkstra@arm.com>
4388 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4389 converstion to Pmode from the buf_addr.
4391 2017-10-17 Richard Biener <rguenther@suse.de>
4393 * graphite-dependences.c (scop_get_reads_and_writes): Change
4394 output parameters to references.
4396 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4398 PR 71026/tree-optimization
4399 * fold-const.c (distribute_real_division): Removed.
4400 (fold_binary_loc): Remove calls to distribute_real_divison.
4402 2017-10-17 Richard Biener <rguenther@suse.de>
4404 * graphite-scop-detection.c
4405 (scop_detection::stmt_has_simple_data_refs_p): Always use
4406 the full nest as region.
4407 (try_generate_gimple_bb): Likewise.
4408 * sese.c (scalar_evolution_in_region): Simplify now that
4409 SCEV can handle instantiation in regions.
4410 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4411 in the non-loop part of a function if requested.
4413 2017-10-17 Richard Biener <rguenther@suse.de>
4415 PR tree-optimization/82563
4416 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4418 (graphite_regenerate_ast_isl): Call it.
4419 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4421 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4423 PR tree-optimization/82549
4424 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4425 Formatting fixes. Instead of calling make_bit_field_ref with negative
4428 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4430 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4431 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4432 _mm_maskz_reduce_ss): New.
4433 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4434 __builtin_ia32_reducess_mask): Ditto..
4435 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4436 * config/i386/sse.md (reduces<mode>): Renamed to ...
4437 (reduces<mode><mask_scalar_name>): ... this.
4438 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4440 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4441 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4443 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4445 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4446 * selftest-run-tests.c (selftest::run_tests): Call
4447 selftest::unique_ptr_tests_cc_tests.
4448 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4449 * unique-ptr-tests.cc: New file.
4451 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4454 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4456 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4457 (make_hard_regno_born, make_hard_regno_dead): Update
4458 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4460 2017-10-16 Jeff Law <law@redhat.com>
4462 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4464 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4466 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4468 2017-10-16 Olivier Hainque <hainque@adacore.com>
4470 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4471 with_cpu if we were configured for an e500v2 target cpu name.
4473 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4475 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4476 * doc/invoke.texi: Document +nodsp as a valid extension for
4479 2017-10-16 Martin Liska <mliska@suse.cz>
4481 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4482 (test_set_range): Likewise.
4483 (test_range_functions): Rename to ...
4484 (test_bit_in_range): ... this.
4485 (sbitmap_c_tests): Add new test.
4487 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4489 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4491 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4492 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4494 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4496 * config/aarch64/aarch64-builtins.c
4497 (aarch64_types_quadopu_lane_qualifiers): New.
4498 (TYPES_QUADOPU_LANE): New.
4499 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4500 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4501 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4502 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4503 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4504 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4505 (Vdottype, DOTPROD): New.
4506 (sur): Add SDOT and UDOT.
4508 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4510 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4511 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4512 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4514 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4515 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4516 Enable TARGET_DOTPROD.
4517 (cortex-a75.cortex-a55): Likewise.
4518 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4520 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4522 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4523 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4525 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4527 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4528 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4529 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4530 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4531 * config/arm/types.md (neon_dot, neon_dot_q): New.
4532 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4534 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4536 * config/arm/arm.h (TARGET_DOTPROD): New.
4537 * config/arm/arm.c (arm_arch_dotprod): New.
4538 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4539 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4540 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4541 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4542 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4543 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4544 * doc/invoke.texi (armv8.2-a): Document dotprod
4546 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4548 * i386.c (ix86_vec_cost): New function.
4549 (ix86_rtx_costs): Handle vector operations better.
4550 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4551 * x86-tune-costs.h: Add new costs to all tables.
4553 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4555 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4557 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4558 divsd, sqrtss and sqrtsd
4559 * x86-tune-costs.h: Add new entries to all costs.
4560 (znver1_cost): Fix to match real instruction latencies.
4562 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4563 Michael Collison <michael.collison@arm.com>
4565 * compare-elim.c: Include emit-rtl.h.
4566 (can_merge_compare_into_arith): New function.
4567 (try_validate_parallel): Likewise.
4568 (try_merge_compare): Likewise.
4569 (try_eliminate_compare): Call the above when no previous clobber
4571 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4574 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4578 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4579 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4583 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4584 to be any operand_equal_p operands. For & (B - 1) require
4585 B to be power of 2. Recognize
4586 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4588 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4591 * optabs.c (expand_memory_blockage): Fix call of
4592 targetm.have_memory_blockage.
4594 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4597 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4598 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4599 objects to extra_objs instead of overwriting it.
4601 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4603 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4604 Use any_fp_register_operand as operand[3] predicate. Simplify
4605 equality test for operands[2] and operands[4] memory location.
4606 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4607 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4608 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4609 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4610 any_fp_register_operand as operand[1] predicate. Simplify
4611 equality test for operands[0] and operands[3] memory location.
4612 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4613 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4614 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4616 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4618 * target-insns.def: Add memory_blockage.
4619 * optabs.c (expand_memory_blockage): New function.
4620 (expand_asm_memory_barrier): Rename ...
4621 (expand_asm_memory_blockage): ... to this.
4622 (expand_mem_thread_fence): Call expand_memory_blockage
4623 instead of expand_asm_memory_barrier.
4624 (expand_mem_singnal_fence): Ditto.
4625 (expand_atomic_load): Ditto.
4626 (expand_atomic_store): Ditto.
4627 * doc/md.texi (Standard Pattern Names For Generation):
4628 Document memory_blockage instruction pattern.
4630 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4632 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4633 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4634 * config/rl78/rl78.md: New define_expand "adddi3".
4636 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4638 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4641 2017-10-13 Jeff Law <law@redhat.com>
4643 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4645 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4648 * internal-fn.c (expand_mul_overflow): If both operands have
4649 the same highpart of -1 or 0 and the topmost bit of lowpart
4650 is different, overflow is if res <= 0 rather than res < 0.
4652 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4654 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4655 TARGET_P9_VECTOR code for unaligned_load case.
4657 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4659 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4661 2017-10-13 Nathan Sidwell <nathan@acm.org>
4663 * tree-core.h (tree_contains_struct): Make bool.
4664 * tree.c (tree_contains_struct): Likewise.
4665 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4666 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4667 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4668 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4669 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4671 2017-10-13 Richard Biener <rguenther@suse.de>
4673 * graphite-isl-ast-to-gimple.c
4674 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4675 parameters and dominance check.
4676 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4677 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4678 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4679 Do not update SSA form here or do intermediate IL verification.
4680 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4681 (graphite_initialize): Remove check on the number of loops in
4682 the function and inline into graphite_transform_loops.
4683 (graphite_finalize): Inline into graphite_transform_loops.
4684 (graphite_transform_loops): Perform SSA update and IL verification
4686 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4688 2017-10-13 Richard Biener <rguenther@suse.de>
4690 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4691 graphite_expression_type_precision): Avoid global constructor
4693 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4694 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4695 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4696 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4697 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4698 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4700 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4703 * config/i386/i386.h (ix86_red_zone_size): New.
4704 * config/i386/i386.md (push peephole2s): Replace
4705 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4707 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4708 Alan Hayward <alan.hayward@arm.com>
4709 David Sherwood <david.sherwood@arm.com>
4711 * combine.c (can_change_dest_mode): Reject changes in
4712 REGMODE_NATURAL_SIZE.
4714 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4715 Alan Hayward <alan.hayward@arm.com>
4716 David Sherwood <david.sherwood@arm.com>
4718 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4719 (expand_debug_source_expr): Likewise.
4720 * combine.c (combine_simplify_rtx): Likewise.
4721 * cse.c (fold_rtx): Likewise.
4722 * fwprop.c (canonicalize_address): Likewise.
4723 * targhooks.c (default_shift_truncation_mask): Likewise.
4725 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4726 Alan Hayward <alan.hayward@arm.com>
4727 David Sherwood <david.sherwood@arm.com>
4729 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4730 (widened_mode): Likewise.
4731 (expand_unop): Likewise.
4732 * ree.c (transform_ifelse): Likewise.
4733 (merge_def_and_ext): Likewise.
4734 (combine_reaching_defs): Likewise.
4735 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4737 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4738 Alan Hayward <alan.hayward@arm.com>
4739 David Sherwood <david.sherwood@arm.com>
4741 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4742 * combine.c (gen_lowpart_for_combine): Likewise.
4743 * dwarf2out.c (rtl_for_decl_location): Likewise.
4744 * final.c (alter_subreg): Likewise.
4745 * rtlhooks.c (gen_lowpart_general): Likewise.
4746 (gen_lowpart_if_possible): Likewise.
4748 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4749 Alan Hayward <alan.hayward@arm.com>
4750 David Sherwood <david.sherwood@arm.com>
4752 * calls.c (expand_call): Use subreg_lowpart_offset.
4753 * cse.c (cse_insn): Likewise.
4754 * regcprop.c (copy_value): Likewise.
4755 (copyprop_hardreg_forward_1): Likewise.
4757 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4760 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4761 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4762 =Q constraints to +Q and into insn condition add check
4763 that operands[0] and operands[1] are equal.
4764 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4765 =Q constraints to +Q and into insn condition add check
4766 that operands[0] is equal to either operands[1] or operands[2].
4769 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4770 instead of handling MINUS_EXPR twice (once for each argument),
4771 canonicalize operand order and handle just once, use rtype where
4772 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4775 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4776 any values of __C while still being pattern recognizable as a simple
4779 2017-10-13 Richard Biener <rguenther@suse.de>
4781 PR tree-optimization/82451
4783 2017-10-02 Richard Biener <rguenther@suse.de>
4785 PR tree-optimization/82355
4786 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4787 a mapping for the enclosing loop but avoid generating one for
4789 (copy_bb_and_scalar_dependences): Remove premature codegen
4790 error on PHIs in blocks duplicated into multiple places.
4791 * graphite-scop-detection.c
4792 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4793 in the region use it as loop and nest to analyze the DR in.
4794 (try_generate_gimple_bb): Likewise.
4795 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4796 (add_loop_constraints): For blocks in a loop not in the region
4797 create a dimension with a single iteration.
4798 * sese.h (gbb_loop_at_index): Remove assert.
4800 * cfgloop.c (loop_preheader_edge): For the loop tree root
4801 return the single successor of the entry block.
4802 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4803 Reset the SCEV hashtable and niters.
4804 * graphite-scop-detection.c
4805 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4806 assert that we only have POLYNOMIAL_CHREC that vary in loops
4807 contained in the region.
4808 (scop_detection::graphite_can_represent_expr): Adjust.
4809 (scop_detection::stmt_has_simple_data_refs_p): For loops
4810 not in the region set loop to NULL. The nest is now the
4811 entry edge to the region.
4812 (try_generate_gimple_bb): Likewise.
4813 * sese.c (scalar_evolution_in_region): Adjust for
4814 instantiate_scev change.
4815 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4816 Make nest parameter the edge into the region.
4817 (create_data_ref): Likewise.
4818 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4819 entry edge into a region and adjust instantiate_scev calls.
4820 (create_data_ref): Likewise.
4821 (graphite_find_data_references_in_stmt): Likewise.
4822 (find_data_references_in_stmt): Pass the loop preheader edge
4823 from the nest argument.
4824 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4825 parameter the edge into the region.
4826 (instantiate_parameters): Use the loop preheader edge as entry.
4827 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4829 (get_instantiated_value_entry): Make instantiate_below parameter
4830 the edge into the region.
4831 (instantiate_scev_name): Likewise. Adjust dominance checks,
4832 when we cannot use loop-based instantiation instantiate by
4833 walking use-def chains.
4834 (instantiate_scev_poly): Adjust.
4835 (instantiate_scev_binary): Likewise.
4836 (instantiate_scev_convert): Likewise.
4837 (instantiate_scev_not): Likewise.
4838 (instantiate_array_ref): Remove.
4839 (instantiate_scev_3): Likewise.
4840 (instantiate_scev_2): Likewise.
4841 (instantiate_scev_1): Likewise.
4842 (instantiate_scev_r): Do not blindly handle N-operand trees.
4843 Do not instantiate array-refs. Handle all constants and invariants.
4844 (instantiate_scev): Make instantiate_below parameter
4845 the edge into the region.
4846 (resolve_mixers): Use the loop preheader edge for the region
4847 parameter to instantiate_scev_r.
4848 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4850 2017-10-13 Richard Biener <rguenther@suse.de>
4852 PR tree-optimization/82525
4853 * graphite-isl-ast-to-gimple.c
4854 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4856 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4857 Fail code generation when we cannot represent the isl integer.
4858 (binary_op_to_tree): Elide modulo operations that are no-ops
4859 in the type we code generate. Remove now superfluous code
4862 2017-10-13 Richard Biener <rguenther@suse.de>
4864 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4865 (scop_detection::harmful_loop_in_region): Remove premature
4866 IV type restriction.
4867 (scop_detection::graphite_can_represent_scev): We can handle
4868 pointer IVs just fine.
4870 2017-10-13 Alan Modra <amodra@gmail.com>
4872 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4873 "Clobbers and Scratch Registers". Add paragraph on
4874 alternative to clobbers for scratch registers and OpenBLAS
4877 2017-10-13 Alan Modra <amodra@gmail.com>
4879 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4880 example of a memory input for a string of known length. Move
4881 commentary out of table. Add a number of new examples
4882 covering array memory inputs.
4884 2017-10-12 Martin Liska <mliska@suse.cz>
4886 PR tree-optimization/82493
4887 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4888 (test_range_functions): New function.
4889 (sbitmap_c_tests): Likewise.
4890 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4891 * selftest.h (sbitmap_c_tests): New function.
4893 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4895 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4897 * config/rs6000/amo.h: Fix spacing issue.
4899 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4902 * config/i386/i386.md (*ashl<mode>3_mask_1,
4903 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4904 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4907 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4909 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4910 (profile_probability): Set max_probability
4911 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4913 * profile-count.c (profile_probability::differs_from_p): Do not
4914 rely on max_probaiblity == 10000
4916 2017-10-12 Jeff Law <law@redhat.com>
4918 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4921 2017-10-12 Martin Sebor <msebor@redhat.com>
4925 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4926 (handle_alias_pairs): Call it.
4927 * common.opt (-Wattribute-alias): New option.
4928 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4929 * doc/invoke.texi (-Wattribute-alias): Document.
4931 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4934 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4936 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4938 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4939 (make_hard_regno_born, make_hard_regno_dead): Update
4940 bb_killed_pseudos and bb_gen_pseudos.
4942 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4944 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4946 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4948 * config/alpha/alpha.c (alpha_split_conditional_move):
4949 Use std::swap instead of manually swapping.
4950 (alpha_stdarg_optimize_hook): Ditto.
4951 (alpha_canonicalize_comparison): Ditto.
4953 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4955 * tree-loop-distribution.c (struct builtin_info): New struct.
4956 (struct partition): Refactor fields into struct builtin_info.
4957 (partition_free): Free struct builtin_info.
4958 (build_size_arg_loc, build_addr_arg_loc): Delete.
4959 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4960 information from struct builtin_info.
4961 (find_single_drs): New function refactored from classify_partition.
4962 Also moved builtin validity checks to this function.
4963 (compute_access_range, alloc_builtin): New functions.
4964 (classify_builtin_st, classify_builtin_ldst): New functions.
4965 (classify_partition): Refactor code into functions find_single_drs,
4966 classify_builtin_st and classify_builtin_ldst.
4967 (distribute_loop): Don't do runtime alias check when distributing
4969 (find_seed_stmts_for_distribution): New function.
4970 (pass_loop_distribution::execute): Refactor code finding seed
4971 stmts into above function. Support distribution for the innermost
4972 two-level loop nest. Adjust dump information.
4974 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4976 * tree-loop-distribution.c: Adjust the general comment.
4977 (NUM_PARTITION_THRESHOLD): New macro.
4978 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4979 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4980 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4981 in call to build_partition_graph.
4982 (finalize_partitions): New parameter. Make loop distribution more
4983 conservative by fusing more partitions.
4984 (distribute_loop): Don't do runtime alias check in case of loop nest
4986 (find_seed_stmts_for_distribution): New function.
4987 (prepare_perfect_loop_nest): New function.
4988 (pass_loop_distribution::execute): Refactor code finding seed stmts
4989 and loop nest into above functions. Support loop nest distribution.
4990 Adjust dump information accordingly.
4992 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4994 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4995 and set PTYPE_SEQUENTIAL for merged partition.
4997 2017-10-12 Richard Biener <rguenther@suse.de>
4999 PR tree-optimization/69728
5001 2017-09-19 Richard Biener <rguenther@suse.de>
5003 PR tree-optimization/69728
5004 * graphite-sese-to-poly.c (schedule_error): New global.
5005 (add_loop_schedule): Handle empty domain by failing the
5007 (build_original_schedule): Handle schedule_error.
5009 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5010 domain by returning an unchanged schedule.
5012 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5014 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5015 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5017 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5019 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5022 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5025 * expr.c (store_field): Don't optimize away bitsize == 0 store
5026 from CALL_EXPR with addressable return type.
5028 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5030 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5031 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5032 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5033 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5034 TARGET_ISEL instead of TARGET_ISEL<sel>.
5036 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5038 * config/rs6000/rs6000.c
5039 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5041 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5043 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5045 (reg_or_cint_operand): Fix comment.
5046 (reg_or_zero_operand): New predicate.
5047 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5048 * config/rs6000/rs6000.c (output_isel): Delete.
5049 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5050 instead of reg_or_cint_operand. Output instruction directly (not via
5052 (isel_unsigned_<mode>): Ditto.
5053 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5054 gpc_reg_operand. Add an instruction alternative for this. Output
5055 instruction directly.
5056 (*isel_reversed_unsigned_<mode>): Ditto.
5058 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5060 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5061 (TARGET_CANONICALIZE_COMPARISON): Define.
5063 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5066 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5067 Check whether the dest is REG before adding REG_EQUIV note.
5069 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5072 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5074 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5075 (make_hard_regno_born, make_hard_regno_dead): Update
5076 bb_killed_pseudos and bb_gen_pseudos.
5078 2017-10-11 Nathan Sidwell <nathan@acm.org>
5080 * incpath.h (enum incpath_kind): Name enum, prefix values.
5081 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5082 * incpath.c (heads, tails): Use INC_MAX.
5083 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5084 (merge_include_chains, split_quote_chain,
5085 register_include_chains): Update incpath_kind names.
5086 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5087 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5089 (add_framework_path, darwin_register_objc_includes): Likewise.
5090 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5092 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5094 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5095 Do not use float_operator operator predicate.
5096 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5097 * config/i386/predicates.md (float_operator): Remove predicate.
5099 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5101 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5102 (*jccxf_i387): Ditto.
5103 (*jcc<mode>_i387): Ditto.
5104 (*jccu<mode>_i387): Ditto.
5105 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5106 (*jcc_*_i387 splitters): Remove.
5107 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5108 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5109 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5112 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5114 * profile-count.h (slow_safe_scale_64bit): New function.
5115 (safe_scale_64bit): New inline.
5116 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5117 * profile-count.c: Include wide-int.h
5118 (slow_safe_scale_64bit): New.
5120 2017-10-11 Nathan Sidwell <nathan@acm.org>
5122 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5123 HAS_DECL_ASSEMBLER_NAME_P.
5124 * gimple-expr.c (gimple_decl_printable_name: Check
5125 HAS_DECL_ASSEMBLER_NAME_P too.
5126 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5127 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5128 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5129 * passes.c (rest_of_decl_compilation): Check
5130 HAS_DECL_ASSEMBLER_NAME_P too.
5131 * recog.c (verify_changes): Likewise.
5132 * tree-pretty-print.c (dump_decl_name): Likewise.
5133 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5135 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5136 (SET_DECL_ASSEMBLER_NAME): Use it.
5137 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5138 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5139 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5141 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5143 * config.gcc (i386, x86_64): Add extra objects.
5144 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5145 (ix86_min_insn_size): Declare.
5146 (ix86_issue_rate): Declare.
5147 (ix86_adjust_cost): Declare.
5148 (ia32_multipass_dfa_lookahead): Declare.
5149 (ix86_macro_fusion_p): Declare.
5150 (ix86_macro_fusion_pair_p): Declare.
5151 (ix86_bd_has_dispatch): Declare.
5152 (ix86_bd_do_dispatch): Declare.
5153 (ix86_core2i7_init_hooks): Declare.
5154 (ix86_atom_sched_reorder): Declare.
5155 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5156 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5157 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5158 (rip_relative_addr_p): Rename to ...
5159 (ix86_rip_relative_addr_p): ... this one; export.
5160 (memory_address_length): Update.
5161 (ix86_issue_rate): Move to x86-tune-sched.c.
5162 (ix86_flags_dependent): Move to x86-tune-sched.c.
5163 (ix86_agi_dependent): Move to x86-tune-sched.c.
5164 (exact_dependency_1): Move to x86-tune-sched.c.
5165 (exact_store_load_dependency): Move to x86-tune-sched.c.
5166 (ix86_adjust_cost): Move to x86-tune-sched.c.
5167 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5168 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5169 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5170 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5171 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5172 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5173 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5174 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5175 (min_insn_size): Rename to ...
5176 (ix86_min_insn_size): ... this one; export.
5177 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5178 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5179 (core2i7_first_cycle_multipass_backtrack): Move to
5180 x86-tune-sched-core.c.
5181 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5182 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5183 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5184 (ix86_avoid_jump_mispredicts): Update.
5185 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5186 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5187 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5188 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5189 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5190 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5191 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5192 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5193 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5194 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5195 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5196 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5197 (BIG): Move to ix86-tune-sched-bd.c.
5198 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5199 (enum insn_path): Move to ix86-tune-sched-bd.c.
5200 (get_mem_group): Move to ix86-tune-sched-bd.c.
5201 (is_cmp): Move to ix86-tune-sched-bd.c.
5202 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5203 (is_branch): Move to ix86-tune-sched-bd.c.
5204 (is_prefetch): Move to ix86-tune-sched-bd.c.
5205 (init_window): Move to ix86-tune-sched-bd.c.
5206 (allocate_window): Move to ix86-tune-sched-bd.c.
5207 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5208 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5209 (process_end_window): Move to ix86-tune-sched-bd.c.
5210 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5211 (find_constant): Move to ix86-tune-sched-bd.c.
5212 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5213 (has_immediate): Move to ix86-tune-sched-bd.c.
5214 (get_insn_path): Move to ix86-tune-sched-bd.c.
5215 (get_insn_group): Move to ix86-tune-sched-bd.c.
5216 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5217 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5218 (add_insn_window): Move to ix86-tune-sched-bd.c.
5219 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5220 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5221 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5222 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5223 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5224 (do_dispatch): Move to ix86-tune-sched-bd.c.
5225 (has_dispatch): Move to ix86-tune-sched-bd.c.
5226 * config/i386/t-i386: Add new object files.
5227 * config/i386/x86-tune-costs.h: New file.
5228 * config/i386/x86-tune-sched-atom.c: New file.
5229 * config/i386/x86-tune-sched-bd.c: New file.
5230 * config/i386/x86-tune-sched-core.c: New file.
5231 * config/i386/x86-tune-sched.c: New file.
5233 2017-10-11 Liu Hao <lh_mouse@126.com>
5235 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5236 the generic version below instead.
5237 (should_colorize): Recognize Windows consoles as terminals
5239 * pretty-print.c [__MINGW32__] (write_all): New function.
5240 [__MINGW32__] (find_esc_head): Likewise.
5241 [__MINGW32__] (find_esc_terminator): Likewise.
5242 [__MINGW32__] (eat_esc_sequence): Likewise.
5243 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5245 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5248 2017-10-11 Richard Biener <rguenther@suse.de>
5250 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5251 Properly call analyze_scalar_evolution with the loop of the stmt.
5253 2017-10-11 Richard Biener <rguenther@suse.de>
5255 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5256 * tree-core.h (tree_base): Add chrec_var union member.
5257 * tree.h (CHREC_VAR): Remove.
5258 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5259 * tree-chrec.h (build_polynomial_chrec): Adjust.
5260 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5261 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5263 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5265 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5266 * match.pd: ... here.
5267 ((T) X == (T) Y): Relax condition.
5269 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5271 PR tree-optimization/82472
5272 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5274 (break_alias_scc_partitions): Update postorder number.
5276 2017-10-11 Martin Liska <mliska@suse.cz>
5279 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5281 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5283 2017-10-11 Martin Liska <mliska@suse.cz>
5285 * print-rtl.c (print_insn): Move declaration of idbuf
5286 to same scope as name.
5288 2017-10-11 Martin Liska <mliska@suse.cz>
5293 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5296 2017-10-11 Martin Liska <mliska@suse.cz>
5299 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5302 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5304 * config.gcc (default_gnu_indirect_function): Default to yes for
5305 arm*-*-linux* with glibc.
5307 2017-10-11 Richard Biener <rguenther@suse.de>
5309 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5310 default-defs and types we do not want to analyze.
5311 (interpret_loop_phi): Replace unreachable code with an assert.
5312 (compute_scalar_evolution_in_loop): Remove and inline ...
5313 (analyze_scalar_evolution_1): ... here, replacing condition with
5314 what makes the intent clearer. Remove handling of cases
5315 get_scalar_evolution now handles.
5317 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5319 PR rtl-optimization/81434
5320 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5321 comment for main loop. In sched_group_found if, also add checks for
5322 pass and min_cost_group.
5324 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5326 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5327 (rs6000_insn_cost): New function.
5328 * config/rs6000/rs6000.md (cost): New attribute.
5330 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5331 H.J. Lu <hongjiu.lu@intel.com>
5335 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5336 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5337 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5338 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5339 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5340 OPTION_MASK_ISA_MMX and some other bit set require both
5341 mmx and the ISAs without the mmx bit.
5342 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5343 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5344 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5345 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5346 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5347 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5348 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5349 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5350 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5351 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5352 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5353 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5354 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5355 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5356 Add OPTION_MASK_ISA_MMX.
5358 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5360 * config.gcc (armv7*-*-freebsd*): New target.
5361 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5363 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5365 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5366 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5369 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5371 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5372 a template parameter.
5373 (WIDE_INT_REF_FOR): Update accordingly.
5374 * tree.h (wi::int_traits <const_tree>): Delete.
5375 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5376 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5377 (wi::tree_to_wide_ref): New typedef.
5378 (wi::to_wide): New function.
5379 * calls.c (get_size_range): Use wi::to_wide when operating on
5381 * cgraph.c (cgraph_node::create_thunk): Likewise.
5382 * config/i386/i386.c (ix86_data_alignment): Likewise.
5383 (ix86_local_alignment): Likewise.
5384 * dbxout.c (stabstr_O): Likewise.
5385 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5386 * expr.c (const_vector_from_tree): Likewise.
5387 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5388 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5389 (fold_negate_expr_1, int_const_binop_1, const_binop)
5390 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5391 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5392 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5393 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5394 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5395 (fold_not_const, round_up_loc): Likewise.
5396 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5397 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5398 (alloca_call_type): Likewise.
5399 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5400 * godump.c (go_output_typedef): Likewise.
5401 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5402 * internal-fn.c (get_min_precision): Likewise.
5403 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5404 * ipa-polymorphic-call.c
5405 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5406 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5407 (ipa_modify_call_arguments): Likewise.
5408 * match.pd: Likewise.
5409 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5410 * print-tree.c (print_node_brief, print_node): Likewise.
5411 * stmt.c (expand_case): Likewise.
5412 * stor-layout.c (layout_type): Likewise.
5413 * tree-affine.c (tree_to_aff_combination): Likewise.
5414 * tree-cfg.c (group_case_labels_stmt): Likewise.
5415 * tree-data-ref.c (dr_analyze_indices): Likewise.
5416 (prune_runtime_alias_test_list): Likewise.
5417 * tree-dump.c (dequeue_and_dump): Likewise.
5418 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5419 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5420 * tree-pretty-print.c (dump_generic_node): Likewise.
5421 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5422 (simple_iv_with_niters): Likewise.
5423 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5424 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5425 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5426 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5427 (refine_value_range_using_guard, number_of_iterations_ne_max)
5428 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5429 (get_cst_init_from_scev, record_nonwrapping_iv)
5430 (scev_var_range_cant_overflow): Likewise.
5431 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5432 * tree-ssa-pre.c (compute_avail): Likewise.
5433 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5434 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5435 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5436 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5437 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5438 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5439 (dump_case_nodes, try_switch_expansion): Likewise.
5440 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5441 (vect_do_peeling): Likewise.
5442 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5443 * tree-vect-stmts.c (vectorizable_load): Likewise.
5444 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5445 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5446 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5447 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5448 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5449 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5450 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5451 (evrp_dom_walker::before_dom_children): Likewise.
5452 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5453 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5454 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5455 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5456 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5457 (get_range_pos_neg): Likewise.
5458 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5459 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5460 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5461 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5462 * config/avr/avr.c (avr_fold_builtin): Likewise.
5463 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5464 * config/msp430/msp430.c (msp430_attr): Likewise.
5465 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5466 * config/powerpcspe/powerpcspe-c.c
5467 (altivec_resolve_overloaded_builtin): Likewise.
5468 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5469 (rs6000_expand_ternop_builtin): Likewise.
5470 * config/rs6000/rs6000-c.c
5471 (altivec_resolve_overloaded_builtin): Likewise.
5472 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5473 (rs6000_expand_ternop_builtin): Likewise.
5474 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5476 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5478 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5479 when copying loop nest with only one inner loop.
5481 2017-10-10 Richard Biener <rguenther@suse.de>
5483 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5484 blocks if SCEV is active.
5485 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5487 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5488 (scev_initialize): Assert we are not yet initialized.
5490 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5492 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5493 inner loop's exit stmt by making it always exit the loop, otherwise
5494 we would generate an infinite empty loop.
5496 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5498 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5499 renaming variables in new preheader if it's deleted.
5501 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5503 * tree-loop-distribution.c (struct partition): Remove unused field
5504 loops of the structure.
5505 (partition_alloc, partition_free): Ditto.
5506 (build_rdg_partition_for_vertex): Ditto.
5508 2017-10-09 Jeff Law <law@redhat.com>
5510 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5511 return type to match prototype and documentation.
5513 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5515 * config/rs6000/rs6000.c (processor_costs): Move to ...
5516 * config/rs6000/rs6000.h: ... here.
5517 (rs6000_cost): Declare.
5519 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5521 * except.c (setjmp_fn): New global variable.
5522 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5523 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5524 if DONT_USE_BUILTIN_SETJMP is defined.
5526 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5528 * target.def (insn_cost): New hook.
5529 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5530 * doc/tm.texi: Regenerate.
5531 * rtlanal.c (insn_cost): Use the new hook.
5533 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5535 * combine.c (combine_validate_cost): Compute the new insn_cost,
5536 not just pattern_cost.
5537 (try_combine): Adjust comment.
5539 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5541 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5543 * combine.c (uid_insn_cost): Adjust comment.
5544 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5546 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5547 * dse.c (find_shift_sequence): Ditto.
5548 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5549 (bb_valid_for_noce_process_p): Use pattern_cost.
5550 * rtl.h (insn_rtx_cost): Delete.
5551 (pattern_cost): New prototype.
5552 (insn_cost): New prototype.
5553 * rtlanal.c (insn_rtx_cost): Rename to...
5554 (pattern_cost): ... this.
5557 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5559 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5560 (*jcc<mode>_0_r_i387): Ditto.
5561 (*jccxf_r_i387): Ditto.
5562 (*jcc<mode>_r_i387): Ditto.
5563 (*jccu<mode>_r_i387): Ditto.
5564 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5565 (*jcc): Rename from *jcc_1.
5567 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5569 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5570 deferred rescans after the lvx/stvx recombination pre-pass.
5572 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5574 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5575 memory operation instruction support.
5576 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5577 (rs6000-ibm-aix[789]*): Likewise.
5578 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5579 Document new functions.
5581 2017-10-09 Richard Biener <rguenther@suse.de>
5583 PR tree-optimization/82397
5584 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5585 equality only for semantically equal trees.
5587 2017-10-09 Richard Biener <rguenther@suse.de>
5589 PR tree-optimization/82449
5590 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5591 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5592 allow constant addresses.
5593 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5596 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5598 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5601 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5604 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5607 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5610 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5612 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5615 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5616 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5618 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5620 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5621 (WI_BINARY_PREDICATE_RESULT): Likewise.
5622 (wi::binary_traits::operator_result): New type.
5623 (wi::binary_traits::predicate_result): Likewise.
5624 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5625 (generic_wide_int::operator==, generic_wide_int::operator!=)
5626 (generic_wide_int::operator&, generic_wide_int::and_not)
5627 (generic_wide_int::operator|, generic_wide_int::or_not)
5628 (generic_wide_int::operator^, generic_wide_int::operator+
5629 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5631 (operator~, unary operator-, operator==, operator!=, operator&)
5632 (operator|, operator^, operator+, binary operator-, operator*): New
5634 * expr.c (get_inner_reference): Use wi::bit_and_not.
5635 * fold-const.c (fold_binary_loc): Likewise.
5636 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5637 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5638 (bit_value_binop): Likewise.
5639 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5640 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5641 (extract_range_from_binary_expr_1): Likewise.
5642 (masked_increment): Likewise.
5643 (simplify_bit_ops_using_ranges): Likewise.
5645 2017-10-09 Martin Jambor <mjambor@suse.cz>
5648 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5649 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5650 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5652 (hsa_fixup_mov_insn_type): New function.
5653 (hsa_op_with_type::get_in_type): Use it.
5654 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5655 immediates in an assert.
5656 (hsa_op_with_type::extend_int_to_32bit): New method.
5657 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5658 types. Convert to dest type if necessary.
5659 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5660 (reg_for_gimple_ssa): Pass false as min32int to
5661 hsa_type_for_scalar_tree_type.
5662 (gen_hsa_addr): Fixup type when creating addresable temporary.
5663 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5664 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5665 necessary. Call hsa_fixup_mov_insn_type.
5666 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5667 extend operands and convert to dest type if necessary.
5668 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5669 to dest type if necessary.
5670 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5671 if conversion nt necessary and size matches.
5672 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5673 to dest type if necessary.
5674 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5675 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5677 (gen_hsa_clrsb): Likewise.
5678 (gen_hsa_ffs): Likewise.
5679 (gen_hsa_divmod): Extend operands and convert to dest type if
5681 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5683 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5685 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5686 Remove empty default arguments. Use a brace block as output
5688 (conditional return): Ditto.
5690 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5691 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5693 (group_ending_nop): Ditto.
5694 (doloop_end): Ditto.
5695 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5696 (splitters for those): Ditto.
5698 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5700 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5701 a conditional jump (and the compare for it) so that pc_rtx is the
5703 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5704 for the deleted and renamed ctr<mode>_internal[234] patterns.
5705 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5706 Delete second conditional return pattern.
5707 (ctr<mode>_internal2): Delete this second bdnz pattern.
5708 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5709 (ctr<mode>_internal4): Delete this second bdz pattern.
5711 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5713 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5714 (always_initialized_rtx_for_ssa_name_p): New predicate.
5715 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5716 (finish_out_of_ssa): Free new field of SA.
5717 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5718 * tree-ssa-coalesce.c: Include tree-ssa.h.
5719 (get_parm_default_def_partitions): Remove extern keyword.
5720 (get_undefined_value_partitions): New function.
5721 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5722 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5724 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5726 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5728 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5730 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5732 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5733 for targets that preffer 128bit.
5735 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5737 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5739 2017-10-08 Olivier Hainque <hainque@adacore.com>
5741 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5742 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5744 (thumb_set_return_address): Likewise.
5746 2017-10-08 Olivier Hainque <hainque@adacore.com>
5748 * common/config/arm/arm-common.c (arm_except_unwind_info):
5749 Handle DWARF2_UNWIND_INFO.
5751 2017-10-07 Michael Collison <michael.collison@arm.com>
5753 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5756 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5758 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5759 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5760 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5761 defined, force the creation of a new block for a dispatch label.
5763 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5765 * invoke.texi (Wsuggest-attribute=cold): Document.
5766 * common.opt (Wsuggest-attribute=cold): New
5767 * ipa-pure-const.c (warn_function_cold): New function.
5768 * predict.c (compute_function_frequency): Use it.
5769 * predict.h (warn_function_cold): Declare.
5771 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5773 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5776 2017-10-06 Martin Liska <mliska@suse.cz>
5778 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5779 keyword for member functions.
5780 (struct sanopt_tree_couple): New struct.
5781 (struct sanopt_tree_couple_hash): New function.
5782 (struct sanopt_ctx): Add new hash_map.
5783 (has_dominating_ubsan_ptr_check): New function.
5784 (record_ubsan_ptr_check_stmt): Likewise.
5785 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5786 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5787 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5789 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5792 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5793 aarch64_simd_valid_immediate on CONST_VECTORs.
5794 (aarch64_reg_or_bic_imm): Likewise.
5796 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5798 PR rtl-optimization/82396
5799 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5801 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5803 * graphite-dependences.c (scop_get_reads): Move code to...
5804 (scop_get_must_writes): Move code to...
5805 (scop_get_may_writes): Move code to...
5806 (scop_get_reads_and_writes): ... here.
5807 (scop_get_dependences): Call scop_get_reads_and_writes.
5809 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5811 PR tree-optimization/82434
5812 * fold-const.h (can_native_encode_type_p,
5813 can_native_encode_string_p): Remove.
5814 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5815 don't encode anything, just return what would be otherwise returned.
5816 (native_encode_fixed, native_encode_complex, native_encode_vector):
5818 (native_encode_string): Likewise. Inline by hand
5819 can_native_encode_string_p.
5820 (can_native_encode_type_p): Remove.
5821 (can_native_encode_string_p): Remove.
5822 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5823 STRING_CSTs using can_native_encode_string_p, test all
5824 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5825 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5826 argument from native_encode_expr.
5827 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5828 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5831 2017-10-06 Richard Biener <rguenther@suse.de>
5833 PR tree-optimization/82397
5834 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5835 operand_equal_p but rely on data_ref_compare_tree for detecting
5837 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5838 to match up with dr_group_sort_cmp.
5840 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5843 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5845 * config/s390/s390-builtin-types.def: Regenerate.
5847 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5850 * config/s390/s390-builtin-types.def: Regenerate.
5851 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5852 Change flag from B_VXE to B_VX.
5853 (s390_vec_min_dbl): Remove B_VXE flag.
5855 2017-10-06 Richard Biener <rguenther@suse.de>
5857 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5858 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5859 translate_isl_ast_to_gimple::is_valid_rename,
5860 translate_isl_ast_to_gimple::get_rename,
5861 translate_isl_ast_to_gimple::get_def_bb_for_const,
5862 translate_isl_ast_to_gimple::get_new_name,
5863 translate_isl_ast_to_gimple::collect_all_ssa_names,
5864 translate_isl_ast_to_gimple::copy_loop_phi_args,
5865 translate_isl_ast_to_gimple::collect_all_ssa_names,
5866 translate_isl_ast_to_gimple::copy_loop_phi_args,
5867 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5868 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5869 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5870 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5871 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5872 translate_isl_ast_to_gimple::copy_cond_phi_args,
5873 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5874 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5875 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5876 translate_isl_ast_to_gimple::rename_uses,
5877 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5878 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5879 (set_rename_for_each_def): Likewise.
5880 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5881 here. Handle rewriting SCEV analyzable uses here.
5882 (copy_bb_and_scalar_dependences): Generate code for PHI
5884 (graphite_regenerate_ast_isl): Adjust.
5885 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5886 (add_write, add_read): New functions.
5887 (build_cross_bb_scalars_def): Use it and simplify.
5888 (build_cross_bb_scalars_use): Likewise.
5889 (graphite_find_cross_bb_scalar_vars): Inline into...
5890 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5891 simulating out-of-SSA. Compute liveout and add dependencies.
5892 (build_scops): Force an empty entry block.
5893 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5895 (sese_build_liveouts): Declare.
5896 (sese_trivially_empty_bb_p): Likewise.
5897 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5898 compute liveout and debug_liveout.
5899 (sese_bad_liveouts_use): Remove.
5900 (sese_reset_debug_liveouts_bb): Likewise.
5901 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5902 (sese_build_liveouts): Build liveout and debug_liveout and store
5904 (new_sese_info): Adjust.
5905 (free_sese_info): Likewise.
5906 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5907 do not build liveout here.
5908 (move_sese_in_condition): Adjust region entry.
5909 (scev_analyzable_p): Match up with chrec_apply requirements.
5910 (sese_trivially_empty_bb_p): New.
5911 * tree-into-ssa.c (get_reaching_def): Properly support generating
5912 default-defs for incremental rewrite of anonymous names.
5914 2017-10-06 Richard Biener <rguenther@suse.de>
5916 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5917 precision do not perform modulo reduction.
5919 2017-10-06 Richard Biener <rguenther@suse.de>
5921 PR tree-optimization/82436
5922 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5923 conservatively choose the vectorization factor when checking
5924 whether we can perform the required load permutation.
5925 (vect_transform_slp_perm_load): Assert when we may not fail.
5927 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5929 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5930 message for incompatible -msdata=* and -mcall-* options.
5932 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5934 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5935 rate for post-reload scheduling.
5937 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5939 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5941 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5943 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5944 to improve monte carlo in scimark.
5946 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5948 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5949 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5950 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5951 pentium4_cost, nocona_cost): Set reassociation width to 1.
5952 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5953 width to 2 for fp operations and 1 otherwise.
5954 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5956 (atom_cost): Set reassociation width to 2.
5957 (slm_cost, generic_cost): Set fp reassociation width
5958 to 2 and 1 otherwise.
5959 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5960 (core_cost): Set fp reassociation width to 4 and vector to 2.
5961 (ix86_reassociation_width): Rewrite using cost table; special case
5962 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5963 and TARGET_AVX128_OPTIMAL.
5964 * config/i386/i386.h (processor_costs): Add
5965 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5966 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5967 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5968 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5969 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5970 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5972 2017-10-05 Nathan Sidwell <nathan@acm.org>
5974 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5976 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5978 * config/arm/arm.c (arm_test_fpu_data): New.
5979 (arm_run_selftests): Call arm_test_fpu_data.
5981 2017-10-04 Nathan Sidwell <nathan@acm.org>
5983 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5985 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5987 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5989 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5990 check type for aarch64_simd_valid_immediate.
5991 (aarch64_output_simd_mov_immediate): Update prototype.
5992 (aarch64_simd_valid_immediate): Update prototype.
5993 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5994 support for ORR-immediate.
5995 (and<mode>3): modified pattern to add support for BIC-immediate.
5996 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5997 now checks for valid immediate for BIC and ORR based on new enum
5999 (aarch64_output_simd_mov_immediate): Function now used to output
6000 BIC/ORR imm as well based on new enum argument.
6001 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6003 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6004 (aarch64_reg_or_bic_imm): Likewise.
6006 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6008 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6009 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6011 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6015 PR rtl-optimization/82396
6016 * haifa-sched.c (autopref_multipass_init): Simplify
6018 (autopref_rank_data): Simplify sort order.
6019 * sched-int.h (autopref_multipass_data_): Remove
6020 multi_mem_insn_p, min_offset and max_offset.
6022 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6024 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6026 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6028 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6031 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6033 * doc/sourcebuild.texi: Document vect_long_mult.
6035 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6037 PR tree-optimization/82413
6038 * fold-const.c (build_range_check): Use widest_int when comparing
6039 the maximum ETYPE value with HIGH.
6041 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6043 PR rtl-optimization/82396
6044 * haifa-sched.c (autopref_multipass_init): Simplify
6046 (autopref_rank_data): Simplify sort order.
6047 * sched-int.h (autopref_multipass_data_): Remove
6048 multi_mem_insn_p, min_offset and max_offset.
6050 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6052 PR tree-optimization/82381
6053 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6054 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6057 PR tree-optimization/82374
6058 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6059 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6060 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6061 current_function_decl to the new decl.
6063 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6065 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6066 helper macro for IEEE float128 hardware built-in functions.
6067 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6069 (TRUNCF128_ODD): Likewise.
6070 (ADDF128_ODD): Likewise.
6071 (SUBF128_ODD): Likewise.
6072 (MULF128_ODD): Likewise.
6073 (DIVF128_ODD): Likewise.
6074 (FMAF128_ODD): Likewise.
6075 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6076 UNSPEC_TRUNC_ROUND_TO_ODD.
6077 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6078 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6079 floating point round to odd instructions.
6080 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6081 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6082 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6083 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6084 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6085 (trunc<mode>sf2_hw): Change the truncate with round to odd
6086 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6087 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6088 to odd hardware instructions.
6089 (sub<mode>3_odd): Likewise.
6090 (mul<mode>3_odd): Likewise.
6091 (div<mode>3_odd): Likewise.
6092 (sqrt<mode>2_odd): Likewise.
6093 (fma<mode>4_odd): Likewise.
6094 (fms<mode>4_odd): Likewise.
6095 (nfma<mode>4_odd): Likewise.
6096 (nfms<mode>4_odd): Likewise.
6097 (trunc<mode>df2_odd): Change the truncate with round to odd
6098 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6100 * doc/extend.texi (PowerPC built-in functions): Update documentation
6101 for existing IEEE float128-bit built-in functions. Add built-in
6102 functions that generate the IEEE 128-bit floating point round to
6105 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6107 PR rtl-optimization/77729
6108 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6109 to (X&(C1&~C2))|C2 transformations.
6111 2017-10-03 Martin Jambor <mjambor@suse.cz>
6113 PR tree-optimization/82363
6114 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6115 mismatch, mark lacc written regardless of racc.
6117 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6119 PR tree-optimization/82381
6120 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6121 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6122 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6123 fallthrough into reassoc_stmt_dominates_stmt_p.
6126 * combine.c (combine_instructions): Don't combine in unreachable
6129 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6132 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6133 function to not use the have_cpu variable. Do not set cpu_index,
6134 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6136 (rs6000_valid_attribute_p): Remove duplicate initializations of
6137 old_optimize and func_optimize.
6138 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6139 (rs6000_activate_target_options): Make global.
6140 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6143 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6145 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6146 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6147 Set *poffset to 0 and *psize and *pmax_size to -1 if
6148 *poffset + *psize overflows in HOST_WIDE_INT.
6150 PR tree-optimization/82387
6151 PR tree-optimization/82388
6152 PR tree-optimization/82389
6153 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6154 instead of live_bytes non-NULL.
6156 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6159 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6162 2017-10-02 Richard Biener <rguenther@suse.de>
6164 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6165 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6166 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6168 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6170 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6171 requested precision matches the type's.
6172 * calls.c (alloc_max_size): Calculate the new candidate size as
6173 a widest_int and use wi::to_widest when comparing it with the
6174 current candidate size.
6175 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6176 zero rather than integer_zero_node.
6177 * match.pd: Check for a no-op conversion before using wi::add
6178 rather than after. Use tree_to_uhwi when summing small shift
6179 counts into an unsigned int.
6181 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6182 Alan Hayward <alan.hayward@arm.com>
6183 David Sherwood <david.sherwood@arm.com>
6186 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6187 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6188 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6189 POINTER_AND_FP_REGS.
6191 2017-10-02 Richard Biener <rguenther@suse.de>
6193 PR tree-optimization/82355
6194 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6195 a mapping for the enclosing loop but avoid generating one for
6197 (copy_bb_and_scalar_dependences): Remove premature codegen
6198 error on PHIs in blocks duplicated into multiple places.
6199 * graphite-scop-detection.c
6200 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6201 in the region use it as loop and nest to analyze the DR in.
6202 (try_generate_gimple_bb): Likewise.
6203 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6204 (add_loop_constraints): For blocks in a loop not in the region
6205 create a dimension with a single iteration.
6206 * sese.h (gbb_loop_at_index): Remove assert.
6208 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6210 * omp-expand.c (adjust_context_scope): New function.
6211 (expand_parallel_call): Call adjust_context_scope.
6213 2017-10-01 Jeff Law <law@redhat.com>
6215 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6216 dom_opt_dom_walker class with direct access to private members.
6217 Add comments. Call test_for_singularity.
6218 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6219 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6220 m_dummy_cond anymore.
6221 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6223 (pass_dominator:execute): Build the dummy_cond here and pass it
6224 to the dom_opt_dom_walker ctor.
6225 (test_for_singularity): New function.
6227 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6228 Maya Rashish <coypu@sdf.org>
6230 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6231 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6232 (alpha*-*-netbsd*) Use nbsd_tm_file.
6233 (arm*-*-netbsdelf*) Likewise.
6234 (i[34567]86-*-netbsdelf*) Likewise.
6235 (x86_64-*-netbsd*) Likewise.
6236 (mips*-*-netbsd*) Likewise.
6237 (powerpc-*-netbsd*) Likewise.
6238 (sh*-*-netbsd*) Likewise.
6239 (sparc-*-netbsdelf*) Likewise.
6240 (sparc64-*-netbsd*) Likewise.
6241 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6243 (vax-*-netbsdelf*) Likewise.
6244 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6245 (UINT_FAST8_TYPE) Likewise.
6246 (INT_FAST16_TYPE) Check CHAR_FAST16.
6247 (UINT_FAST16_TYPE) Likewise.
6249 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6252 * config/i386/i386.md
6253 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6254 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6255 *divmodsi4_zext_2): New define_insn_and_split.
6256 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6257 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6258 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6259 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6260 New define_insn_and_split.
6261 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6262 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6263 operands[1] having DImode when mode is SImode.
6265 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6266 always SImode for DIV and MOD in REG_EQUAL notes.
6268 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6271 * match.pd: Fix handling of NaNs in pattern.
6273 2017-09-29 Jeff Law <law@redhat.com>
6275 * sbitmap.c (bitmap_bit_in_range_p): New function.
6276 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6277 * tree-ssa-dse.c (live_bytes_read): New function.
6278 (dse_classify_store): Ignore reads of dead bytes.
6280 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6281 typos and whitespace errors.
6282 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6283 * config/s390/s390.c (s390_emit_prologue): Likewise.
6285 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6288 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6289 with a symbol for LRA.
6291 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6293 PR rtl-optimization/82338
6294 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6296 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6298 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6299 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6300 (qsort_chk): Declare.
6301 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6302 (qsort_chk): New function.
6304 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6306 PR tree-optimization/82337
6307 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6308 phi definition if the PHI result appears in an abnormal PHI.
6309 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6310 basis appears in an abnormal PHI.
6312 2017-09-29 Richard Biener <rguenther@suse.de>
6314 * graphite-isl-ast-to-gimple.c
6315 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6316 (binary_op_to_tree): Use it.
6317 (get_rename_from_scev): Likewise.
6318 (copy_loop_phi_nodes): Likewise.
6319 (copy_bb_and_scalar_dependences): Likewise.
6320 (translate_pending_phi_nodes): Likewise.
6322 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6325 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6326 for movabsq $(i32 << shift), r64.
6328 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6330 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6331 index when encoding %esp as %rsp to avoid 0x67 prefix.
6333 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6335 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6336 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6338 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6340 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6341 Extensions with more than 16 double VFP registers.
6342 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6343 to_clear_mask and all code related to it. Replace the remaining
6344 entry by a sbitmap and adapt code accordingly.
6346 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6348 * brig-builtins.def: Change pure attributes to const.
6350 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6352 * config.gcc (default_gnu_indirect_function): Default to yes for
6353 sparc*-*-linux* with glibc.
6355 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6357 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6358 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6359 when creating .init_array and .fini_array sections with priority
6362 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6365 * config/aarch64/aarch64.c
6366 (aarch64_builtin_support_vector_misalignment): Always return false
6367 when misalignment is unknown.
6369 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6371 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6372 this function to return false if the definition used by the swap
6373 instruction is artificial, or if the memory address from which the
6374 constant value is loaded is not represented by a base address held
6375 in a register or if the base address register is a frame or stack
6376 pointer. Additionally, return false if the base address of the
6377 loaded constant is a SYMBOL_REF but is not considered to be a
6379 (replace_swapped_load_constant): New function.
6380 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6381 loaded constant vector with a load of a swapped constant vector.
6383 2017-09-27 Carl Love <cel@us.ibm.com>
6385 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6386 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6387 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6390 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6392 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6393 first, always call autopref_rank_data otherwise.
6395 2017-09-27 Richard Biener <rguenther@suse.de>
6397 * graphite-scop-detection.c (find_scop_parameters): Move
6398 loop bound handling ...
6399 (gather_bbs::before_dom_children): ... here, avoiding the need
6400 to build scop_info->loop_nest.
6401 (record_loop_in_sese): Remove.
6402 * sese.h (sese_info_t::loop_nest): Remove.
6403 * sese.c (new_sese_info): Do not allocate loop_nest.
6404 (free_sese_info): Do not free loop_nest.
6406 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6409 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6410 lhs from calls if the lhs has addressable type.
6412 2017-09-27 Richard Biener <rguenther@suse.de>
6414 * graphite.h (scop::max_alias_set): New member.
6415 * graphite-scop-detection.c: Remove references to non-existing
6416 --param in comments.
6417 (build_alias_sets): Record the maximum alias set used for drs.
6418 (build_scops): Support zero as unlimited for
6419 --param graphite-max-arrays-per-scop.
6420 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6422 (build_poly_sr_1): ... here. Compute alias set based on the
6423 maximum alias set used for drs rather than
6424 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6426 2017-09-27 Richard Biener <rguenther@suse.de>
6428 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6429 --param loop-block-tile-size=0 to disable tiling.
6431 2017-09-27 Richard Biener <rguenther@suse.de>
6433 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6434 (graphite-max-nb-scop-params): Document special value zero.
6435 * domwalk.h (dom_walker::STOP): New symbolical constant.
6436 (dom_walker::dom_walker): Add optional parameter for bb to
6438 (dom_walker::~dom_walker): Declare.
6439 (dom_walker::before_dom_children): Document STOP return value.
6440 (dom_walker::m_user_bb_to_rpo): New member.
6441 (dom_walker::m_bb_to_rpo): Likewise.
6442 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6443 mapping here if not provided by the user.
6444 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6445 provided by the user.
6446 (dom_walker::STOP): Define.
6447 (dom_walker::walk): Do not compute bb to RPO mapping here.
6448 Support STOP return value from before_dom_children to stop
6450 * graphite-optimize-isl.c (optimize_isl): If the schedule
6451 is the same still generate code if -fgraphite-identity
6452 or -floop-parallelize-all are given.
6453 * graphite-scop-detection.c: Include cfganal.h.
6454 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6456 (gather_bbs::before_dom_children): Return STOP for BBs
6458 (build_scops): Compute bb to RPO mapping and pass it to
6459 the domwalk. Treat --param graphite-max-nb-scop-params=0
6460 as not limiting the number of params.
6461 * graphite.c (graphite_initialize): Remove limit on the
6462 number of basic-blocks in a function.
6463 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6464 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6465 default value of 10.
6467 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6469 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6470 Adjust code to eliminate needing to do the shift right 32-bits
6471 operation after XSCVDPSPN.
6473 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6475 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6476 ((X / Y) != 0 -> X >= Y): Likewise.
6478 2017-09-26 Carl Love <cel@us.ibm.com>
6480 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6481 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6482 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6483 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6484 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6485 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6486 definitions and overloading.
6487 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6488 statement for P9V_BUILTIN_XST_LEN_R.
6489 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6490 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6491 define_expand and define_insn for the instructions and builtins.
6492 * doc/extend.texi: Update the built-in documentation file for the new
6494 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6495 define_insn for the instructions
6497 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6500 * gcc/config/netbsd-protos.h: New file.
6501 * gcc/config/netbsd.c: New file.
6502 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6503 * gcc/config/t-netbsd: New file.
6504 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6505 (tmake_file) Add t-netbsd.
6506 (extra_objs) Add netbsd.o.
6508 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6512 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6514 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6516 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6517 sign extension from a vector register to a GPR by doing a 32-bit
6518 direct move and then an EXTSW.
6519 (extendsi<mode>2 splitter): Likewise.
6520 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6521 right or vector extract after doing XSCVDPSPN. Use
6522 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6524 (movdi_from_sf_zero_ext): Likewise.
6525 (reload_gpr_from_vsxsf): Likewise.
6526 (p8_mfvsrd_4_disf): Delete, no longer used.
6527 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6528 then needing to move the SFmode to a GPR to use the XSCVDPSP
6529 instruction instead of FRSP and XSCVDPSPN.
6530 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6531 it is adjacent to the other XSCVSPDP insns.
6532 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6533 SFmode to be in traditional Altivec registers.
6534 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6535 (vsx_xscvspdpn): Likewise.
6536 (vsx_xscvspdpn_scalar): Likewise.
6538 2017-09-26 Martin Jambor <mjambor@suse.cz>
6540 * tree-sra.c (compare_access_positions): Put integral types first,
6541 stabilize sorting of integral types, remove conditions putting
6542 non-full-precision integers last.
6543 (sort_and_splice_var_accesses): Disable scalarization if a
6544 non-integert would be represented by a non-full-precision integer.
6546 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6548 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6549 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6550 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6551 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6552 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6553 conditionals inside the function instead of around it. Call
6554 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6555 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6557 2017-09-26 Richard Biener <rguenther@suse.de>
6559 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6561 (scop_detection::build_scop_breadth): ... this. Removed.
6562 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6563 (scop_detection::harmful_stmt_in_bb): Likewise.
6564 (scop_detection::graphite_can_represent_stmt): Likewise.
6565 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6566 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6567 (scop_detection::can_represent_loop_1): ... this. Removed.
6568 (scop_detection::harmful_loop_in_region): Simplify after inlining
6569 the above and remove more quadraticness.
6570 (build_scops): Adjust.
6571 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6574 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6577 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6578 REGNO (base) == SP_REG if base is a REG.
6581 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6582 if it is different SSA_NAME.
6583 (optimize_range_tests_cmp_bitwise): New function.
6584 (optimize_range_tests): Call it.
6586 2017-09-26 Richard Biener <rguenther@suse.de>
6588 PR tree-optimization/82321
6589 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6590 for the def being inside the loop.
6592 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6594 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6596 * config/s390/s390-builtins.def: Fix constraint on op4.
6598 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6600 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6601 independent expanders.
6602 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6603 ("vec_ordered", "vec_unordered"): New expanders.
6605 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6607 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6610 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6612 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6613 vec_unpacks_lo_v16qi.
6614 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6616 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6618 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6619 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6620 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6622 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6624 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6626 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6628 ("*vec_slb<mode>"): New insn pattern.
6629 ("vec_shr_<mode>"): New expander.
6630 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6631 and force the shift count operand to V16QImode.
6632 ("vec_srb<mode>"): Set shift count mode to V16QI.
6634 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6636 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6637 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6638 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6640 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6643 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6645 2017-09-26 Richard Biener <rguenther@suse.de>
6647 PR tree-optimization/82320
6648 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6651 2017-09-25 Jeff Law <law@redhat.com>
6653 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6654 prototype for new argument.
6655 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6656 mostly extracted from rs6000_emit_allocate_stack.
6657 (rs6000_emit_probe_stack_range_stack_clash): New function.
6658 (rs6000_emit_allocate_stack): Call
6659 rs6000_emit_probe_stack_range_stack_clash as needed.
6660 (rs6000_emit_probe_stack_range): Add additional argument
6661 to call to gen_probe_stack_range{si,di}.
6662 (output_probe_stack_range): New.
6663 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6664 (output_probe_stack_range_stack_clash): New.
6665 (rs6000_emit_prologue): Emit notes into dump file as requested.
6666 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6667 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6668 Add additional operand and pass it to output_probe_stack_range.
6670 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6672 PR tree-optimization/82163
6673 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6674 (checking_verify_loop_closed_ssa): New parameter.
6675 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6676 (check_loop_closed_ssa_stmt): Delete.
6677 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6678 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6679 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6682 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6684 * brig-builtins.def: Treat HSAIL barrier builtins as
6685 setjmp/longjump style functions.
6687 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6689 * target.def (constant_alignment): New hook.
6690 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6691 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6692 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6693 * doc/tm.texi: Regenerate.
6694 * targhooks.h (default_constant_alignment): Declare.
6695 (constant_alignment_word_strings): Likewise.
6696 * targhooks.c (default_constant_alignment): New function.
6697 (constant_alignment_word_strings): Likewise.
6698 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6699 instead of CONSTANT_ALIGNMENT.
6700 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6701 (force_const_mem): Likewise.
6702 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6703 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6704 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6705 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6706 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6708 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6709 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6710 constant_alignment_word_strings.
6711 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6712 (CONSTANT_ALIGNMENT): Likewise.
6713 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6714 (arm_constant_alignment): New function.
6715 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6716 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6717 constant_alignment_word_strings.
6718 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6719 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6720 constant_alignment_word_strings.
6721 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6722 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6723 (cris_constant_alignment): New function.
6724 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6725 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6726 (epiphany_constant_alignment): New function.
6727 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6728 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6729 constant_alignment_word_strings.
6730 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6731 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6732 constant_alignment_word_strings.
6733 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6734 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6735 constant_alignment_word_strings.
6736 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6737 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6738 * config/i386/i386.c (ix86_constant_alignment): Make static.
6739 Use the same interface as the target hook.
6740 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6741 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6742 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6743 constant_alignment_word_strings.
6744 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6745 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6746 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6747 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6748 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6749 constant_alignment_word_strings.
6750 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6751 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6752 constant_alignment_word_strings.
6753 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6754 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6755 constant_alignment_word_strings.
6756 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6757 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6759 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6760 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6761 * config/mips/mips.c (mips_constant_alignment): New function.
6762 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6763 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6764 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6765 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6766 (mmix_constant_alignment): Make static. Use the same interface
6768 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6769 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6770 constant_alignment_word_strings.
6771 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6772 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6773 constant_alignment_word_strings.
6774 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6775 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6776 constant_alignment_word_strings.
6777 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6778 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6779 (rs6000_constant_alignment): New function.
6780 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6781 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6782 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6783 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6784 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6785 (rs6000_constant_alignment): New function.
6786 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6787 * config/s390/s390.c (s390_constant_alignment): New function.
6788 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6789 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6790 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6791 constant_alignment_word_strings.
6792 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6793 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6794 (sparc_constant_alignment): New function.
6795 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6796 * config/spu/spu.c (spu_constant_alignment): New function.
6797 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6798 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6799 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6800 constant_alignment_word_strings.
6801 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6802 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6803 constant_alignment_word_strings.
6804 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6805 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6806 constant_alignment_word_strings.
6807 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6808 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6809 (visium_constant_alignment): New function.
6810 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6811 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6812 (xtensa_constant_alignment): New function.
6813 * system.h (CONSTANT_ALIGNMENT): Poison.
6815 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6817 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6818 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6819 (rs6000_builtin_valid_without_lhs): New helper function.
6820 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6821 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6823 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6825 * target.h (vec_perm_indices): Use unsigned short rather than
6827 (auto_vec_perm_indices): Likewise.
6828 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6829 Use unsigned int rather than unsigned char.
6830 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6832 2017-09-25 Richard Biener <rguenther@suse.de>
6834 * cfgloop.h (sort_sibling_loops): Declare.
6835 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6836 (sort_sibling_loops): New function sorting the sibling loop list
6838 * graphite.c (graphite_transform_loops): Sort sibling loops.
6840 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6842 * target.def (vec_perm_const_ok): Change sel parameter to
6844 * optabs-query.c (can_vec_perm_p): Update accordingly.
6845 * doc/tm.texi: Regenerate.
6846 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6847 auto_vec_perm_indices and remove separate nelt field.
6848 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6849 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6850 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6851 (aarch64_expand_vec_perm_const): Update accordingly.
6852 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6853 to vec_perm_indices.
6854 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6855 auto_vec_perm_indices and remove separate nelt field.
6856 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6857 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6858 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6860 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6861 to vec_perm_indices.
6862 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6863 sel to vec_perm_indices.
6864 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6865 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6866 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6868 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6871 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6874 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6875 on the FUNCTION_DECL function context if it has a DIE that is a
6878 2017-09-25 Richard Biener <rguenther@suse.de>
6880 PR tree-optimization/82285
6881 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6884 2017-09-25 Tom de Vries <tom@codesourcery.com>
6888 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6891 2017-09-25 Richard Biener <rguenther@suse.de>
6893 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6894 ISL errors other than isl_error_quota happen. Dump if the
6895 schedule is the same.
6896 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6897 errors instead of aborting inside ISL.
6899 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6902 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6903 of libgcc_eh for m64.
6904 * config/i386/darwin64.h: Likewise.
6906 2017-09-25 Richard Biener <rguenther@suse.de>
6909 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6910 attribute for incomplete types nor twice for complete ones.
6912 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6915 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6916 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6919 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6922 * config/i386/i386.opt (mprefer-avx256): Use
6923 ix86_target_flags variable.
6924 * config/i386/i386.c (ix86_target_string): Move
6925 -mprefer-avx256 to flag2_opts.
6927 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6930 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6931 and x != -1 | y != -1 into (x & y) != -1.
6933 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6935 * config.gcc: Add new case statement to set
6936 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6937 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6938 s390x-*-linux* case statements. Added aarch64 to the list of
6939 supported architectures.
6941 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6943 PR tree-optimization/82289
6944 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6945 STMT_VINFO_RELEVANT_P.
6947 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6952 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6953 (extract_range_from_binary_expr_1): Don't call
6954 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6956 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6957 Alan Hayward <alan.hayward@arm.com>
6958 David Sherwood <david.sherwood@arm.com>
6960 * target.def (preferred_vector_alignment): New hook.
6961 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6963 * doc/tm.texi: Regenerate.
6964 * targhooks.h (default_preferred_vector_alignment): Declare.
6965 * targhooks.c (default_preferred_vector_alignment): New function.
6966 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6968 (DR_TARGET_ALIGNMENT): New macro.
6969 (aligned_access_p): Update commentary.
6970 (vect_known_alignment_in_bytes): New function.
6971 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6973 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6974 Calculate the misalignment based on the target alignment rather than
6976 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6977 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6978 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6979 the target alignment, rather than masking the element misalignment
6980 with the number of elements in a vector. Also use the target
6981 alignment when calculating the maximum number of peels.
6982 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6983 instead of TYPE_ALIGN_UNIT.
6984 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6985 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6986 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6987 (vect_create_data_ref_ptr): Likewise.
6988 (vect_setup_realignment): Realign by ANDing with
6989 -DR_TARGET_MISALIGNMENT.
6990 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6991 the number of peels based on DR_TARGET_ALIGNMENT.
6992 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6993 with the guaranteed alignment boundary when deciding whether
6995 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6996 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6997 (ensure_base_align): Remove stmt_info parameter. Get the
6998 target base alignment from DR_TARGET_ALIGNMENT.
6999 (vectorizable_store): Update call accordingly. Interpret
7000 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7002 (vectorizable_load): Likewise.
7004 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7005 Alan Hayward <alan.hayward@arm.com>
7006 David Sherwood <david.sherwood@arm.com>
7008 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7009 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7010 (vect_enhance_data_refs_alignment): Likewise.
7012 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7014 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7015 error. Only quit immediately if parsing is complete.
7016 (BEGIN): Initialize fatal_err and parse_done.
7017 (begin fpu, end fpu): Check number of arguments.
7018 (begin arch, end arch): Likewise.
7019 (begin cpu, end cpu): Likewise.
7020 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7021 (optalias): Likewise.
7023 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7025 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7026 * config/arm/arm-isa.h: Delete. Move definitions to ...
7027 * arm-cpus.in: ... here. Use new feature and fgroup values.
7028 * config/arm/arm.c (arm_option_override): Use lower case for feature
7030 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7031 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7032 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7034 (print_isa_bits_for): New function.
7035 (gen_isa): New function.
7036 (gen_comm_data): Use print_isa_bits_for.
7037 (define feature): New keyword.
7038 (define fgroup): New keyword.
7039 * config/arm/t-arm (TM_H): Remove.
7040 (GTM_H): Add arm-isa.h.
7041 (arm-isa.h): Add rule to generate file.
7042 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7043 case for feature bit names.
7045 2017-09-22 Richard Biener <rguenther@suse.de>
7047 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7049 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7050 print of no dependency loops ...
7051 * graphite.c (graphite_transform_loops): ... here.
7052 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7054 (same_close_phi_node, remove_duplicate_close_phi,
7055 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7056 (canonicalize_loop_closed_ssa): ... here and simplify.
7057 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7058 (optimize_isl): Use dump_printf_loc to tell when we stopped
7059 optimizing because of an ISL timeout.
7061 2017-09-22 Richard Biener <rguenther@suse.de>
7063 PR tree-optimization/82291
7064 * tree-if-conv.c (predicate_mem_writes): Make sure to
7065 remove writes in blocks predicated with false.
7067 2017-09-22 Richard Biener <rguenther@suse.de>
7069 * sese.c: Include cfganal.h.
7070 (if_region_set_false_region): Remove.
7071 (create_if_region_on_edge): Likewise.
7072 (move_sese_in_condition): Re-implement without destroying
7075 2017-09-22 Richard Biener <rguenther@suse.de>
7077 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7078 Verify both BBs contain loop PHI nodes before dispatching to
7080 (graphite_regenerate_ast_isl): Do not recompute dominators,
7081 do not verify three times. Restructure for clarity.
7082 * graphite-scop-detection.c (same_close_phi_node,
7083 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7084 defined_in_loop_p, canonicalize_loop_closed_ssa,
7085 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7086 checking and SSA rewrite, move to ...
7087 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7088 (graphite_initialize): Do not pass in ctx, do not reset the
7089 SCEV cache, compute only dominators.
7090 (graphite_transform_loops): Allocate ISL ctx after
7091 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7092 Maintain post-dominators only around build_scops.
7093 * sese.c (if_region_set_false_region): Make static. Free
7094 and recompute dominators.
7095 (move_sese_in_condition): Assert we don't get called with
7096 post-dominators computed.
7097 * sese.h (if_region_set_false_region): Remove.
7099 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7101 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7102 mode attribute for TARGET_AVX512VL.
7104 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7106 * config/i386/i386.opt (mprefer-avx256): New option.
7107 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7109 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7110 for TARGET_PREFER_AVX256.
7111 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7113 2017-09-21 Jeff Law <law@redhat.com>
7115 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7116 Fix dump output if the only stack space is for pushed registers.
7118 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7120 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7123 2017-09-21 Martin Sebor <msebor@redhat.com>
7126 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7127 code (in C++) or code that triggers warnings.
7129 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7131 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7133 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7135 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7136 * sched-rgn.c: Ditto.
7137 * sel-sched-ir.c: Ditto.
7139 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7141 * toplev.h (set_random_seed): Adjust return type.
7142 * toplev.c (init_local_tick): Move eager initialization of random_seed
7143 to get_random_seed. Adjust comment.
7144 (init_random_seed): Inline to get_random_seed, delete.
7145 (get_random_seed): Initialize random_seed lazily.
7146 (set_random_seed): Do not return previous value.
7147 (print_switch_value): Do not call get_random_seed.
7149 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7151 * cgraph.c (delete_function_version): New, broken out from...
7152 (cgraph_node::delete_function_version): ...here. Rename to
7153 cgraph_node::delete_function_version_by_decl. Update all uses.
7154 (cgraph_node::remove): Call delete_function_version.
7156 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7159 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7160 VAR_DECLs to which addressable non-volatile parameters are mapped
7161 and for id->retvar after the return value assignment. Clear
7162 id->retval and id->retbnd after inlining.
7164 2017-09-21 Richard Biener <rguenther@suse.de>
7166 PR tree-optimization/82276
7167 PR tree-optimization/82244
7168 * tree-vrp.c (build_assert_expr_for): Set
7169 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7171 (remove_range_assertions): Revert earlier change.
7173 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7176 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7178 2017-09-21 Richard Biener <rguenther@suse.de>
7180 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7181 Restore valid IL after code generation errors.
7182 * graphite.c (graphite_transform_loops): Diagnose code
7183 generation issues as MSG_MISSED_OPTIMIZATION and continue
7184 with processing SCOPs.
7186 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7187 Alan Hayward <alan.hayward@arm.com>
7188 David Sherwood <david.sherwood@arm.com>
7190 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7191 rather than choosing between plus_constant and gen_rtx_<CODE>.
7192 * expr.c (emit_push_insn): Likewise.
7193 (expand_expr_real_2): Likewise.
7195 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7196 Alan Hayward <alan.hayward@arm.com>
7197 David Sherwood <david.sherwood@arm.com>
7199 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7201 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7202 Alan Hayward <alan.hayward@arm.com>
7203 David Sherwood <david.sherwood@arm.com>
7205 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7206 calling tree_to_uhwi.
7208 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7209 Alan Hayward <alan.hayward@arm.com>
7210 David Sherwood <david.sherwood@arm.com>
7212 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7213 INTEGER_CST rather than a negative test for ADDR_EXPR.
7215 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7216 Alan Hayward <alan.hayward@arm.com>
7217 David Sherwood <david.sherwood@arm.com>
7219 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7220 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7222 2017-09-21 Richard Biener <rguenther@suse.de>
7224 PR tree-optimization/71351
7225 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7226 graphite_create_new_loop_guard): Remove, fold remaining parts
7228 (translate_isl_ast_node_for): ... here and simplify.
7230 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7233 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7234 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7235 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7236 alternative always use QI mode, for -Os imov (=R,R) alternative
7237 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7240 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7241 Jeff Law <law@redhat.com>
7243 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7244 (allocate_stack_space): New function, partially extracted from
7246 (s390_emit_prologue): Track offset to most recent stack probe.
7247 Code to allocate space moved into allocate_stack_space.
7248 Dump actions when no stack is allocated.
7249 (s390_prologue_plus_offset): New function.
7250 (s390_emit_stack_probe): Likewise.
7252 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7254 * common.opt (Wa, Wl, Wp, g, gz=): Add
7256 (gno-column-info): Remove.
7257 (gcolumn-info): Drop RejectNegative.
7259 (gno-record-gcc-switches): Remove.
7260 (grecord-gcc-switches): Drop RejectNegative.
7261 (gno-split-dwarf): Remove.
7262 (gsplit-dwarf): Drop RejectNegative.
7263 (gno-strict-dwarf): Remove.
7264 (gstrict-dwarf): Drop RejectNegative.
7265 * config/darwin.opt (gfull, gused): Add RejectNegative.
7266 * dwarf2out.c (gen_producer_string): Drop
7267 gno-record-gcc-switches handler.
7268 * optc-gen.awk: Add g to prefixes with negative forms.
7269 * opts-common.c (remapping_prefix_p): New.
7270 (find_opt): Check it.
7271 (generate_canonical_option): Test g prefix.
7272 (option_map): Add -gno- mapping.
7273 (add_misspelling_candidates): Check remapping_prefix_p.
7275 2017-09-20 Jeff Law <law@redhat.com>
7277 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7278 thinko in stack clash protection support.
7280 * explow.c (compute_stack_clash_protection_loop_data): Use
7281 CONST_INT_P instead of explicit test. Verify object is a
7282 CONST_INT_P before looking at INTVAL.
7283 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7284 instead of explicit test.
7286 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7289 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7290 address instead of to r1 and r11.
7292 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7294 * config.gcc: Support "knm".
7295 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7296 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7298 * config/i386/i386.c (m_KNM): Define.
7299 (processor_target_table): Add "knm".
7301 (ix86_option_override_internal): Add "knm".
7302 (ix86_issue_rate): Add PROCESSOR_KNM.
7303 (ix86_adjust_cost): Ditto.
7304 (ia32_multipass_dfa_lookahead): Ditto.
7305 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7306 (fold_builtin_cpu): Add M_INTEL_KNM.
7307 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7308 (processor_type): Add PROCESSOR_KNM.
7309 * config/i386/x86-tune.def: Add m_KNM.
7310 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7312 2017-09-20 Richard Biener <rguenther@suse.de>
7314 PR tree-optimization/80213
7315 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7316 are allowed in empty BBs as well.
7317 (canonicalize_loop_closed_ssa): Also look for other complex
7319 (scop_detection::get_sese): Include the loop-closed PHI block
7321 (scop_detection::merge_sese): Remove code adding extra blocks.
7322 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7323 (build_scops): Assert the final returned scop is invalid.
7325 2017-09-20 Richard Biener <rguenther@suse.de>
7327 PR tree-optimization/82264
7328 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7330 (vn_phi_lookup): Likewise.
7331 (vn_phi_insert): Likewise.
7333 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7335 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7336 that fits into uhwi or shwi, add DW_AT_const_value regardless
7337 of early_dwarf without going through RTL, using add_AT_unsigned
7340 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7341 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7342 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7343 (DEBUG_MACRO_SECTION): Likewise.
7344 (DEBUG_LTO_MACRO_SECTION): Likewise.
7345 (DEBUG_STR_DWO_SECTION): Likewise.
7346 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7347 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7348 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7349 (DEBUG_LTO_LINE_STR_SECTION): Define.
7350 (init_sections_and_labels): Initialize debug_line_str_section
7351 variable. Initialize debug_loc_section for -gdwarf-5 to
7352 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7354 2017-09-20 Richard Biener <rguenther@suse.de>
7356 * graphite-sese-to-poly.c (extract_affine): Properly handle
7357 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7359 2017-09-20 Richard Biener <rguenther@suse.de>
7361 PR tree-optimization/81373
7362 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7363 Force SESE live-out defs to be handled even if they are
7366 2017-09-19 Jeff Law <law@redhat.com>
7368 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7369 nothing for stack adjustments with REG_STACK_CHECK.
7370 * sched-deps.c (parse_add_or_inc): Reject insns with
7371 REG_STACK_CHECK from dependency breaking.
7372 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7373 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7374 * reg-notes.def (STACK_CHECK): New note.
7376 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7377 (ix86_expand_prologue): Dump stack clash info as needed.
7378 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7380 * function.c (dump_stack_clash_frame_info): New function.
7381 * function.h (dump_stack_clash_frame_info): Prototype.
7382 (enum stack_clash_probes): New enum.
7384 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7385 flag_stack_clash_protection.
7386 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7387 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7388 (arm_frame_pointer_required): Likewise.
7389 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7390 (ia64_expand_prologue): Likewise.
7391 * config/mips/mips.c (mips_expand_prologue): Likewise.
7392 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7393 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7394 (sparc_flat_expand_prologue): Likewise.
7395 * config/spu/spu.c (spu_expand_prologue): Likewise.
7397 * explow.c: Include "params.h".
7398 (anti_adjust_stack_and_probe_stack_clash): New function.
7399 (get_stack_check_protect): Likewise.
7400 (compute_stack_clash_protection_loop_data): Likewise.
7401 (emit_stack_clash_protection_loop_start): Likewise.
7402 (emit_stack_clash_protection_loop_end): Likewise.
7403 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7404 Use anti_adjust_stack_and_probe_stack_clash.
7405 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7406 (emit_stack_clash_protection_loop_start): Likewise.
7407 (emit_stack_clash_protection_loop_end): Likewise.
7408 * rtl.h (get_stack_check_protect): Prototype.
7409 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7410 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7411 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7413 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7415 * doc/tm.texi: Rebuilt.
7416 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7417 get_stack_check_protect.
7418 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7419 * config/arm/arm.c (arm_expand_prologue): Likewise.
7420 (arm_frame_pointer_required): Likewise.
7421 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7422 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7423 * config/mips/mips.c (mips_expand_prologue): Likewise.
7424 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7425 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7426 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7427 (sparc_flat_expand_prologue): Likewise.
7429 * common.opt (-fstack-clash-protection): New option.
7430 * flag-types.h (enum stack_check_type): Note difference between
7431 -fstack-check= and -fstack-clash-protection.
7432 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7433 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7434 * toplev.c (process_options): Issue warnings/errors for cases
7435 not handled with -fstack-clash-protection.
7436 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7437 (-fstack-check): Note additional problem with -fstack-check=generic.
7438 Note that -fstack-check is primarily for Ada and refer users
7439 to -fstack-clash-protection for stack-clash-protection.
7440 Document new params for stack clash protection.
7442 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7444 * config/i386/i386.c (ix86_split_long_move): Do not handle
7445 address used for LEA in a special way.
7447 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7449 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7451 2017-09-19 Martin Sebor <msebor@redhat.com>
7454 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7455 of incompatible types.
7457 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7459 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7460 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7461 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7462 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7464 2017-09-19 Richard Biener <rguenther@suse.de>
7466 PR tree-optimization/82244
7467 * tree-vrp.c (remove_range_assertions): Do not propagate
7468 a constant to abnormals but replace the assert with a copy.
7470 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7472 PR rtl-optimization/57878
7473 PR rtl-optimization/68988
7474 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7475 avoidance test involving non_reload_pseudos. Move frequency test
7476 below the general fragmentation avoidance test.
7478 2017-09-19 Richard Biener <rguenther@suse.de>
7480 PR tree-optimization/69728
7481 * graphite-sese-to-poly.c (schedule_error): New global.
7482 (add_loop_schedule): Handle empty domain by failing the
7484 (build_original_schedule): Handle schedule_error.
7486 2017-09-19 Richard Biener <rguenther@suse.de>
7488 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7489 Do not iterate to sibling loops but only to siblings of inner
7492 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7495 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7496 that the registers are different.
7498 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7500 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7501 to processor_model and "amdfam17h" to arch_names_table.
7502 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7504 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7507 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7509 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7510 Alan Hayward <alan.hayward@arm.com>
7511 David Sherwood <david.sherwood@arm.com>
7513 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7515 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7517 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7518 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7519 vect_schedule_slp_instance.
7520 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7521 Update call to vect_slp_analyze_node_operations. Simplify return
7523 (vect_slp_analyze_bb_1): Update call accordingly.
7524 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7525 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7526 (vect_schedule_slp): Update call accordingly.
7528 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7529 Alan Hayward <alan.hayward@arm.com>
7530 David Sherwood <david.sherwood@arm.com>
7532 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7533 with types that aren't in fact scalar.
7535 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7537 * tree-vect-slp.c (vect_record_max_nunits): New function,
7539 (vect_build_slp_tree_1): ...here.
7540 (vect_build_slp_tree_2): Call it for phis too.
7542 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7544 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7545 to vect_get_vec_def_for_operand when getting the mask operand.
7547 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7548 Alan Hayward <alan.hayward@arm.com>
7549 David Sherwood <david.sherwood@arm.com>
7551 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7554 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7555 Alan Hayward <alan.hayward@arm.com>
7556 David Sherwood <david.sherwood@arm.com>
7558 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7559 calculation for vector booleans.
7561 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7562 Alan Hayward <alan.hayward@arm.com>
7563 David Sherwood <david.sherwood@arm.com>
7565 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7567 (vect_transform_stmt): ...here.
7568 (vect_analyze_stmt): Use it instead of calling
7569 vectorizable_live_operation directly.
7571 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7573 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7574 non-SIMT targets in acc vector loops.
7576 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7578 * configure.ac: Add arc and check if assembler supports gdwarf2.
7579 * configure: Regenerate.
7581 2017-09-18 Richard Biener <rguenther@suse.de>
7583 PR tree-optimization/82220
7584 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7585 epilogue niters from the min_profitable_iters compute.
7587 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7590 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7591 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7592 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7594 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7597 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7598 switching to a new text section.
7600 2017-09-18 Richard Biener <rguenther@suse.de>
7602 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7604 (build_alias_set): Reject aliases with no access function.
7606 2017-09-18 Richard Biener <rguenther@suse.de>
7608 PR tree-optimization/79622
7609 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7611 (build_cross_bb_scalars_use): Likewise.
7613 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7615 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7617 2017-09-18 Alan Modra <amodra@gmail.com>
7620 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7621 stack_pointer_rtx for count 0. Update comments. Break up
7622 large rtl expression.
7624 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7626 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7627 Increase to 20 bytes.
7628 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7629 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7630 or avx version of the stub.
7632 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7635 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7636 compute the minimum stack alignment. Also update preferred stack
7637 boundary for leaf functions.
7639 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7641 PR tree-optimization/82228
7642 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7645 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7647 * common/config/nds32/nds32-common.c
7648 (nds32_option_optimization_table): Refine formatting.
7649 (nds32_option_optimization_table): Use -fsched-pressure and
7650 -fomit-frame-pointer for specific optimization level.
7652 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7654 * config/nds32/nds32.c: Refine formatting and comments.
7655 * config/nds32/nds32.h: Likewise.
7656 * config/nds32/nds32.md: Likewise.
7657 * config/nds32/nds32-cost.c: Likewise.
7658 * config/nds32/nds32-isr.c: Likewise.
7659 * config/nds32/nds32-md-auxiliary.c: Likewise.
7660 * config/nds32/nds32-multiple.md: Likewise.
7661 * config/nds32/nds32-predicates.c: Likewise.
7663 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7664 Jakub Jelinek <jakub@redhat.com>
7666 Add support for -std=c++2a.
7667 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7669 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7671 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7674 * doc/extend.texi (Common Function Attributes): Add
7675 references to ARM, AArch64, and S/390 specific attributes.
7676 (Function Specific Option Pragmas): Add AArch64 and S/390
7677 to list of back ends that support the target pragma.
7679 2017-09-15 Nathan Sidwell <nathan@acm.org>
7681 * doc/standards.texi: Fix C++17 description. Update URLs for
7684 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7686 * common.opt (Wcast-align=strict): New warning option.
7687 * doc/invoke.texi: Document -Wcast-align=strict.
7689 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7691 * cgraph.h (cgraph_thunk_info): Add comments.
7692 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7693 assert for VIRTUAL_* arguments stricter.
7695 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7697 PR tree-optimization/71026
7698 * match.pd: Move RDIV patterns from fold-const.c
7699 * fold-const.c (distribute_real_division): Removed.
7700 (fold_binary_loc): Remove calls to distribute_real_divison.
7702 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7704 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7705 c++1z and gnu++1z as deprecated. Change other references to
7706 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7707 Change -Wc++1z-compat to -Wc++17-compat.
7708 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7709 * dwarf2out.c (highest_c_language): Handle C++17.
7710 (gen_compile_unit_die): Likewise.
7712 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7714 PR rtl-optimization/82192
7715 * combine.c (make_extraction): Don't look through non-paradoxical
7716 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7719 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7720 Alan Hayward <alan.hayward@arm.com>
7721 David Sherwood <david.sherwood@arm.com>
7723 * target.def (function_arg_offset): New hook.
7724 * targhooks.h (default_function_arg_offset): Declare.
7725 * targhooks.c (default_function_arg_offset): New function.
7726 * function.c (locate_and_pad_parm): Use
7727 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7728 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7729 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7730 * doc/tm.texi: Regenerate.
7731 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7732 * config/spu/spu.c (spu_function_arg_offset): New function.
7733 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7734 * system.h (FUNCTION_ARG_OFFSET): Poison.
7736 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7737 Alan Hayard <alan.hayward@arm.com>
7738 David Sherwood <david.sherwood@arm.com>
7740 * target.def (truly_noop_truncation): New hook.
7741 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7742 than TRULY_NOOP_TRUNCATION.
7743 * hooks.h (hook_bool_uint_uint_true): Declare.
7744 * hooks.c (hook_bool_uint_uint_true): New function.
7745 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7746 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7747 * doc/tm.texi: Regenerate.
7748 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7749 rather than TRULY_NOOP_TRUNCATION in comments.
7750 (simplify_comparison): Likewise.
7751 (record_truncated_value): Likewise.
7752 * expmed.c (extract_bit_field_1): Likewise.
7753 (extract_split_bit_field): Likewise.
7754 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7755 instead of TRULY_NOOP_TRUNCATION.
7756 * function.c (assign_parm_setup_block): Likewise.
7757 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7758 * rtlhooks.c: Include target.h.
7759 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7760 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7761 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7762 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7763 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7764 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7765 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7766 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7767 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7768 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7769 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7770 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7771 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7772 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7773 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7774 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7775 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7776 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7777 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7778 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7779 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7780 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7781 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7782 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7783 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7784 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7785 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7786 rather than TRULY_NOOP_TRUNCATION in comments.
7787 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7788 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7789 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7790 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7791 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7792 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7793 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7794 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7795 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7796 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7797 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7798 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7799 rather than TRULY_NOOP_TRUNCATION in comments.
7800 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7801 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7802 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7803 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7804 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7805 TRULY_NOOP_TRUNCATION condition.
7806 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7807 (TRULY_NOOP_TRUNCATION): Delete.
7808 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7809 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7810 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7811 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7812 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7813 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7814 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7815 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7816 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7817 rather than TRULY_NOOP_TRUNCATION in comments.
7818 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7819 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7820 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7821 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7822 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7823 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7825 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7828 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7829 (*cmp_ior): Likewise.
7830 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7831 (*ior_scc_scc_cmp): Likewise.
7832 (*and_scc_scc): Likewise.
7833 (*and_scc_scc_cmp): Likewise.
7835 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7836 Alan Hayard <alan.hayward@arm.com>
7837 David Sherwood <david.sherwood@arm.com>
7839 * target.def (can_change_mode_class): New hook.
7840 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7841 (hard_regno_nregs): Likewise.
7842 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7843 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7844 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7845 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7846 (LOAD_EXTEND_OP): Update accordingly.
7847 * doc/tm.texi: Regenerate.
7848 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7849 CANNOT_CHANGE_MODE_CLASS.
7850 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7851 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7852 * combine.c (simplify_set): Update accordingly.
7853 * emit-rtl.c (validate_subreg): Likewise.
7854 * recog.c (general_operand): Likewise.
7855 * regcprop.c (mode_change_ok): Likewise.
7856 * reload1.c (choose_reload_regs): Likewise.
7857 (inherit_piecemeal_p): Likewise.
7858 * rtlanal.c (simplify_subreg_regno): Likewise.
7859 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7860 instead of CANNOT_CHANGE_MODE_CLASS.
7861 (reload_cse_simplify_operands): Likewise.
7862 * reload.c (push_reload): Use targetm.can_change_mode_class
7863 instead of CANNOT_CHANGE_MODE_CLASS.
7864 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7865 REG_CANNOT_CHANGE_MODE_P.
7866 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7867 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7868 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7869 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7870 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7871 (arm_can_change_mode_class): New function.
7872 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7873 than CANNOT_CHANGE_MODE_CLASS in comments.
7874 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7875 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7876 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7877 (ix86_can_change_mode_class): ...this new function, inverting the
7878 sense of the return value.
7879 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7880 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7881 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7882 (ia64_can_change_mode_class): New function.
7883 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7884 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7885 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7886 (m32c_can_change_mode_class): ...this new function, inverting the
7887 sense of the return value.
7888 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7889 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7890 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7891 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7892 (mips_can_change_mode_class): ...this new function, inverting the
7893 sense of the return value.
7894 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7895 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7896 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7897 (msp430_can_change_mode_class): New function.
7898 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7899 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7900 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7901 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7902 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7903 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7904 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7905 (pa_cannot_change_mode_class): Replace with...
7906 (pa_can_change_mode_class): ...this new function, inverting the
7907 sense of the return value.
7908 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7909 than CANNOT_CHANGE_MODE_CLASS in comments.
7910 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7911 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7912 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7913 (pdp11_cannot_change_mode_class): Replace with...
7914 (pdp11_can_change_mode_class): ...this new function, inverting the
7915 sense of the return value.
7916 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7917 * config/powerpcspe/powerpcspe-protos.h
7918 (rs6000_cannot_change_mode_class_ptr): Delete.
7919 * config/powerpcspe/powerpcspe.c
7920 (rs6000_cannot_change_mode_class_ptr): Delete.
7921 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7922 (rs6000_option_override_internal): Assign to
7923 targetm.can_change_mode_class instead of
7924 rs6000_cannot_change_mode_class_ptr.
7925 (rs6000_cannot_change_mode_class): Replace with...
7926 (rs6000_can_change_mode_class): ...this new function, inverting the
7927 sense of the return value.
7928 (rs6000_debug_cannot_change_mode_class): Replace with...
7929 (rs6000_debug_can_change_mode_class): ...this new function.
7930 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7931 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7932 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7933 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7934 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7936 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7937 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7938 (rs6000_option_override_internal): Assign to
7939 targetm.can_change_mode_class instead of
7940 rs6000_cannot_change_mode_class_ptr.
7941 (rs6000_cannot_change_mode_class): Replace with...
7942 (rs6000_can_change_mode_class): ...this new function, inverting the
7943 sense of the return value.
7944 (rs6000_debug_cannot_change_mode_class): Replace with...
7945 (rs6000_debug_can_change_mode_class): ...this new function.
7946 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7947 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7948 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7949 (s390_can_change_mode_class): ...this new function, inverting the
7950 sense of the return value.
7951 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7952 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7953 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7954 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7955 (sh_cannot_change_mode_class): Replace with...
7956 (sh_can_change_mode_class): ...this new function, inverting the
7957 sense of the return value.
7958 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7959 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7960 (sparc_can_change_mode_class): New function.
7961 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7962 * config/spu/spu.c (spu_can_change_mode_class): New function.
7963 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7964 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7965 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7966 (visium_can_change_mode_class): New function.
7967 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7969 2017-09-15 Richard Biener <rguenther@suse.de>
7971 PR tree-optimization/82217
7972 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7973 but not undefined case.
7975 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7978 * postreload.c (reload_cse_simplify_operands): Skip
7979 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7981 2017-09-15 Richard Biener <rguenther@suse.de>
7983 PR tree-optimization/68823
7984 * graphite-scop-detection.c (build_alias_set): If we have a
7985 possible dependence check whether we can handle them by just
7986 looking at the DRs DR_ACCESS_FNs.
7987 (build_scops): If build_alias_set fails, fail the SCOP.
7989 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7991 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7992 to support float128 built-in functions that require the ISA 3.0
7994 (BU_FLOAT128_3_HW): Likewise.
7995 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7997 (FMAF128): Likewise.
7999 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8000 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8001 floating point instructions.
8002 (rs6000_invalid_builtin): Likewise.
8003 (rs6000_builtin_mask_names): Likewise.
8004 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8005 (RS6000_BTM_FLOAT128_HW): Likewise.
8006 (RS6000_BTM_COMMON): Likewise.
8007 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8009 * doc/extend.texi (RS/6000 built-in functions): Document the
8010 IEEE 128-bit floating point square root and fused multiply-add
8013 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8015 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8016 reg (r2) isn't in the set of registers defined in the prologue.
8018 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8019 Alan Hayward <alan.hayward@arm.com>
8020 David Sherwood <david.sherwood@arm.com>
8022 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8023 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8024 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8025 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8026 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8028 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8029 max_vectorization_factor.
8030 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8032 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8033 Alan Hayward <alan.hayward@arm.com>
8034 David Sherwood <david.sherwood@arm.com>
8036 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8037 (vect_worthwhile_without_simd_p): Declare.
8038 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8039 (vectorizable_reduction): Use it.
8040 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8041 (vectorizable_operation): Likewise.
8043 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8044 Alan Hayward <alan.hayward@arm.com>
8045 David Sherwood <david.sherwood@arm.com>
8047 * tree-vectorizer.h (vect_get_num_copies): New function.
8048 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8049 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8050 (vectorizable_induction): Likewise.
8051 (vectorizable_live_operation): Likewise.
8052 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8053 (vectorizable_bswap): Likewise.
8054 (vectorizable_call): Likewise.
8055 (vectorizable_conversion): Likewise.
8056 (vectorizable_assignment): Likewise.
8057 (vectorizable_shift): Likewise.
8058 (vectorizable_operation): Likewise.
8059 (vectorizable_store): Likewise.
8060 (vectorizable_load): Likewise.
8061 (vectorizable_condition): Likewise.
8062 (vectorizable_comparison): Likewise.
8063 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8065 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8066 Alan Hayward <alan.hayward@arm.com>
8067 David Sherwood <david.sherwood@arm.com>
8069 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8070 of vect_init_vector.
8072 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8073 Alan Hayward <alan.hayward@arm.com>
8074 David Sherwood <david.sherwood@arm.com>
8076 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8077 an inline wrapper that provides a location.
8078 (gimple_build_vector): Likewise.
8079 * gimple-fold.c (gimple_build_vector_from_val): New function.
8080 (gimple_build_vector): Likewise.
8081 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8082 functions to build the initial value. Always return a gimple value.
8083 (get_initial_defs_for_reduction): Likewise. Only compute
8085 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8086 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8087 (vectorizable_induction): Use gimple_build_vector rather than
8090 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8091 Alan Hayward <alan.hayward@arm.com>
8092 David Sherwood <david.sherwood@arm.com>
8094 * target.h (vec_perm_indices): New typedef.
8095 (auto_vec_perm_indices): Likewise.
8096 * optabs-query.h: Include target.h
8097 (can_vec_perm_p): Take a vec_perm_indices *.
8098 * optabs-query.c (can_vec_perm_p): Likewise.
8099 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8100 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8101 * tree-vect-generic.c (lower_vec_perm): Likewise.
8102 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8103 (vect_grouped_load_supported): Likewise.
8104 (vect_shift_permute_load_chain): Likewise.
8105 (vect_permute_store_chain): Use auto_vec_perm_indices.
8106 (vect_permute_load_chain): Likewise.
8107 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8108 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8109 Update uses of can_vec_perm_p.
8110 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8111 mode with a number of elements. Take a vec_perm_indices *.
8112 (vect_create_epilog_for_reduction): Update accordingly.
8113 Use auto_vec_perm_indices.
8114 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8115 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8116 (vect_transform_slp_perm_load): Likewise.
8117 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8118 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8119 (vect_gen_perm_mask_checked): Likewise.
8120 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8121 (vect_gen_perm_mask_checked): Likewise.
8122 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8123 (vectorizable_store): Likewise.
8124 (vectorizable_load): Likewise.
8125 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8126 (vectorizable_bswap): Likewise.
8128 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8129 Alan Hayward <alan.hayward@arm.com>
8130 David Sherwood <david.sherwood@arm.com>
8132 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8133 * tree.c (build_vector): Likewise.
8134 (build_vector_from_ctor): Update accordingly.
8135 (build_vector_from_val): Likewise.
8136 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8137 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8138 * tree-vect-generic.c (add_rshift): Likewise.
8139 (expand_vector_divmod): Likewise.
8140 (optimize_vector_constructor): Likewise.
8141 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8142 (vect_transform_slp_perm_load): Likewise.
8143 (vect_schedule_slp_instance): Likewise.
8144 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8145 (vectorizable_call): Likewise.
8146 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8147 * expmed.c (make_tree): Likewise.
8148 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8149 a vector passed to build_vector.
8150 (fold_convert_const): Likewise.
8151 (exact_inverse): Likewise.
8152 (fold_ternary_loc): Likewise.
8153 (fold_relational_const): Likewise.
8154 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8155 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8156 (const_unop): Likewise. Store the reduction accumulator in a
8157 variable rather than an array.
8158 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8159 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8160 the new vector, rather than constructing it after the input arrays.
8161 (native_interpret_vector): Use auto_vec<tree> when building
8162 a vector passed to build_vector. Add elements in order.
8163 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8164 auto_vec<tree> when building a vector passed to build_vector.
8165 (vect_create_epilog_for_reduction): Likewise.
8166 (vectorizable_induction): Likewise.
8167 (get_initial_def_for_reduction): Likewise. Fix indentation of
8169 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8171 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8172 passed to build_vector.
8174 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * tree-core.h (tree_base::u): Add an "nelts" field.
8179 (tree_vector): Use VECTOR_CST_NELTS as the length.
8180 * tree.c (tree_size): Likewise.
8181 (make_vector): Initialize VECTOR_CST_NELTS.
8182 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8183 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8184 TYPE_VECTOR_SUBPARTS.
8185 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8186 as the number of units, setting it from VECTOR_CST_NELTS.
8187 (const_vector_from_tree): Likewise.
8188 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8189 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8190 (fold_negate_expr_1): Likewise.
8191 (fold_convert_const): Likewise.
8192 (const_binop): Likewise. Differentiate the number of output and
8194 (const_unop): Likewise.
8195 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8196 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8197 in cases that did the opposite.
8199 2017-09-14 Richard Biener <rguenther@suse.de>
8201 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8204 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8206 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8208 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8211 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8212 if and where to split a bb, except for splitting before debug insn
8213 sequences followed by non-label real insn. Delete debug insns
8214 in between basic blocks.
8216 * combine.c (make_compound_operation_int): Formatting fixes.
8218 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8219 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8220 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8221 * config/sol2.h (LINK_EH_SPEC): Likewise.
8222 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8223 * config/s390/linux.h (LINK_SPEC): Likewise.
8224 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8225 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8226 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8227 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8228 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8229 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8230 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8231 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8233 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8235 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8237 (ENDFILE_LINUX_SPEC): Likewise.
8238 (LINK_EH_SPEC): Likewise.
8239 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8240 (LINK_OS_LINUX_SPEC32): Likewise.
8241 (LINK_OS_LINUX_SPEC64): Likewise.
8242 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8243 (LINK_OS_LINUX_SPEC): Likewise.
8245 2017-09-13 Martin Liska <mliska@suse.cz>
8248 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8249 CASE_HIGH is NULL_TREE.
8251 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8252 Alan Hayward <alan.hayward@arm.com>
8253 David Sherwood <david.sherwood@arm.com>
8255 * target.def (secondary_memory_needed): New hook.
8256 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8257 instead of SECONDARY_MEMORY_NEEDED.
8258 (secondary_memory_needed_mode): Likewise.
8259 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8260 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8261 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8262 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8263 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8264 * doc/tm.texi: Regenerate.
8265 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8266 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8267 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8268 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8269 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8270 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8271 mode argument first and change the reg_class arguments to reg_class_t.
8272 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8273 Make static. Update the call to inline_secondary_memory_needed.
8274 (ix86_register_move_cost): Update the call to
8275 inline_secondary_memory_needed.
8276 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8277 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8279 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8280 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8282 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8283 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8284 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8285 and match hook interface. Add comment from mips.h.
8286 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8287 * config/mmix/mmix.md (truncdfsf2): Refer to
8288 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8290 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8291 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8292 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8293 (pa_secondary_memory_needed): New function.
8294 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8295 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8296 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8297 (pdp11_secondary_memory_needed): Make static and match hook interface.
8298 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8299 * config/powerpcspe/powerpcspe-protos.h
8300 (rs6000_secondary_memory_needed_ptr): Delete.
8301 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8303 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8304 (rs6000_option_override_internal): Assign to
8305 targetm.secondary_memory_needed rather than
8306 rs6000_secondary_memory_needed_ptr.
8307 (rs6000_secondary_memory_needed): Match hook interface.
8308 (rs6000_debug_secondary_memory_needed): Likewise.
8309 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8310 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8311 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8312 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8313 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8314 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8316 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8317 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8318 (rs6000_option_override_internal): Assign to
8319 targetm.secondary_memory_needed rather than
8320 rs6000_secondary_memory_needed_ptr.
8321 (rs6000_secondary_memory_needed): Match hook interface.
8322 (rs6000_debug_secondary_memory_needed): Likewise.
8323 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8324 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8325 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8326 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8327 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8328 (sparc_secondary_memory_needed): New function.
8329 * lra-constraints.c (check_and_process_move): Refer to
8330 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8332 (curr_insn_transform): Likewise.
8333 (process_alt_operands): Use targetm.secondary_memory_needed
8334 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8335 (check_secondary_memory_needed_p): Likewise.
8336 (choose_split_class): Likewise.
8337 * reload.c: Unconditionally include code that was previously
8338 conditional on SECONDARY_MEMORY_NEEDED.
8339 (push_secondary_reload): Use targetm.secondary_memory_needed
8340 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8341 (push_reload): Likewise.
8342 * reload1.c: Unconditionally include code that was previously
8343 conditional on SECONDARY_MEMORY_NEEDED.
8344 (choose_reload_regs): Use targetm.secondary_memory_needed
8345 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8346 (gen_reload): Likewise.
8347 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8349 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8350 Alan Hayward <alan.hayward@arm.com>
8351 David Sherwood <david.sherwood@arm.com>
8353 * target.def (secondary_memory_needed_mode): New hook:
8354 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8355 * targhooks.h (default_secondary_memory_needed_mode): New function.
8356 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8357 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8358 * doc/tm.texi: Regenerate.
8359 * lra-constraints.c (check_and_process_move): Use
8360 targetm.secondary_memory_needed_mode instead of
8361 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8362 (curr_insn_transform): Likewise.
8363 * reload.c (get_secondary_mem): Likewise.
8364 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8365 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8367 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8368 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8369 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8370 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8371 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8373 * config/powerpcspe/powerpcspe-protos.h
8374 (rs6000_secondary_memory_needed_mode): Delete.
8375 * config/powerpcspe/powerpcspe.c
8376 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8377 (rs6000_secondary_memory_needed_mode): Make static.
8378 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8379 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8381 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8383 (rs6000_secondary_memory_needed_mode): Make static.
8384 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8385 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8386 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8387 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8388 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8390 (sparc_secondary_memory_needed_mode): New function.
8391 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8393 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8395 * config/aarch64/constraints.md (Umq): New constraint.
8396 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8398 (mov<mode>): Update condition.
8400 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8402 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8403 when bitposition is the same.
8405 2017-09-13 Richard Biener <rguenther@suse.de>
8407 * dwarf2out.c (output_die_symbol): Remove.
8408 (output_die): Do not output a DIEs symbol.
8410 2017-09-13 Richard Biener <rguenther@suse.de>
8413 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8414 default-def to avoid breaking iterator update with the weird
8415 interaction with cgraph_update_edges_for_call_stmt_node.
8417 2017-09-13 Richard Biener <rguenther@suse.de>
8419 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8420 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8421 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8422 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8424 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8426 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8427 Disable pc relative literal load irrespective of
8428 TARGET_FIX_ERR_A53_84341 for default.
8430 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8432 * config/sparc/sparc.c (output_return): Output the source location of
8433 the insn in the delay slot, if any.
8434 (output_sibcall): Likewise.
8436 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8439 * common.opt (-static-pie): New alias.
8440 (shared): Negate static-pie.
8441 (-no-pie): Update help text.
8443 (static-pie): New option.
8444 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8445 -static-pie support.
8446 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8447 (LINK_EH_SPEC): Likewise.
8448 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8449 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8450 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8451 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8452 (init_gcc_specs): Likewise.
8453 (init_spec): Likewise.
8454 (display_help): Update help message for -pie.
8455 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8458 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8460 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8461 (movdi_aarch64): Likewise.
8462 (movti_aarch64): Likewise.
8464 2017-09-12 Simon Wright <simon@pushface.org>
8467 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8468 calculation of the minor version, always output as 0.
8470 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8474 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8475 on it early, rather than manual conversion late. For
8476 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8477 instead of performing manual conversion.
8479 2017-09-12 Carl Love <cel@us.ibm.com>
8481 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8482 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8484 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8485 VMULOSW): Add definitions.
8486 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8487 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8488 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8490 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8492 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8494 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8496 (movti_aarch64): Likewise.
8497 (movdf_aarch64): Likewise.
8498 (movtf_aarch64): Likewise.
8499 (load_pairdi): Likewise.
8500 (store_pairdi): Likewise.
8501 (load_pairdf): Likewise.
8502 (store_pairdf): Likewise.
8503 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8504 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8505 (ldr_got_small_<mode>): Likewise.
8506 (ldr_got_small_28k_<mode>): Likewise.
8507 (ldr_got_tiny): Likewise.
8508 * config/aarch64/iterators.md (ldst_sz): New.
8509 (ldpstp_sz): Likewise.
8510 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8512 (thunderx_load): Split load_8 to load_16.
8513 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8515 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8516 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8517 (xgene1_store_pair): Split store_8 to store_16.
8518 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8519 (falkor_st_0_st_sd): Split store_8 to store_16.
8521 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8523 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8524 and store1/2/3/4 to store_4/8/12/16.
8525 * config/aarch64/aarch64.md: Update for rename.
8526 * config/arm/arm.md: Likewise.: Likewise.
8527 * config/arm/arm.c: Likewise.
8528 * config/arm/thumb1.md: Likewise.
8529 * config/arm/thumb2.md: Likewise.
8530 * config/arm/vfp.md: Likewise.
8531 * config/arm/arm-generic.md: Likewise.
8532 * config/arm/arm1020e.md: Likewise.
8533 * config/arm/arm1026ejs.md: Likewise.
8534 * config/arm/arm1136jfs.md: Likewise.
8535 * config/arm/arm926ejs.md: Likewise.
8536 * config/arm/cortex-a15.md: Likewise.
8537 * config/arm/cortex-a17.md: Likewise.
8538 * config/arm/cortex-a5.md: Likewise.
8539 * config/arm/cortex-a53.md: Likewise.
8540 * config/arm/cortex-a57.md: Likewise.
8541 * config/arm/cortex-a7.md: Likewise.
8542 * config/arm/cortex-a8.md: Likewise.
8543 * config/arm/cortex-a9.md: Likewise.
8544 * config/arm/cortex-m4.md: Likewise.
8545 * config/arm/cortex-m7.md: Likewise.
8546 * config/arm/cortex-r4.md: Likewise.
8547 * config/arm/exynos-m1.md: Likewise.
8548 * config/arm/fa526.md: Likewise.
8549 * config/arm/fa606te.md: Likewise.
8550 * config/arm/fa626te.md: Likewise.
8551 * config/arm/fa726te.md: Likewise.
8552 * config/arm/fmp626.md: Likewise.
8553 * config/arm/iwmmxt.md: Likewise.
8554 * config/arm/ldmstm.md: Likewise.
8555 * config/arm/marvell-pj4.md: Likewise.
8556 * config/arm/xgene1.md: Likewise.
8557 * config/aarch64/thunderx.md: Likewise.
8558 * config/aarch64/thunderx2t99.md: Likewise.
8559 * config/aarch64/falkor.md: Likewise.
8561 2017-09-12 Martin Liska <mliska@suse.cz>
8563 * attribs.c (private_lookup_attribute): New function.
8564 * attribs.h (private_lookup_attribute): Declared here.
8565 (lookup_attribute): Called from this place.
8567 2017-09-12 Richard Biener <rguenther@suse.de>
8569 PR tree-optimization/82157
8570 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8571 stmts with side-effects.
8573 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8574 Alan Hayward <alan.hayward@arm.com>
8575 David Sherwood <david.sherwood@arm.com>
8577 * target.def (hard_regno_nregs): New hook.
8578 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8579 * targhooks.h (default_hard_regno_nregs): Declare.
8580 * targhooks.c (default_hard_regno_nregs): New function.
8581 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8582 (TARGET_HARD_REGNO_NREGS): ...this hook.
8583 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8584 (CLASS_MAX_NREGS): Likewise.
8585 * doc/tm.texi: Regenerate.
8586 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8587 instead of HARD_REGNO_NREGS.
8588 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8589 HARD_REGNO_NREGS in the comment.
8590 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8591 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8592 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8593 Return an unsigned int.
8594 (TARGET_HARD_REGNO_NREGS): Redefine.
8595 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8596 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8597 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8598 (arc_hard_regno_nregs): New function.
8599 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8600 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8601 (arm_hard_regno_nregs): New function.
8602 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8603 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8604 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8605 (TARGET_HARD_REGNO_NREGS): Redefine.
8606 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8607 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8608 (HARD_REGNO_NREGS): Delete.
8609 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8610 (cr16_hard_regno_nregs): New function.
8611 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8612 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8613 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8614 (cris_hard_regno_nregs): New function.
8615 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8616 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8617 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8618 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8619 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8620 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8621 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8622 (frv_hard_regno_nregs): Make static. Take and return an
8624 (frv_class_max_nregs): Remove outdated copy of documentation.
8625 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8626 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8627 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8628 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8629 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8630 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8631 (TARGET_HARD_REGNO_NREGS): Redefine.
8632 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8633 (CLASS_MAX_NREGS): Update comment.
8634 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8635 (ia64_hard_regno_nregs): New function.
8636 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8637 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8638 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8639 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8640 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8642 (m32c_hard_regno_nregs): Likewise. Make static.
8643 (TARGET_HARD_REGNO_NREGS): Redefine.
8644 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8645 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8646 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8647 (m68k_hard_regno_nregs): New function.
8648 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8649 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8650 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8651 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8652 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8653 Take and return an unsigned int.
8654 (TARGET_HARD_REGNO_NREGS): Redefine.
8655 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8656 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8657 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8658 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8659 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8660 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8661 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8662 (msp430_hard_regno_nregs): Make static. Take and return an
8664 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8665 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8666 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8667 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8668 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8669 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8670 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8671 (TARGET_HARD_REGNO_NREGS): Redefine.
8672 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8673 (PA_HARD_REGNO_NREGS): ...this.
8674 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8675 (PA_HARD_REGNO_NREGS): ...this.
8676 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8677 (pa_hard_regno_nregs): New function.
8678 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8679 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8680 (pdp11_hard_regno_nregs): New function.
8681 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8682 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8683 (rs6000_hard_regno_nregs_hook): New function.
8684 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8685 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8686 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8687 Take and return an unsigned int. Move earlier in file.
8688 (TARGET_HARD_REGNO_NREGS): Redefine.
8689 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8690 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8691 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8692 (rl78_hard_regno_nregs): Make static. Take and return an
8694 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8695 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8696 (rs6000_hard_regno_nregs_hook): New function.
8697 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8698 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8699 (TARGET_HARD_REGNO_NREGS): Redefine.
8700 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8701 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8702 instead of HARD_REGNO_NREGS.
8703 (s390_hard_regno_nregs): New function.
8704 (s390_hard_regno_mode_ok): Add comment from s390.h.
8705 (TARGET_HARD_REGNO_NREGS): Redefine.
8706 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8707 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8708 (sh_hard_regno_nregs): New function.
8709 (sh_pass_in_reg_p): Use it.
8710 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8711 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8712 (sparc_hard_regno_nregs): New function.
8713 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8714 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8715 (spu_function_arg_advance): Use it, supplying a valid register number.
8716 (TARGET_HARD_REGNO_NREGS): Redefine.
8717 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8718 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8719 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8720 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8721 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8722 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8723 (CLASS_MAX_NREGS): Remove copy of old documentation.
8724 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8725 (visium_hard_regno_nregs): New function.
8726 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8727 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8728 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8729 xtensa_hard_regno_nregs): New function.
8730 * system.h (HARD_REGNO_NREGS): Poison.
8732 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8734 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8735 hard_regno_nregs instead of HARD_REGNO_NREGS.
8736 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8737 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8738 (c6x_expand_epilogue): Likewise.
8739 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8740 (frv_read_iacc_argument): Likewise.
8741 * config/sh/sh.c: Include regs.h.
8742 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8743 (regs_used): Likewise.
8744 (output_stack_adjust): Likewise.
8745 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8746 * expmed.c: Include regs.h.
8747 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8748 * ree.c: Include regs.h.
8749 (combine_reaching_defs): Use hard_regno_nregs instead of
8751 (add_removable_extension): Likewise.
8753 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8755 * regs.h (hard_regno_nregs): Turn into a function.
8756 (end_hard_regno): Update accordingly.
8757 * caller-save.c (setup_save_areas): Likewise.
8758 (save_call_clobbered_regs): Likewise.
8759 (replace_reg_with_saved_mem): Likewise.
8760 (insert_restore): Likewise.
8761 (insert_save): Likewise.
8762 * combine.c (can_change_dest_mode): Likewise.
8763 (move_deaths): Likewise.
8764 (distribute_notes): Likewise.
8765 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8766 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8767 (rs6000_split_multireg_move): Likewise.
8768 (rs6000_register_move_cost): Likewise.
8769 (rs6000_memory_move_cost): Likewise.
8770 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8771 (rs6000_split_multireg_move): Likewise.
8772 (rs6000_register_move_cost): Likewise.
8773 (rs6000_memory_move_cost): Likewise.
8774 * cselib.c (cselib_reset_table): Likewise.
8775 (cselib_lookup_1): Likewise.
8776 * emit-rtl.c (set_mode_and_regno): Likewise.
8777 * function.c (aggregate_value_p): Likewise.
8778 * ira-color.c (setup_profitable_hard_regs): Likewise.
8779 (check_hard_reg_p): Likewise.
8780 (calculate_saved_nregs): Likewise.
8781 (assign_hard_reg): Likewise.
8782 (improve_allocation): Likewise.
8783 (calculate_spill_cost): Likewise.
8784 * ira-emit.c (modify_move_list): Likewise.
8785 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8786 (ira_hard_reg_in_set_p): Likewise.
8787 * ira.c (setup_reg_mode_hard_regset): Likewise.
8788 (clarify_prohibited_class_mode_regs): Likewise.
8789 (check_allocation): Likewise.
8790 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8791 (lra_setup_reg_renumber): Likewise.
8792 (setup_try_hard_regno_pseudos): Likewise.
8793 (spill_for): Likewise.
8794 (assign_hard_regno): Likewise.
8795 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8796 * lra-constraints.c (in_class_p): Likewise.
8797 (lra_constraint_offset): Likewise.
8798 (simplify_operand_subreg): Likewise.
8799 (lra_constraints): Likewise.
8800 (split_reg): Likewise.
8801 (split_if_necessary): Likewise.
8802 (invariant_p): Likewise.
8803 (inherit_in_ebb): Likewise.
8804 * lra-lives.c (process_bb_lives): Likewise.
8805 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8806 (get_hard_regs): Likewise.
8807 (do_remat): Likewise.
8808 * lra-spills.c (assign_spill_hard_regs): Likewise.
8809 * mode-switching.c (create_pre_exit): Likewise.
8810 * postreload.c (reload_combine_recognize_pattern): Likewise.
8811 * recog.c (peep2_find_free_register): Likewise.
8812 * regcprop.c (kill_value_regno): Likewise.
8813 (set_value_regno): Likewise.
8814 (copy_value): Likewise.
8815 (maybe_mode_change): Likewise.
8816 (find_oldest_value_reg): Likewise.
8817 (copyprop_hardreg_forward_1): Likewise.
8818 * regrename.c (check_new_reg_p): Likewise.
8819 (regrename_do_replace): Likewise.
8820 * reload.c (push_reload): Likewise.
8821 (combine_reloads): Likewise.
8822 (find_dummy_reload): Likewise.
8823 (operands_match_p): Likewise.
8824 (find_reloads): Likewise.
8825 (find_equiv_reg): Likewise.
8826 (reload_adjust_reg_for_mode): Likewise.
8827 * reload1.c (count_pseudo): Likewise.
8828 (count_spilled_pseudo): Likewise.
8829 (find_reg): Likewise.
8830 (clear_reload_reg_in_use): Likewise.
8831 (free_for_value_p): Likewise.
8832 (allocate_reload_reg): Likewise.
8833 (choose_reload_regs): Likewise.
8834 (reload_adjust_reg_for_temp): Likewise.
8835 (emit_reload_insns): Likewise.
8836 (delete_output_reload): Likewise.
8837 * rtlanal.c (subreg_get_info): Likewise.
8838 * sched-deps.c (sched_analyze_reg): Likewise.
8839 * sel-sched.c (init_regs_for_mode): Likewise.
8840 (mark_unavailable_hard_regs): Likewise.
8841 (choose_best_reg_1): Likewise.
8842 (verify_target_availability): Likewise.
8843 * valtrack.c (dead_debug_insert_temp): Likewise.
8844 * var-tracking.c (track_loc_p): Likewise.
8845 (emit_note_insn_var_location): Likewise.
8846 * varasm.c (make_decl_rtl): Likewise.
8847 * reginfo.c (choose_hard_reg_mode): Likewise.
8848 (init_reg_modes_target): Refer directly to
8849 this_target_regs->x_hard_regno_nregs.
8851 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8853 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8854 instead of hard_regno_nregs.
8856 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8858 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8859 end_hard_regno instead of hard_regno_nregs.
8860 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8861 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8862 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8863 * ira-color.c (improve_allocation): Likewise.
8864 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8865 * lra-lives.c (mark_regno_live): Likewise.
8866 (mark_regno_dead): Likewise.
8867 * lra-remat.c (operand_to_remat): Likewise.
8868 * lra.c (collect_non_operand_hard_regs): Likewise.
8869 * postreload.c (reload_combine_note_store): Likewise.
8870 (move2add_valid_value_p): Likewise.
8871 * reload.c (regno_clobbered_p): Likewise.
8873 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8875 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8877 * config/v850/v850.c (v850_reorg): Likewise.
8878 * reload.c (refers_to_regno_for_reload_p): Likewise.
8879 (find_equiv_reg): Likewise.
8880 * reload1.c (reload_reg_reaches_end_p): Likewise.
8882 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8884 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8886 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8887 * config/arm/arm.c (output_move_neon): Likewise.
8888 (arm_attr_length_move_neon): Likewise.
8889 (neon_split_vcombine): Likewise.
8890 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8891 (c6x_mark_reg_written): Likewise.
8892 (c6x_dwarf_register_span): Likewise.
8893 * config/i386/i386.c (ix86_save_reg): Likewise.
8894 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8895 (rws_access_reg): Likewise.
8896 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8897 * mode-switching.c (create_pre_exit): Likewise.
8898 * ree.c (combine_reaching_defs): Likewise.
8899 (add_removable_extension): Likewise.
8900 * regcprop.c (find_oldest_value_reg): Likewise.
8901 (copyprop_hardreg_forward_1): Likewise.
8902 * reload.c (reload_inner_reg_of_subreg): Likewise.
8903 (push_reload): Likewise.
8904 (combine_reloads): Likewise.
8905 (find_dummy_reload): Likewise.
8906 (reload_adjust_reg_for_mode): Likewise.
8907 * reload1.c (find_reload_regs): Likewise.
8908 (forget_old_reloads_1): Likewise.
8909 (reload_reg_free_for_value_p): Likewise.
8910 (reload_adjust_reg_for_temp): Likewise.
8911 (emit_reload_insns): Likewise.
8912 (delete_output_reload): Likewise.
8913 * sel-sched.c (choose_best_reg_1): Likewise.
8914 (choose_best_pseudo_reg): Likewise.
8916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8917 Alan Hayward <alan.hayward@arm.com>
8918 David Sherwood <david.sherwood@arm.com>
8920 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8921 * target.def (slow_unaligned_access): New hook.
8922 * targhooks.h (default_slow_unaligned_access): Declare.
8923 * targhooks.c (default_slow_unaligned_access): New function.
8924 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8925 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8926 * doc/tm.texi: Regenerate.
8927 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8928 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8929 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8931 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8932 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8934 (rs6000_slow_unaligned_access): New function.
8935 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8936 (expand_block_compare): Likewise.
8937 (expand_strn_compare): Likewise.
8938 (rs6000_rtx_costs): Likewise.
8939 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8940 (riscv_slow_unaligned_access): Likewise.
8941 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8942 (riscv_slow_unaligned_access_p): ...this and make static.
8943 (riscv_option_override): Update accordingly.
8944 (riscv_slow_unaligned_access): New function.
8945 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8946 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8947 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8948 (rs6000_slow_unaligned_access): New function.
8949 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8950 (rs6000_rtx_costs): Likewise.
8951 * config/rs6000/rs6000-string.c (expand_block_compare)
8952 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8953 of SLOW_UNALIGNED_ACCESS.
8954 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8955 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8956 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8957 of SLOW_UNALIGNED_ACCESS.
8958 * expmed.c (simple_mem_bitfield_p): Likewise.
8959 * expr.c (alignment_for_piecewise_move): Likewise.
8960 (emit_group_load_1): Likewise.
8961 (emit_group_store): Likewise.
8962 (copy_blkmode_from_reg): Likewise.
8963 (emit_push_insn): Likewise.
8964 (expand_assignment): Likewise.
8965 (store_field): Likewise.
8966 (expand_expr_real_1): Likewise.
8967 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8968 * lra-constraints.c (simplify_operand_subreg): Likewise.
8969 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8970 * gimple-ssa-store-merging.c: Likewise in block comment at start
8972 * tree-ssa-strlen.c: Include target.h.
8973 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8974 of SLOW_UNALIGNED_ACCESS.
8975 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8977 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8979 PR rtl-optimization/82185
8980 * expmed.c (emit_store_flag_int): Only test tem if it has been
8983 2017-09-12 Richard Biener <rguenther@suse.de>
8986 * match.pd ((FTYPE) N CMP CST): Fix typo.
8988 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8990 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8992 (mips_near_type_p): Add 'short_call' attribute as a synonym
8994 * doc/extend.texi (short_call): Document new function attribute.
8996 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8999 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9000 assertion check that in the condition.
9001 (get_atomic_generic_size): Likewise. Before testing if parameter
9002 has pointer type, if it has array type, call for C++
9003 default_conversion to perform array-to-pointer conversion.
9005 2017-09-12 Richard Biener <rguenther@suse.de>
9007 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9008 for operations we cannot scalarize.
9010 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9012 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9013 vectors heap vectors. Clean up comments.
9014 Make visited_bbs a reference.
9015 (profitable_jump_thread_path): Make GC
9016 vectors heap vectors. Clean up comments.
9018 (convert_and_register_jump_thread_path): Make GC vectors heap
9020 (check_subpath_and_update_thread_path): Same. Clean up comments.
9021 Make visited_bbs a reference.
9022 (handle_phi): Abstract common code to to
9023 register_jump_thread_path_if_profitable.
9024 Rename VAR_BB to DEF_BB.
9026 Make GC vectors heap vectors.
9027 Make visited_bbs a reference.
9028 (handle_assignment): Same.
9029 (register_jump_thread_path_if_profitable): New.
9030 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9032 Make GC vectors heap vectors. Clean up comments.
9033 Make visited_bbs a reference.
9034 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9035 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9038 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9041 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9042 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9044 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9046 Revert r251800 and r251799.
9048 2017-09-11 Martin Jambor <mjambor@suse.cz>
9051 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9052 arguments in advance.
9053 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9054 use it to find predecessor edges.
9055 (naive_outof_ssa): Collect vector of predecessors.
9057 2017-09-08 Jason Merrill <jason@redhat.com>
9059 PR c++/70029 - ICE with ref-qualifier and -flto
9060 * langhooks.h (struct lang_hooks_for_types): Add
9061 copy_lang_qualifiers.
9062 * attribs.c (build_type_attribute_qual_variant): Use it.
9063 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9065 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9066 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9068 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9071 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9072 (*mulsi3_sp64): New instruction.
9073 (mulsi3): New expander.
9075 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9077 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9079 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9081 * sancov.c: Include memmodel.h.
9083 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9086 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9089 2017-09-07 Carl Love <cel@us.ibm.com>
9091 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9092 the sldi instruction.
9094 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9096 * sancov.c: Include tm_p.h.
9098 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9101 * output.h (switch_to_other_text_partition): New declaration.
9102 * varasm.c (switch_to_other_text_partition): New function.
9103 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9104 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9105 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9106 to the other text partition before emitting LCL label and switch back
9107 after emitting the word after it.
9109 2017-09-07 Richard Biener <rguenther@suse.de>
9111 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9112 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9113 critical edges here, after loop init.
9114 (pass_data_pre): Remove PROP_no_crit_edges flags.
9115 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9116 for valueization of call args to avoid leaking VN_TOP.
9117 (visit_use): Assert we do not visit default defs.
9118 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9119 Use error_mark_node to more easily detect leaking VN_TOP.
9120 All default-defs are varying, not VN_TOP. Mark them visited.
9121 (run_scc_vn): Make code match comment.
9123 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9125 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9126 OPTION_MASK_FLOAT128_KEYWORD.
9127 (POWERPC_MASKS): Likewise.
9128 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9129 support for the -mfloat128-type option, and make -mfloat128
9130 default on PowerPC Linux systems. Define or undefine
9131 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9132 Define __float128 to be __ieee128 if IEEE 128-bit support is
9133 enabled, or undefine it.
9134 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9135 Delete defining __FLOAT128_TYPE__.
9136 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9137 -mfloat128-type option and make -mfloat128 default on PowerPC
9139 (TARGET_FLOAT128_TYPE): Likewise.
9140 (-mfloat128-type): Likewise.
9141 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9142 Delete the -mfloat128-type option and make -mfloat128 default on
9143 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9144 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9145 128-bit floating point is enabled. Change tests from using
9146 -mfloat128-type to -mfloat128.
9147 (rs6000_mangle_type): Use the correct mangling for the __float128
9148 type even if normal long double is restricted to 64-bits.
9149 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9151 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9152 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9153 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9155 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9156 documentation for -mfloat128.
9158 2017-09-06 Olivier Hainque <hainque@adacore.com>
9160 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9162 2017-09-06 Wish Wu <wishwu007@gmail.com>
9163 Jakub Jelinek <jakub@redhat.com>
9165 * asan.c (initialize_sanitizer_builtins): Add
9166 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9167 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9168 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9169 BT_FN_VOID_UINT64_PTR variables.
9170 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9171 (BT_FN_VOID_UINT16_UINT16): Likewise.
9172 (BT_FN_VOID_UINT32_UINT32): Likewise.
9173 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9174 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9175 (BT_FN_VOID_UINT64_PTR): Likewise.
9176 * common.opt (flag_sanitize_coverage): New variable.
9177 (fsanitize-coverage=trace-pc): Remove.
9178 (fsanitize-coverage=): Add.
9179 * flag-types.h (enum sanitize_coverage_code): New enum.
9180 * fold-const.c (fold_range_test): Disable non-short-circuit
9181 optimization if flag_sanitize_coverage.
9182 (fold_truth_andor): Likewise.
9183 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9184 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9185 (coverage_sanitizer_opts): New array.
9186 (get_closest_sanitizer_option): Add OPTS argument, handle also
9187 OPT_fsanitize_coverage_.
9188 (parse_sanitizer_options): Adjusted to also handle
9189 OPT_fsanitize_coverage_.
9190 (common_handle_option): Add OPT_fsanitize_coverage_.
9191 * sancov.c (instrument_comparison, instrument_switch): New function.
9192 (sancov_pass): Add trace-cmp support.
9193 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9194 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9195 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9196 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9197 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9198 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9199 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9200 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9201 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9202 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9204 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9206 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9207 error. Only quit immediately if parsing is complete.
9208 (BEGIN): Initialize fatal_err and parse_done.
9209 (begin fpu, end fpu): Check number of arguments.
9210 (begin arch, end arch): Likewise.
9211 (begin cpu, end cpu): Likewise.
9212 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9213 (optalias): Likewise.
9215 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9217 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9218 * config/arm/arm-isa.h: Delete. Move definitions to ...
9219 * arm-cpus.in: ... here. Use new feature and fgroup values.
9220 * config/arm/arm.c (arm_option_override): Use lower case for feature
9222 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9223 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9224 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9226 (print_isa_bits_for): New function.
9227 (gen_isa): New function.
9228 (gen_comm_data): Use print_isa_bits_for.
9229 (define feature): New keyword.
9230 (define fgroup): New keyword.
9231 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9232 (arm-isa.h): Add rule to generate file.
9233 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9234 case for feature bit names.
9236 2017-09-06 Richard Biener <rguenther@suse.de>
9238 * tree-ssa-pre.c (NECESSARY): Remove.
9239 (create_expression_by_pieces): Do not touch pass-local flags.
9240 (insert_into_preds_of_block): Likewise.
9241 (do_pre_regular_insertion): Likewise.
9242 (eliminate_insert): Likewise.
9243 (eliminate_dom_walker::before_dom_children): Likewise.
9244 (fini_eliminate): Do not look at inserted_exprs.
9245 (mark_operand_necessary): Remove.
9246 (remove_dead_inserted_code): Replace with simple work-list
9247 algorithm based on inserted_exprs and SSA uses.
9248 (pass_pre::execute): Re-order fini_eliminate and
9249 remove_dead_inserted_code.
9251 2017-09-06 Olivier Hainque <hainque@adacore.com>
9253 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9254 for VxWorks 7. Adjust surrounding comments.
9256 2017-09-06 Richard Biener <rguenther@suse.de>
9258 * gimple-ssa-strength-reduction.c
9259 (find_candidates_dom_walker::before_dom_children): Also allow
9262 2017-09-06 Richard Biener <rguenther@suse.de>
9264 PR tree-optimization/82108
9265 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9266 for gap in the non-permutation SLP case.
9268 2017-09-06 Martin Jambor <mjambor@suse.cz>
9270 PR tree-optimization/82078
9271 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9272 add_access_to_work_queue...
9273 (build_accesses_from_assign): ...here.
9274 (propagate_all_subaccesses): Make sure racc is the group
9275 representative, if there is one.
9277 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9280 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9283 2017-09-06 Richard Biener <rguenther@suse.de>
9285 * gimple-ssa-strength-reduction.c
9286 (find_candidates_dom_walker::before_doom_children): Use a
9287 type and not a mode check.
9289 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9292 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9294 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9296 2017-09-05 Jeff Law <law@redhat.com>
9298 PR tree-optimization/64910
9299 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9300 cases where we have 3 or more operands.
9302 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9305 * omp-low.c (lower_omp_for): Recompute tree invariant if
9306 gimple_omp_for_initial/final is ADDR_EXPR.
9309 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9310 into gimple val before gimplification fo the COND_EXPR.
9312 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9314 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9315 REGION_COPY argument.
9316 (thread_through_all_blocks): Remove unused argument to
9317 duplicate_thread_path.
9319 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9320 Alan Hayward <alan.hayward@arm.com>
9321 David Sherwood <david.sherwood@arm.com>
9323 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9324 Take a scalar_mode rather than a machine_mode.
9325 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9326 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9327 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9328 (aarch64_gen_adjusted_ldpstp): Likewise.
9329 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9331 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9332 Alan Hayward <alan.hayward@arm.com>
9333 David Sherwood <david.sherwood@arm.com>
9335 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9336 Take a scalar_int_mode instead of a machine_mode.
9337 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9338 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9339 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9340 (aarch64_simd_attr_length_rglist): Delete.
9341 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9342 a scalar_int_mode instead of a machine_mode.
9343 (aarch64_add_offset): Likewise.
9344 (aarch64_internal_mov_immediate): Likewise
9345 (aarch64_add_constant_internal): Likewise.
9346 (aarch64_add_constant): Likewise.
9347 (aarch64_movw_imm): Likewise.
9348 (aarch64_rtx_arith_op_extract_p): Likewise.
9349 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9350 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9351 Remove assert that the mode isn't a vector.
9352 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9353 (aarch64_expand_mov_immediate): Update calls after above changes.
9354 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9355 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9356 (aarch64_move_imm): Likewise.
9357 (aarch64_can_const_movi_rtx_p): Likewise.
9358 (aarch64_strip_extend): Likewise.
9359 (aarch64_extr_rtx_p): Likewise.
9360 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9361 a CONST_INT when the mode parameter is VOIDmode.
9362 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9364 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9366 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9367 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9368 (bitwise_type_for_mode): Update accordingly.
9370 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9372 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9373 * stor-layout.c (mode_for_size_tree): Likewise.
9374 (mode_for_array): Update accordingly.
9375 (layout_decl): Likewise.
9376 (compute_record_mode): Likewise. Only set the mode once.
9378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9380 * target.def (get_mask_mode): Change return type to opt_mode.
9382 * doc/tm.texi: Regenerate.
9383 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9384 * targhooks.c (default_get_mask_mode): Likewise.
9385 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9386 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9387 targetm.get_mask_mode.
9388 * tree.c (build_truth_vector_type): Likewise.
9390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9392 * machmode.h (mode_for_vector): Return an opt_mode.
9393 * stor-layout.c (mode_for_vector): Likewise.
9394 (mode_for_int_vector): Update accordingly.
9395 (layout_type): Likewise.
9396 * config/i386/i386.c (emit_memmov): Likewise.
9397 (ix86_expand_set_or_movmem): Likewise.
9398 (ix86_expand_vector_init): Likewise.
9399 (ix86_get_mask_mode): Likewise.
9400 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9402 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9403 * expmed.c (extract_bit_field_1): Likewise.
9404 * expr.c (expand_expr_real_2): Likewise.
9405 * optabs-query.c (can_vec_perm_p): Likewise.
9406 (can_vec_mask_load_store_p): Likewise.
9407 * optabs.c (expand_vec_perm): Likewise.
9408 * targhooks.c (default_get_mask_mode): Likewise.
9409 * tree-vect-stmts.c (vectorizable_store): Likewise.
9410 (vectorizable_load): Likewise.
9411 (get_vectype_for_scalar_type_and_size): Likewise.
9413 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9415 * machmode.h (mode_for_int_vector): New function.
9416 * stor-layout.c (mode_for_int_vector): Likewise.
9417 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9418 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9419 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9420 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9421 (s390_expand_vcond): Likewise.
9423 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9425 * machmode.h (opt_machine_mode): New type.
9426 (opt_mode<T>): Allow construction from anything that can be
9428 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9429 (mode_for_size): Return an opt_machine_mode.
9430 * stor-layout.c (mode_for_size): Likewise.
9431 (mode_for_size_tree): Update call accordingly.
9432 (bitwise_mode_for_mode): Likewise.
9433 (make_fract_type): Likewise.
9434 (make_accum_type): Likewise.
9435 * caller-save.c (replace_reg_with_saved_mem): Update call
9437 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9438 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9439 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9440 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9441 * expmed.c (extract_bit_field_1): Likewise.
9442 * reload.c (get_secondary_mem): Likewise.
9443 * varasm.c (assemble_integer): Likewise.
9444 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9447 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9449 * machmode.h (decimal_float_mode_for_size): New function.
9450 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9451 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9452 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9453 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9454 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9455 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9457 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9459 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9460 (get_builtin_sync_mode): Likewise.
9461 (expand_ifn_atomic_compare_exchange): Likewise.
9462 (expand_builtin_atomic_clear): Likewise.
9463 (expand_builtin_atomic_test_and_set): Likewise.
9464 (fold_builtin_atomic_always_lock_free): Likewise.
9465 * calls.c (compute_argument_addresses): Likewise.
9466 (emit_library_call_value_1): Likewise.
9467 (store_one_arg): Likewise.
9468 * combine.c (combine_instructions): Likewise.
9469 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9470 * config/arm/arm.c (arm_function_value): Likewise.
9471 (aapcs_allocate_return_reg): Likewise.
9472 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9473 * config/i386/i386.c (construct_container): Likewise.
9474 (ix86_gimplify_va_arg): Likewise.
9475 (ix86_expand_sse_cmp): Likewise.
9476 (emit_memmov): Likewise.
9477 (emit_memset): Likewise.
9478 (expand_small_movmem_or_setmem): Likewise.
9479 (ix86_expand_pextr): Likewise.
9480 (ix86_expand_pinsr): Likewise.
9481 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9482 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9484 * config/mips/mips.c (mips_function_value_1) Likewise.
9485 (mips_block_move_straight): Likewise.
9486 (mips_expand_ins_as_unaligned_store): Likewise.
9487 * config/powerpcspe/powerpcspe.c
9488 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9489 (rs6000_darwin64_record_arg_flush): Likewise.
9490 * config/rs6000/rs6000.c
9491 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9492 (rs6000_darwin64_record_arg_flush): Likewise.
9493 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9494 (sparc_function_value_1): Likewise.
9495 * config/spu/spu.c (adjust_operand): Likewise.
9496 (spu_emit_branch_or_set): Likewise.
9497 (arith_immediate_p): Likewise.
9498 * emit-rtl.c (gen_lowpart_common): Likewise.
9499 * expr.c (expand_expr_real_1): Likewise.
9500 * function.c (assign_parm_setup_block): Likewise.
9501 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9502 * reload1.c (alter_reg): Likewise.
9503 * stor-layout.c (mode_for_vector): Likewise.
9504 (layout_type): Likewise.
9506 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9508 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9509 (spu_convert_move): Likewise.
9510 * lower-subreg.c (resolve_simple_move): Likewise.
9512 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9515 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9516 define_insn to a define_expand.
9517 (altivec_vsum2sws_direct): New define_insn.
9518 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9520 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9522 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9525 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9530 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9531 realign stack if stack alignment needed is less than incoming
9534 2017-09-05 Marek Polacek <polacek@redhat.com>
9537 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9540 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9542 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9544 2017-09-05 Richard Biener <rguenther@suse.de>
9546 PR tree-optimization/82084
9547 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9549 2017-09-05 Richard Biener <rguenther@suse.de>
9551 PR tree-optimization/82102
9552 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9554 2017-09-05 Martin Liska <mliska@suse.cz>
9556 PR tree-optimization/82032
9557 * tree-cfg.c (generate_range_test): New function.
9558 * tree-cfg.h (generate_range_test): Declared here.
9559 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9560 (cleanup_control_expr_graph): Use it.
9561 * tree-switch-conversion.c (try_switch_expansion): Remove
9563 (emit_case_nodes): Use generate_range_test.
9565 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9568 * config/i386/i386.md (*<btsc><mode>_mask): Add
9569 TARGET_USE_BT to insn constraint.
9570 (*btr<mode>_mask): Ditto.
9572 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9574 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9575 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9577 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9580 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9581 TARGET_NEON and TARGET_IWMMXT.
9582 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9583 TARGET_NEON and TARGET_IWMMXT.
9584 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9586 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9588 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9589 (ix86_rewrite_tls_address): Ditto.
9590 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9591 (ix86_rewrite_tls_address_1): Ditto.
9592 (ix86_rewrite_tls_address): Ditto.
9593 * config/i386/predicates.md (tls_address_pattern): New predicate.
9594 * config/i386/i386.md (TLS address splitter): New splitter.
9596 2017-09-04 Richard Biener <rguenther@suse.de>
9598 PR tree-optimization/82084
9599 * fold-const.h (can_native_encode_string_p): Declare.
9600 * fold-const.c (can_native_encode_string_p): Factor out from ...
9601 (native_encode_string): ... here.
9602 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9603 vectorizing stores from constants we later cannot handle.
9605 2017-09-04 Marek Polacek <polacek@redhat.com>
9608 * doc/invoke.texi: Update -Wtautological-compare documentation.
9610 2017-09-04 Jeff Law <law@redhat.com>
9612 PR tree-optimization/64910
9613 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9614 swap the first and last operand if the last is a constant.
9616 2017-09-04 Marek Polacek <polacek@redhat.com>
9619 * convert.c (do_narrow): When sanitizing signed integer overflows,
9620 bail out for signed types.
9621 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9623 2017-09-04 Richard Biener <rguenther@suse.de>
9625 PR tree-optimization/82060
9626 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9627 Move devirtualization after stmt folding and before EH/AB/noreturn
9628 cleanup to get the stmt refs canonicalized. Use a bool instead
9629 of gimple_modified_p since that doesn't work for NOPs. Schedule
9630 NOPs generated by folding for removal.
9632 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9633 Alan Hayward <alan.hayward@arm.com>
9634 David Sherwood <david.sherwood@arm.com>
9636 * coretypes.h (pad_direction): New enum.
9637 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9638 (FUNCTION_ARG_PADDING): Likewise.
9639 * target.def (function_arg_padding): New hook.
9640 * targhooks.h (default_function_arg_padding): Declare.
9641 * targhooks.c (default_function_arg_padding): New function.
9642 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9643 (TARGET_FUNCTION_ARG_PADDING): ...this.
9644 * doc/tm.texi: Regenerate.
9645 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9646 instead of direction.
9647 (compute_argument_addresses): Likewise.
9648 (load_register_parameters): Likewise.
9649 (emit_library_call_value_1): Likewise.
9650 (store_one_arg): Use targetm.calls.function_arg_padding instead
9651 of FUNCTION_ARG_PADDING.
9652 (must_pass_in_stack_var_size_or_pad): Likewise.
9653 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9654 (emit_group_store): Likewise.
9655 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9656 instead of FUNCTION_ARG_PADDING.
9657 (emit_push_insn): Likewise, and propagate enum change throughout
9659 * function.h (direction): Delete.
9660 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9662 * function.c (assign_parm_find_stack_rtl): Likewise.
9663 (assign_parm_setup_block_p): Likewise.
9664 (assign_parm_setup_block): Likewise.
9665 (gimplify_parameters): Likewise.
9666 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9667 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9669 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9670 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9671 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9672 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9673 (aarch64_function_arg_padding): ...this new function.
9674 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9675 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9676 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9677 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9678 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9679 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9680 (arm_pad_arg_upward): Replace with...
9681 (arm_function_arg_padding): ...this new function.
9682 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9684 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9685 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9686 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9687 (ia64_hpux_function_arg_padding): Replace with...
9688 (ia64_function_arg_padding): ...this new function. Use pad_direction
9689 instead of direction. Check for TARGET_HPUX.
9690 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9691 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9692 (iq2000_function_arg_padding): New function.
9693 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9694 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9695 (mips_function_arg_padding): ...this new function.
9696 (mips_pad_reg_upward): Update accordingly.
9697 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9698 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9699 targetm.calls.function_arg_padding.
9700 (FUNCTION_ARG_PADDING): Delete.
9701 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9702 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9703 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9704 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9705 (nios2_block_reg_padding): Return pad_direction instead of direction.
9706 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9707 instead of direction.
9708 (nios2_function_arg_padding): Likewise. Make static.
9709 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9710 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9711 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9712 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9713 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9714 (pa_function_arg_padding): Make static. Return pad_direction instead
9716 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9717 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9718 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9719 instead of direction. Use targetm.calls.function_arg_padding.
9720 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9721 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9722 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9723 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9724 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9726 (function_arg_padding): Rename to...
9727 (rs6000_function_arg_padding): ...this. Make static. Return
9728 pad_direction instead of direction.
9729 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9730 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9731 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9732 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9733 instead of direction. Use targetm.calls.function_arg_padding.
9734 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9735 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9736 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9737 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9738 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9739 (function_arg_padding): Rename to...
9740 (rs6000_function_arg_padding): ...this. Make static. Return
9741 pad_direction instead of direction.
9742 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9743 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9744 * config/s390/s390.c (s390_function_arg_padding): New function.
9745 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9746 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9747 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9748 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9749 (function_arg_padding): Rename to...
9750 (sparc_function_arg_padding): ...this. Make static. Return
9751 pad_direction instead of direction.
9752 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9753 * config/spu/spu.c (spu_function_arg_padding): New function.
9754 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9755 * system.h (FUNCTION_ARG_PADDING): Poison.
9757 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9758 Alan Hayward <alan.hayward@arm.com>
9759 David Sherwood <david.sherwood@arm.com>
9761 * target.def (modes_tieable_p): New hook.
9762 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9763 (TARGET_MODES_TIEABLE_P): ...this.
9764 * doc/tm.texi.in: Regenerate.
9765 * hooks.h (hook_bool_mode_mode_true): Declare.
9766 * hooks.c (hook_bool_mode_mode_true): New function.
9767 * combine.c (subst): Use targetm.modes_tieable_p instead of
9769 * dse.c (find_shift_sequence): Likewise.
9770 * expmed.c (extract_low_bits): Likewise.
9771 * lower-subreg.c: Include target.h.
9772 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9774 * rtlanal.c (rtx_cost): Likewise.
9775 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9776 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9777 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9778 (TARGET_MODES_TIEABLE_P): Redefine.
9779 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9780 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9781 (TARGET_MODES_TIEABLE_P): Redefine.
9782 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9783 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9784 (arc_modes_tieable_p): New function.
9785 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9786 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9787 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9788 (arm_modes_tieable_p): Make static.
9789 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9790 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9791 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9792 (TARGET_MODES_TIEABLE_P): Redefine.
9793 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9794 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9795 (TARGET_MODES_TIEABLE_P): Redefine.
9796 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9797 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9798 (cr16_modes_tieable_p): New function.
9799 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9800 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9801 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9802 (TRULY_NOOP_TRUNCATION): Update comment.
9803 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9804 (TRULY_NOOP_TRUNCATION): Update comment.
9805 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9806 (frv_modes_tieable_p): New function.
9807 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9808 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9809 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9810 (TARGET_MODES_TIEABLE_P): Redefine.
9811 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9812 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9813 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9814 (TARGET_MODES_TIEABLE_P): Redefine.
9815 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9816 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9817 (ia64_modes_tieable_p): New function.
9818 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9819 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9820 (iq2000_modes_tieable_p): New function.
9821 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9822 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9823 (lm32_modes_tieable_p): New function.
9824 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9825 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9826 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9827 (TARGET_MODES_TIEABLE_P): Redefine.
9828 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9829 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9830 (m32r_modes_tieable_p): New function.
9831 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9832 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9833 (m68k_modes_tieable_p): New function.
9834 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9835 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9836 (mcore_modes_tieable_p): New function.
9837 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9838 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9840 (TARGET_MODES_TIEABLE_P): Redefine.
9841 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9842 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9843 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9844 (TARGET_MODES_TIEABLE_P): Redefine.
9845 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9846 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9847 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9848 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9849 (mn10300_modes_tieable_p): ...this and make static.
9850 (TARGET_MODES_TIEABLE_P): Redefine.
9851 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9852 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9853 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9854 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9855 (msp430_modes_tieable_p): Make static.
9856 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9857 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9858 (TARGET_MODES_TIEABLE_P): Redefine.
9859 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9860 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9861 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9862 (TARGET_MODES_TIEABLE_P): Redefine.
9863 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9864 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9865 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9866 (TARGET_MODES_TIEABLE_P): Redefine.
9867 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9868 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9869 (pdp11_modes_tieable_p): New function.
9870 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9871 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9872 (rs6000_modes_tieable_p): New function.
9873 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9874 * config/powerpcspe/powerpcspe.md: Update comment.
9875 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9876 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9877 (TARGET_MODES_TIEABLE_P): Redefine.
9878 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9879 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9880 (rl78_modes_tieable_p): New function.
9881 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9882 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9883 (rs6000_modes_tieable_p): New function.
9884 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9885 * config/rs6000/rs6000.md: Update comment.
9886 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9887 * config/rx/rx.c (rx_modes_tieable_p): New function.
9888 (TARGET_MODES_TIEABLE_P): Redefine.
9889 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9890 * config/s390/s390.c (s390_modes_tieable_p): New function.
9891 (TARGET_MODES_TIEABLE_P): Redefine.
9892 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9893 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9894 (sh_modes_tieable_p): New function.
9895 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9896 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9897 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9898 (sparc_modes_tieable_p): Make static.
9899 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9900 * config/spu/spu.c (spu_modes_tieable_p): New function.
9901 (TARGET_MODES_TIEABLE_P): Redefine.
9902 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9903 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9904 (TARGET_MODES_TIEABLE_P): Redefine.
9905 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9906 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9907 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9908 * config/v850/v850.c (v850_modes_tieable_p): New function.
9909 (TARGET_MODES_TIEABLE_P): Redefine.
9910 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9911 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9912 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9913 (visium_modes_tieable_p): New function.
9914 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9915 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9916 (xtensa_modes_tieable_p): New function.
9917 * system.h (MODES_TIEABLE_P): Poison.
9919 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9920 Alan Hayward <alan.hayward@arm.com>
9921 David Sherwood <david.sherwood@arm.com>
9923 * target.def (hard_regno_mode_ok): New hook.
9924 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9925 (TARGET_HARD_REGNO_MODE_OK): ...this.
9926 * doc/tm.texi.in: Regenerate.
9927 * hooks.h (hook_bool_uint_mode_true): Declare.
9928 * hooks.c (hook_bool_uint_mode_true): New function.
9929 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9931 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9932 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9933 instead of HARD_REGNO_MODE_OK.
9934 * caller-save.c: Include target.h.
9935 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9937 * combine.c (can_combine_p): Likewise.
9938 (combinable_i3pat): Likewise.
9939 (can_change_dest_mode): Likewise.
9940 * expr.c (init_expr_target): Likewise.
9941 (convert_move): Likewise.
9942 (convert_modes): Likewise.
9943 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9944 (setup_prohibited_mode_move_regs): Likewise.
9945 * ira.h (target_ira): Likewise.
9946 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9947 * lra-constraints.c (process_alt_operands): Likewise.
9948 (split_reg): Likewise.
9949 * recog.c (peep2_find_free_register): Likewise.
9950 * ree.c (combine_reaching_defs): Likewise.
9951 * regcprop.c (maybe_mode_change): Likewise.
9952 * reginfo.c (init_reg_sets_1): Likewise.
9953 (choose_hard_reg_mode): Likewise.
9954 (simplifiable_subregs): Likewise.
9955 * regrename.c (check_new_reg_p): Likewise.
9956 * reload.c (find_valid_class): Likewise.
9957 (find_valid_class_1): Likewise.
9958 (reload_inner_reg_of_subreg): Likewise.
9959 (push_reload): Likewise.
9960 (combine_reloads): Likewise.
9961 (find_dummy_reload): Likewise.
9962 (find_reloads): Likewise.
9963 * reload1.c (find_reg): Likewise.
9964 (set_reload_reg): Likewise.
9965 (allocate_reload_reg): Likewise.
9966 (choose_reload_regs): Likewise.
9967 (reload_adjust_reg_for_temp): Likewise.
9968 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9969 (simplify_subreg_regno): Likewise.
9970 * sel-sched.c (init_regs_for_mode): Likewise.
9971 * varasm.c (make_decl_rtl): Likewise.
9972 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9973 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9975 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9976 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9977 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9978 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9979 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9981 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9982 (arc_mode_class): Delete.
9983 (HARD_REGNO_MODE_OK): Delete.
9984 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9985 (arc_hard_regno_mode_ok): Rename old array to...
9986 (arc_hard_regno_mode_ok_modes): ...this.
9987 (arc_conditional_register_usage): Update accordingly.
9988 (arc_mode_class): Make static.
9989 (arc_hard_regno_mode_ok): New function.
9990 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9991 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9992 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9993 (arm_hard_regno_mode_ok): Make static.
9994 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9996 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9997 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9998 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10001 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10002 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10003 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10004 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10005 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10006 * config/bfin/predicates.md (valid_reg_operand): Use
10007 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10008 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10009 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10010 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10011 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10012 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10013 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10014 (cr16_hard_regno_mode_ok): Make static and return a bool.
10015 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10016 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10017 (cris_hard_regno_mode_ok): New function.
10018 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10019 (epiphany_mode_class): Delete.
10020 (HARD_REGNO_MODE_OK): Delete.
10021 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10022 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10023 (hard_regno_mode_ok): Rename to...
10024 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10025 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10026 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10027 HARD_REGNO_MODE_OK.
10028 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10029 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10030 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10031 (frv_hard_regno_mode_ok): Make static and return a bool.
10032 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10033 HARD_REGNO_MODE_OK.
10034 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10035 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10036 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10037 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10040 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10041 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10042 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10044 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10045 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10046 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10047 (ia64_hard_regno_mode_ok): New function.
10048 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10049 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10050 (iq2000_hard_regno_mode_ok): New function.
10051 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10052 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10053 (lm32_hard_regno_mode_ok): New function.
10054 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10055 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10056 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10057 instead of HARD_REGNO_MODE_OK.
10058 (m32c_hard_regno_ok): Rename to...
10059 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10060 (m32c_cannot_change_mode_class): Update accordingly.
10061 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10062 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10063 (m32r_mode_class): Delete.
10064 (HARD_REGNO_MODE_OK): Delete.
10065 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10066 (m32r_hard_regno_mode_ok): Rename to...
10067 (m32r_hard_regno_modes): ...this.
10068 (m32r_mode_class): Make static.
10069 (m32r_hard_regno_mode_ok): New function.
10070 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10071 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10072 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10073 (m68k_hard_regno_mode_ok): Make static.
10074 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10075 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10076 (mcore_hard_regno_mode_ok): New function.
10077 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10078 (HARD_REGNO_MODE_OK): Delete.
10079 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10081 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10082 (microblaze_hard_regno_mode_ok): New function.
10083 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10084 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10085 (mips_hard_regno_mode_ok): Delete.
10086 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10087 (mips_hard_regno_mode_ok_p): ...this and make static.
10088 (mips_hard_regno_mode_ok_p): Rename to...
10089 (mips_hard_regno_mode_ok_uncached): ...this.
10090 (mips_hard_regno_mode_ok): New function.
10091 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10092 of HARD_REGNO_MODE_OK.
10093 (mips_option_override): Update after above name changes.
10094 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10095 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10096 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10097 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10098 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10099 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10100 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10101 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10102 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10103 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10104 (msp430_hard_regno_mode_ok): Make static and return a bool.
10105 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10106 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10107 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10109 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10110 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10111 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10112 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10113 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10114 (PA_HARD_REGNO_MODE_OK): ...this
10115 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10116 (PA_HARD_REGNO_MODE_OK): ...this.
10117 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10118 (pa_hard_regno_mode_ok): New function.
10119 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10120 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10121 (pdp11_hard_regno_mode_ok): New function.
10122 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10123 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10125 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10127 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10128 (rs6000_hard_regno_mode_ok): Rename to...
10129 (rs6000_hard_regno_mode_ok_uncached): ...this.
10130 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10131 (rs6000_hard_regno_mode_ok): New function.
10132 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10133 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10134 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10135 (riscv_hard_regno_mode_ok): ...this and make static.
10136 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10137 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10138 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10139 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10140 (rl78_hard_regno_mode_ok): Make static and return bool.
10141 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10142 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10144 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10145 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10146 (rs6000_hard_regno_mode_ok): Rename to...
10147 (rs6000_hard_regno_mode_ok_uncached): ...this.
10148 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10149 (rs6000_hard_regno_mode_ok): New function.
10150 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10151 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10152 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10153 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10154 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10155 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10156 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10157 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10158 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10159 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10160 (sh_hard_regno_mode_ok): Make static.
10161 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10162 instead of HARD_REGNO_MODE_OK.
10163 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10164 (sparc_mode_class): Delete.
10165 (HARD_REGNO_MODE_OK): Delete.
10166 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10167 (hard_regno_mode_classes): Make static.
10168 (sparc_mode_class): Likewise.
10169 (sparc_hard_regno_mode_ok): New function.
10170 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10171 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10172 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10174 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10175 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10176 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10177 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10178 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10180 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10181 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10182 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10183 (visium_hard_regno_mode_ok): New function.
10184 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10185 instead of HARD_REGNO_MODE_OK.
10186 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10187 (HARD_REGNO_MODE_OK): Delete.
10188 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10189 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10190 (xtensa_option_override): Update accordingly.
10191 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10192 (xtensa_hard_regno_mode_ok): New function.
10193 * system.h (HARD_REGNO_MODE_OK): Poison.
10195 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10196 Alan Hayward <alan.hayward@arm.com>
10197 David Sherwood <david.sherwood@arm.com>
10199 * target.def (hard_regno_call_part_clobbered): New hook.
10200 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10201 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10202 * doc/tm.texi: Regenerate.
10203 * hooks.h (hook_bool_uint_mode_false): Declare.
10204 * hooks.c (hook_bool_uint_mode_false): New function.
10205 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10206 * cselib.c (cselib_process_insn): Use
10207 targetm.hard_regno_call_part_clobbered instead of
10208 HARD_REGNO_CALL_PART_CLOBBERED.
10209 * ira-conflicts.c (ira_build_conflicts): Likewise.
10210 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10211 * lra-constraints.c (need_for_call_save_p): Likewise.
10212 * lra-lives.c: Include target.h.
10213 (check_pseudos_live_through_calls): Use
10214 targetm.hard_regno_call_part_clobbered instead of
10215 HARD_REGNO_CALL_PART_CLOBBERED.
10216 * regcprop.c: Include target.h.
10217 (copyprop_hardreg_forward_1): Use
10218 targetm.hard_regno_call_part_clobbered instead of
10219 HARD_REGNO_CALL_PART_CLOBBERED.
10220 * reginfo.c (choose_hard_reg_mode): Likewise.
10221 * regrename.c (check_new_reg_p): Likewise.
10222 * reload.c (find_equiv_reg): Likewise.
10223 * reload1.c (emit_reload_insns): Likewise.
10224 * sched-deps.c (deps_analyze_insn): Likewise.
10225 * sel-sched.c (init_regs_for_mode): Likewise.
10226 (mark_unavailable_hard_regs): Likewise.
10227 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10228 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10229 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10231 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10232 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10233 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10235 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10237 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10238 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10239 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10241 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10242 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10243 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10246 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10248 * config/powerpcspe/powerpcspe.c
10249 (rs6000_hard_regno_call_part_clobbered): New function.
10250 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10251 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10252 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10254 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10255 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10256 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10258 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10259 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10260 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10262 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10263 Alan Hayward <alan.hayward@arm.com>
10264 David Sherwood <david.sherwood@arm.com>
10266 * rtl.h (subreg_memory_offset): Declare.
10267 * emit-rtl.c (subreg_memory_offset): New function.
10268 * expmed.c (store_bit_field_1): Use it.
10269 * expr.c (undefined_operand_subword_p): Likewise.
10270 * simplify-rtx.c (simplify_subreg): Likewise.
10272 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10274 PR rtl-optimization/57448
10277 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10278 using atomic_load pattern.
10279 (expand_atomic_store): Likewise.
10281 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10284 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10285 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10286 handling. Use replace_call_with_value with NULL instead of
10287 gsi_replace, unlink_stmt_vdef and release_defs.
10289 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10292 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10294 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10297 * rtl.h (emit_library_call_value_1): Declare.
10298 (emit_library_call): Replace declaration with a series of overloads.
10299 Remove the parameter count argument.
10300 (emit_library_call_value): Likewise.
10301 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10302 with an "rtx_mode_t *".
10303 (emit_library_call_value): Delete.
10304 (emit_library_call): Likewise.
10305 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10306 (asan_emit_allocas_unpoison): Likewise.
10307 * builtins.c (expand_builtin_powi): Likewise.
10308 (expand_asan_emit_allocas_unpoison): Likewise.
10309 * cfgexpand.c (expand_main_function): Likewise.
10310 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10311 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10312 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10313 * config/arm/arm.c (arm_trampoline_init): Likewise.
10314 (arm_call_tls_get_addr): Likewise.
10315 (arm_expand_divmod_libfunc): Likewise.
10316 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10317 (smulsi3_highpart): Likewise.
10318 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10319 (c6x_expand_compare): Likewise.
10320 (c6x_expand_movmem): Likewise.
10321 * config/frv/frv.c (frv_trampoline_init): Likewise.
10322 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10323 (ix86_expand_divmod_libfunc): Likewise.
10324 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10325 (ia64_expand_compare): Likewise.
10326 (ia64_profile_hook): Likewise.
10327 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10328 (nonlocal_goto): Likewise.
10329 (restore_stack_nonlocal): Likewise.
10330 * config/m32r/m32r.c (block_move_call): Likewise.
10331 (m32r_trampoline_init): Likewise.
10332 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10333 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10334 (m68k_call_m68k_read_tp): Likewise.
10335 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10336 (microblaze_expand_divide): Likewise.
10337 * config/mips/mips.h (mips_args): Likewise.
10338 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10339 (MIPS_ICACHE_SYNC): Likewise.
10340 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10341 (nios2_trampoline_init): Likewise.
10342 * config/pa/pa.c (hppa_tls_call): Likewise.
10343 (pa_trampoline_init): Likewise.
10344 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10345 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10346 (expand_strn_compare): Likewise.
10347 (rs6000_generate_compare): Likewise.
10348 (rs6000_expand_float128_convert): Likewise.
10349 (output_profile_hook): Likewise.
10350 (rs6000_trampoline_init): Likewise.
10351 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10352 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10353 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10354 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10355 (rs6000_generate_compare): Likewise.
10356 (rs6000_expand_float128_convert): Likewise.
10357 (output_profile_hook): Likewise.
10358 (rs6000_trampoline_init): Likewise.
10359 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10360 * config/sh/sh.c (sh_trampoline_init): Likewise.
10361 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10362 (sparc_emit_float_lib_cmp): Likewise.
10363 (sparc32_initialize_trampoline): Likewise.
10364 (sparc64_initialize_trampoline): Likewise.
10365 (sparc_profile_hook): Likewise.
10366 * config/spu/spu.c (ea_load_store): Likewise.
10367 * config/spu/spu.md (floatunssidf2): Likewise.
10368 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10369 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10370 * config/visium/visium.c (expand_block_move_4): Likewise.
10371 (expand_block_move_2): Likewise.
10372 (expand_block_move_1): Likewise.
10373 (expand_block_set_4): Likewise.
10374 (expand_block_set_2): Likewise.
10375 (expand_block_set_1): Likewise.
10376 (visium_trampoline_init): Likewise.
10377 (visium_profile_hook): Likewise.
10378 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10379 (xtensa_setup_frame_addresses): Likewise.
10380 (xtensa_trampoline_init): Likewise.
10381 * except.c (sjlj_emit_function_enter): Likewise.
10382 (sjlj_emit_function_exit): Likewise.
10383 * explow.c (allocate_dynamic_stack_space): Likewise.
10384 (probe_stack_range): Likewise.
10385 * expr.c (convert_mode_scalar): Likewise.
10386 * optabs.c (expand_binop): Likewise.
10387 (expand_twoval_binop_libfunc): Likewise.
10388 (expand_unop): Likewise.
10389 (prepare_cmp_insn): Likewise.
10390 (prepare_float_lib_cmp): Likewise.
10391 (expand_float): Likewise.
10392 (expand_fix): Likewise.
10393 (expand_fixed_convert): Likewise.
10394 (maybe_emit_sync_lock_test_and_set): Likewise.
10395 (expand_atomic_compare_and_swap): Likewise.
10396 (expand_mem_thread_fence): Likewise.
10397 (expand_atomic_fetch_op): Likewise.
10399 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10401 * doc/generic.texi (OpenACC): Adjust URL.
10402 * doc/invoke.texi (C Dialect Options): Ditto.
10404 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10406 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10407 predicate for operand 1. Add (m,<S>) constraint.
10408 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10409 Prevent memory operand 1 with register operand 2.
10411 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10413 PR rtl-optimization/82024
10414 * combine.c (try_combine): If the combination result is a PARALLEL,
10415 and we only need to retain the SET in there that would be placed
10416 at I2, check that we can place that at I3 instead, before doing so.
10418 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10421 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10423 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10424 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10427 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10428 Jeff Law <law@redhat.com>
10430 * varasm.c (bss_initializer_p): Do not put constants into .bss
10431 (categorize_decl_for_section): Handle bss_initializer_p returning
10432 false when DECL_INITIAL is NULL.
10434 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10437 * config/s390/s390.c (s390_can_inline_p): New function.
10439 2017-09-01 Jeff Law <law@redhat.com>
10441 PR tree-optimization/82052
10442 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10443 Always initialize the returned slot after a hash table miss
10444 when INSERT is true.
10446 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10448 * config/s390/s390.md (mem_signal_fence): Remove.
10449 * doc/md.texi (mem_signal_fence): Remove.
10450 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10452 * target-insns.def (mem_signal_fence): Remove.
10454 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10457 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10460 * asan.c (create_odr_indicator): Strip name encoding from assembler
10461 name before appending it after __odr_asan_.
10463 2017-09-01 Martin Liska <mliska@suse.cz>
10465 PR tree-optimization/82059
10466 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10467 frequency only when an edge is redirected.
10469 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10471 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10472 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10473 (arc_conditional_register_usage): Remove ARC600 lp_count
10475 (arc_file_start): Emit Tag_ARC_CPU_variation.
10476 (arc_can_use_doloop_p): New conditions to use ZOLs.
10477 (hwloop_fail): New function.
10478 (hwloop_optimize): Likewise.
10479 (hwloop_pattern_reg): Likewise.
10480 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10481 (arc_reorg_loops): New function, calls reorg_loops.
10482 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10483 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10485 (arc_loop_hazard): Remove function, functionality moved into
10487 (arc_hazard): Remove arc_loop_hazard call.
10488 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10489 into hwloop_optimize.
10490 (arc_label_align): Remove ZOL handling.
10491 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10492 * config/arc/arc.md (doloop_begin): Remove pattern.
10493 (doloop_begin_i): Likewise.
10494 (doloop_end_i): Likewise.
10495 (doloop_fallback): Likewise.
10496 (doloop_fallback_m): Likewise.
10497 (doloop_end): Reimplement expand.
10498 (arc_lp): New pattern for LP instruction.
10499 (loop_end): New pattern.
10500 (loop_fail): Likewise.
10501 (decrement_and_branch_until_zero): Likewise.
10502 * config/arc/arc.opt (mlpc-width): New option.
10503 * doc/invoke.texi (mlpc-width): Document option.
10505 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10507 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10508 (arc_ccfsm_advance): Fix checking for delay slots.
10509 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10511 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10513 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10515 (movhi_insn): Update store instruction constraint which are saving
10516 6-bit short immediates.
10517 (movsi_insn): Consider also short scaled load operations.
10518 (zero_extendhisi2_i): Use Usd constraint instead of T.
10519 (extendhisi2_i): Add q constraint.
10520 (arc_clzsi2): Add type and length attributes.
10521 (arc_ctzsi2): Likewise.
10522 * config/arc/constraints.md (Usc): Update constraint, the
10523 assembler can parse two relocations for a single instruction.
10525 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10527 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10528 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10530 2017-08-31 Olivier Hainque <hainque@adacore.com>
10532 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10533 match as powerpc-wrs-vxworks*.
10535 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10537 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10538 register constraint for by-element operand.
10539 (aarch64_mls_elt_merge<mode>): Likewise.
10541 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10543 * config/arc/arc.c (arc_can_follow_jump): Check for short
10546 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10548 * config.gcc: Use g.opt for arc.
10549 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10550 functionality moved to ...
10551 (legitimate_scaled_address_p): New function, ...here.
10552 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10553 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10554 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10556 (arc_override_options): Handle G option.
10557 (arc_output_pic_addr_const): Correct function definition.
10558 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10559 (arc_decl_anon_ns_mem_p): Delete.
10560 (arc_in_small_data_p): Overhaul this function to take into
10561 consideration the value given via G option.
10562 (arc_rewrite_small_data_1): Renamed and corrected old
10563 arc_rewrite_small_data function.
10564 (arc_rewrite_small_data): New function.
10565 (small_data_pattern): Don't use pic_offset_table_rtx.
10566 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10567 * config/arc/simdext.md (movmisalignv2hi): Use
10568 prepare_move_operands function.
10570 (movmisalign*): Likewise.
10571 * doc/invoke.texi (ARC options): Document -G option.
10573 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10575 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10577 * config/arc/arc.c (arc_print_operand): Output scalled address for
10578 sdata whenever is possible.
10579 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10580 load/stores are available.
10581 (compact_sda_memory_operand): Check for the alignment required by
10582 code density instructions.
10583 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10585 * config/arc/constraints.md (Usd): Update constraint.
10586 (Us0): New constraint.
10587 (Usc): Update constraint.
10589 2017-08-31 Richard Biener <rguenther@suse.de>
10591 PR middle-end/82054
10592 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10593 function only once.
10595 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10597 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10598 Resize type_signature.
10600 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10601 Alan Hayward <alan.hayward@arm.com>
10602 David Sherwood <david.sherwood@arm.com>
10604 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10605 subregs whose inner modes can be stored in GPRs.
10606 (aarch64_classify_index): Likewise.
10608 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10609 Alan Hayward <alan.hayward@arm.com>
10610 David Sherwood <david.sherwood@arm.com>
10612 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10613 (V_INT_EQUIV): ...this.
10614 (v_cmp_result): Rename to...
10615 (v_int_equiv): ...this.
10616 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10617 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10618 (copysign<mode>3): Likewise.
10619 (aarch64_simd_bsl<mode>_internal): Likewise.
10620 (aarch64_simd_bsl<mode>): Likewise.
10621 (vec_cmp<mode><mode>): Likewise.
10622 (vcond<mode><mode>): Likewise.
10623 (vcond<v_cmp_mixed><mode>): Likewise.
10624 (vcondu<mode><v_cmp_mixed>): Likewise.
10625 (aarch64_cm<optab><mode>): Likewise.
10626 (aarch64_cmtst<mode>): Likewise.
10627 (aarch64_fac<optab><mode>): Likewise.
10628 (vec_perm_const<mode>): Likewise.
10629 (vcond_mask_<mode><v_cmp_result>): Rename to...
10630 (vcond_mask_<mode><v_int_equiv>): ...this.
10631 (vec_cmp<mode><v_cmp_result>): Rename to...
10632 (vec_cmp<mode><v_int_equiv>): ...this.
10634 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10635 Alan Hayward <alan.hayward@arm.com>
10636 David Sherwood <david.sherwood@arm.com>
10638 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10640 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10641 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10642 (UNSPEC_LD4_DREG): New unspecs.
10643 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10644 (aarch64_ld2<mode>_dreg_be): Replace with...
10645 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10647 (aarch64_ld3<mode>_dreg_le)
10648 (aarch64_ld3<mode>_dreg_be): Replace with...
10649 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10651 (aarch64_ld4<mode>_dreg_le)
10652 (aarch64_ld4<mode>_dreg_be): Replace with...
10653 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10656 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10658 PR tree-optimization/81987
10659 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10660 insert an initializer in a location not dominated by the stride
10663 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10665 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10666 on the entire header of the finally block in the fallthru case.
10668 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10670 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10672 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10674 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10675 rs6000_emit_move_from_cr and call renamed function.
10676 (rs6000_emit_prologue): Call renamed functions.
10677 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10678 movesi_from_cr, remove volatile CRs.
10680 2017-08-30 Jon Beniston <jon@beniston.com>
10681 Richard Biener <rguenther@suse.de>
10683 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10684 of VECTOR_MODE_P check.
10685 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10686 element vector types.
10688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10690 * df.h (df_read_modify_subreg_p): Remove in favor of...
10691 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10692 const_rtx instead of an rtx.
10693 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10694 * df-problems.c (df_word_lr_mark_ref): Likewise.
10695 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10696 (mark_pseudo_reg_dead): Likewise.
10697 (mark_ref_dead): Likewise.
10698 * reginfo.c (init_subregs_of_mode): Likewise.
10699 * sched-deps.c (sched_analyze_1): Likewise.
10700 * df-scan.c (df_def_record_1): Likewise.
10701 (df_uses_record): Likewise.
10702 (df_read_modify_subreg_p): Remove in favor of...
10703 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10704 const_rtx instead of an rtx.
10706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10707 Alan Hayward <alan.hayward@arm.com>
10708 David Sherwood <david.sherwood@arm.com>
10710 * rtl.h (partial_subreg_p): New function.
10711 * caller-save.c (save_call_clobbered_regs): Use it.
10712 * calls.c (expand_call): Likewise.
10713 * combine.c (combinable_i3pat): Likewise.
10714 (simplify_set): Likewise.
10715 (make_extraction): Likewise.
10716 (make_compound_operation_int): Likewise.
10717 (gen_lowpart_or_truncate): Likewise.
10718 (force_to_mode): Likewise.
10719 (make_field_assignment): Likewise.
10720 (reg_truncated_to_mode): Likewise.
10721 (record_truncated_value): Likewise.
10722 (move_deaths): Likewise.
10723 * cse.c (record_jump_cond): Likewise.
10724 (cse_insn): Likewise.
10725 * cselib.c (cselib_lookup_1): Likewise.
10726 * expmed.c (extract_bit_field_using_extv): Likewise.
10727 * function.c (assign_parm_setup_reg): Likewise.
10728 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10729 * ira-build.c (create_insn_allocnos): Likewise.
10730 * lra-coalesce.c (merge_pseudos): Likewise.
10731 * lra-constraints.c (match_reload): Likewise.
10732 (simplify_operand_subreg): Likewise.
10733 (curr_insn_transform): Likewise.
10734 * lra-lives.c (process_bb_lives): Likewise.
10735 * lra.c (new_insn_reg): Likewise.
10736 (lra_substitute_pseudo): Likewise.
10737 * regcprop.c (mode_change_ok): Likewise.
10738 (maybe_mode_change): Likewise.
10739 (copyprop_hardreg_forward_1): Likewise.
10740 * reload.c (push_reload): Likewise.
10741 (find_reloads): Likewise.
10742 (find_reloads_subreg_address): Likewise.
10743 * reload1.c (alter_reg): Likewise.
10744 (eliminate_regs_1): Likewise.
10745 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10747 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10749 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10750 back to if statements, including unpack.
10752 2017-08-30 Martin Liska <mliska@suse.cz>
10754 PR inline-asm/82001
10755 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10757 (func_checker::compare_asm_inputs_outputs): ... this function.
10758 (func_checker::compare_gimple_asm): Use the function to compare
10759 also ASM constrains.
10760 * ipa-icf-gimple.h: Rename the function.
10762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10763 Alan Hayward <alan.hayward@arm.com>
10764 David Sherwood <david.sherwood@arm.com>
10766 * coretypes.h (complex_mode): New type.
10767 * gdbhooks.py (build_pretty_printer): Handle it.
10768 * machmode.h (complex_mode): New class.
10769 (complex_mode::includes_p): New function.
10770 (is_complex_int_mode): Likewise.
10771 (is_complex_float_mode): Likewise.
10772 * genmodes.c (get_mode_class): Handle complex mode classes.
10773 * function.c (expand_function_end): Use is_complex_int_mode.
10775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10776 Alan Hayward <alan.hayward@arm.com>
10777 David Sherwood <david.sherwood@arm.com>
10779 * coretypes.h (scalar_mode_pod): New typedef.
10780 * gdbhooks.py (build_pretty_printer): Handle it.
10781 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10782 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10783 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10784 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10785 as_a <scalar_mode>.
10787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10788 Alan Hayward <alan.hayward@arm.com>
10789 David Sherwood <david.sherwood@arm.com>
10791 * machmode.h (mode_for_vector): Take a scalar_mode instead
10793 * stor-layout.c (mode_for_vector): Likewise.
10794 * explow.c (promote_mode): Use as_a <scalar_mode>.
10795 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10798 Alan Hayward <alan.hayward@arm.com>
10799 David Sherwood <david.sherwood@arm.com>
10801 * target.def (preferred_simd_mode): Take a scalar_mode
10802 instead of a machine_mode.
10803 * targhooks.h (default_preferred_simd_mode): Likewise.
10804 * targhooks.c (default_preferred_simd_mode): Likewise.
10805 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10806 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10807 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10808 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10809 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10810 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10811 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10812 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10814 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10815 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10816 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10817 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10818 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10819 * doc/tm.texi: Regenerate.
10820 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10824 Alan Hayward <alan.hayward@arm.com>
10825 David Sherwood <david.sherwood@arm.com>
10827 * target.def (scalar_mode_supported_p): Take a scalar_mode
10828 instead of a machine_mode.
10829 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10830 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10831 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10832 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10833 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10834 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10835 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10836 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10837 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10838 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10839 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10840 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10841 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10842 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10844 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10845 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10846 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10847 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10848 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10850 * doc/tm.texi: Regenerate.
10852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * coretypes.h (opt_scalar_mode): New typedef.
10857 * gdbhooks.py (build_pretty_printers): Handle it.
10858 * machmode.h (mode_iterator::get_2xwider): Add overload for
10860 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10862 * expr.c (convert_mode_scalar): Likewise.
10863 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10864 * optabs.c (expand_float): Likewise.
10865 (expand_fix): Likewise.
10866 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10869 Alan Hayward <alan.hayward@arm.com>
10870 David Sherwood <david.sherwood@arm.com>
10872 * optabs.c (expand_float): Explicitly check for scalars before
10873 using a branching expansion.
10874 (expand_fix): Likewise.
10876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10877 Alan Hayward <alan.hayward@arm.com>
10878 David Sherwood <david.sherwood@arm.com>
10880 * expr.c (convert_mode): Split scalar handling out into...
10881 (convert_mode_scalar): ...this new function. Treat the modes
10884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10885 Alan Hayward <alan.hayward@arm.com>
10886 David Sherwood <david.sherwood@arm.com>
10888 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10890 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10893 Alan Hayward <alan.hayward@arm.com>
10894 David Sherwood <david.sherwood@arm.com>
10896 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10897 rather than a machine_mode.
10898 (fixed_from_string): Likewise.
10899 (fixed_convert): Likewise.
10900 (fixed_convert_from_int): Likewise.
10901 (fixed_convert_from_real): Likewise.
10902 (real_convert_from_fixed): Likewise.
10903 * fixed-value.c (fixed_from_double_int): Likewise.
10904 (fixed_from_string): Likewise.
10905 (fixed_convert): Likewise.
10906 (fixed_convert_from_int): Likewise.
10907 (fixed_convert_from_real): Likewise.
10908 (real_convert_from_fixed): Likewise.
10909 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10912 Alan Hayward <alan.hayward@arm.com>
10913 David Sherwood <david.sherwood@arm.com>
10915 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10916 of separate mode class checks. Do not allow vector modes here.
10917 (immed_wide_int_const): Use as_a <scalar_mode>.
10918 * explow.c (trunc_int_for_mode): Likewise.
10919 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10920 (wi::shwi): Likewise.
10921 (wi::min_value): Likewise.
10922 (wi::max_value): Likewise.
10923 * dwarf2out.c (loc_descriptor): Likewise.
10924 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10925 for CONST_WIDE_INT.
10927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10928 Alan Hayward <alan.hayward@arm.com>
10929 David Sherwood <david.sherwood@arm.com>
10931 * tree.h (SCALAR_TYPE_MODE): New macro.
10932 * expr.c (expand_expr_addr_expr_1): Use it.
10933 (expand_expr_real_2): Likewise.
10934 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10935 (fold_convert_const_fixed_from_int): Likewise.
10936 (fold_convert_const_fixed_from_real): Likewise.
10937 (native_encode_fixed): Likewise
10938 (native_encode_complex): Likewise
10939 (native_encode_vector): Likewise.
10940 (native_interpret_fixed): Likewise.
10941 (native_interpret_real): Likewise.
10942 (native_interpret_complex): Likewise.
10943 (native_interpret_vector): Likewise.
10944 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10945 (simd_clone_adjust_argument_types): Likewise.
10946 (simd_clone_init_simd_arrays): Likewise.
10947 (simd_clone_adjust): Likewise.
10948 * stor-layout.c (layout_type): Likewise.
10949 * tree.c (build_minus_one_cst): Likewise.
10950 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10951 * tree-inline.c (estimate_move_cost): Likewise.
10952 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10953 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10954 (vectorizable_reduction): Likewise.
10955 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10956 (vect_recog_mixed_size_cond_pattern): Likewise.
10957 (check_bool_pattern): Likewise.
10958 (adjust_bool_pattern): Likewise.
10959 (search_type_for_mask_1): Likewise.
10960 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10961 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10962 (vectorizable_load): Likewise.
10963 (vectorizable_store): Likewise.
10964 * ubsan.c (ubsan_encode_value): Likewise.
10965 * varasm.c (output_constant): Likewise.
10967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10968 Alan Hayward <alan.hayward@arm.com>
10969 David Sherwood <david.sherwood@arm.com>
10971 * coretypes.h (scalar_mode): New class.
10972 * machmode.h (scalar_mode): Likewise.
10973 (scalar_mode::includes_p): New function.
10974 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10975 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10976 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10977 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10978 * expmed.c (store_bit_field_1): Likewise.
10979 (extract_bit_field_1): Likewise.
10980 * expr.c (write_complex_part): Likewise.
10981 (read_complex_part): Likewise.
10982 (emit_move_complex_push): Likewise.
10983 (expand_expr_real_2): Likewise.
10984 * function.c (assign_parm_setup_reg): Likewise.
10985 (assign_parms_unsplit_complex): Likewise.
10986 * optabs.c (expand_binop): Likewise.
10987 * rtlanal.c (subreg_get_info): Likewise.
10988 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10989 * varasm.c (output_constant_pool_2): Likewise.
10991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10992 Alan Hayward <alan.hayward@arm.com>
10993 David Sherwood <david.sherwood@arm.com>
10995 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10997 (expmed_mult_highpart_optab): Likewise.
10998 (expmed_mult_highpart): Likewise.
11000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11001 Alan Hayward <alan.hayward@arm.com>
11002 David Sherwood <david.sherwood@arm.com>
11004 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11005 instead of a machine_mode.
11006 (builtin_memset_read_str): Likewise.
11007 * builtins.c (c_readstr): Likewise.
11008 (builtin_memcpy_read_str): Likewise.
11009 (builtin_strncpy_read_str): Likewise.
11010 (builtin_memset_read_str): Likewise.
11011 (builtin_memset_gen_str): Likewise.
11012 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11013 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11014 instead of a machine_mode.
11015 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11017 (make_extraction): Likewise.
11018 (try_widen_shift_mode): Take and return scalar_int_modes instead
11020 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11021 a scalar_int_mode instead of a machine_mode.
11022 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11023 (avr_addr_space_pointer_mode): Likewise.
11024 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11025 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11026 (msp430_unwind_word_mode): Likewise.
11027 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11028 (spu_addr_space_pointer_mode): Likewise.
11029 (spu_addr_space_address_mode): Likewise.
11030 (spu_libgcc_cmp_return_mode): Likewise.
11031 (spu_libgcc_shift_count_mode): Likewise.
11032 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11033 (rl78_addr_space_pointer_mode): Likewise.
11034 (fl78_unwind_word_mode): Likewise.
11035 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11037 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11038 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11039 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11040 (mips_valid_pointer_mode): Likewise.
11041 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11042 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11043 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11045 (ft32_addr_space_address_mode): Likewise.
11046 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11047 scalar_int_mode instead of a machine_mode.
11048 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11050 (m32c_addr_space_address_mode): Likewise.
11051 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11052 (rs6000_eh_return_filter_mode): Likewise.
11053 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11054 (rs6000_eh_return_filter_mode): Likewise.
11055 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11056 (s390_libgcc_shift_count_mode): Likewise.
11057 (s390_unwind_word_mode): Likewise.
11058 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11060 * target.def (mode_rep_extended): Likewise.
11061 (valid_pointer_mode): Likewise.
11062 (addr_space.valid_pointer_mode): Likewise.
11063 (eh_return_filter_mode): Return a scalar_int_mode rather than
11065 (libgcc_cmp_return_mode): Likewise.
11066 (libgcc_shift_count_mode): Likewise.
11067 (unwind_word_mode): Likewise.
11068 (addr_space.pointer_mode): Likewise.
11069 (addr_space.address_mode): Likewise.
11070 * doc/tm.texi: Regenerate.
11071 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11073 (do_jump): Use scalar_int_mode for local variables.
11074 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11075 rather than a machine_mode.
11076 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11077 (scompare_loc_descriptor_wide): Likewise.
11078 (scompare_loc_descriptor_narrow): Likewise.
11079 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11081 * except.c (sjlj_emit_dispatch_table): Likewise.
11082 (expand_builtin_eh_copy_values): Likewise.
11083 * explow.c (convert_memory_address_addr_space_1): Likewise.
11084 Take a scalar_int_mode rather than a machine_mode.
11085 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11086 than a machine_mode.
11087 (memory_address_addr_space): Use scalar_int_mode for local variables.
11088 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11089 rather than a machine_mode.
11090 * expmed.c (mask_rtx): Likewise.
11091 (init_expmed_one_conv): Likewise.
11092 (expand_mult_highpart_adjust): Likewise.
11093 (extract_high_half): Likewise.
11094 (expmed_mult_highpart_optab): Likewise.
11095 (expmed_mult_highpart): Likewise.
11096 (expand_smod_pow2): Likewise.
11097 (expand_sdiv_pow2): Likewise.
11098 (emit_store_flag_int): Likewise.
11099 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11101 (extract_low_bits): Likewise.
11102 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11104 * expr.c (pieces_addr::adjust): Likewise.
11105 (can_store_by_pieces): Likewise.
11106 (store_by_pieces): Likewise.
11107 (clear_by_pieces_1): Likewise.
11108 (expand_expr_addr_expr_1): Likewise.
11109 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11110 (expand_expr_real_1): Likewise.
11111 (try_casesi): Likewise.
11112 * final.c (shorten_branches): Likewise.
11113 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11114 type of "mode" to machine_mode.
11115 * internal-fn.c (expand_arith_overflow_result_store): Take a
11116 scalar_int_mode rather than a machine_mode.
11117 (expand_mul_overflow): Use scalar_int_mode for local variables.
11118 * loop-doloop.c (doloop_modify): Likewise.
11119 (doloop_optimize): Likewise.
11120 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11121 than a machine_mode.
11122 (expand_doubleword_shift_condmove): Likewise.
11123 (expand_doubleword_shift): Likewise.
11124 (expand_doubleword_clz): Likewise.
11125 (expand_doubleword_popcount): Likewise.
11126 (expand_doubleword_parity): Likewise.
11127 (expand_absneg_bit): Use scalar_int_mode for local variables.
11128 (prepare_float_lib_cmp): Likewise.
11129 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11130 rather than a machine_mode.
11131 (convert_memory_address_addr_space): Likewise.
11132 (get_mode_bounds): Likewise.
11133 (get_address_mode): Return a scalar_int_mode rather than a
11135 * rtlanal.c (get_address_mode): Likewise.
11136 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11137 than a machine_mode.
11138 * targhooks.c (default_mode_rep_extended): Likewise.
11139 (default_valid_pointer_mode): Likewise.
11140 (default_addr_space_valid_pointer_mode): Likewise.
11141 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11142 than a machine_mode.
11143 (default_libgcc_cmp_return_mode): Likewise.
11144 (default_libgcc_shift_count_mode): Likewise.
11145 (default_unwind_word_mode): Likewise.
11146 (default_addr_space_pointer_mode): Likewise.
11147 (default_addr_space_address_mode): Likewise.
11148 * targhooks.h (default_eh_return_filter_mode): Likewise.
11149 (default_libgcc_cmp_return_mode): Likewise.
11150 (default_libgcc_shift_count_mode): Likewise.
11151 (default_unwind_word_mode): Likewise.
11152 (default_addr_space_pointer_mode): Likewise.
11153 (default_addr_space_address_mode): Likewise.
11154 (default_mode_rep_extended): Take a scalar_int_mode rather than
11156 (default_valid_pointer_mode): Likewise.
11157 (default_addr_space_valid_pointer_mode): Likewise.
11158 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11160 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11161 rather than a machine_mode.
11162 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11163 for local variables.
11164 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11165 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11166 than a machine_mode.
11168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11169 Alan Hayward <alan.hayward@arm.com>
11170 David Sherwood <david.sherwood@arm.com>
11172 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11173 the mode argument to scalar_int_mode.
11174 (do_jump_by_parts_zero_rtx): Likewise.
11175 (do_jump_by_parts_equality_rtx): Likewise.
11176 (do_jump_by_parts_greater): Take a mode argument.
11177 (do_jump_by_parts_equality): Likewise.
11178 (do_jump_1): Update calls accordingly.
11180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11181 Alan Hayward <alan.hayward@arm.com>
11182 David Sherwood <david.sherwood@arm.com>
11184 * is-a.h (safe_dyn_cast): New function.
11185 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11186 (jump_table_for_label): Likewise.
11187 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11188 instead of an rtx_insn *.
11189 (shorten_branches): Use dyn_cast instead of LABEL_P and
11190 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11191 rtx_jump_table_data::get_data_mode.
11192 (final_scan_insn): Likewise.
11194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11195 Alan Hayward <alan.hayward@arm.com>
11196 David Sherwood <david.sherwood@arm.com>
11198 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11199 trying to combine a full-register integer set with a subreg
11202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11203 Alan Hayward <alan.hayward@arm.com>
11204 David Sherwood <david.sherwood@arm.com>
11206 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11207 that is always either address_mode or pointer_mode.
11209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11210 Alan Hayward <alan.hayward@arm.com>
11211 David Sherwood <david.sherwood@arm.com>
11213 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11214 when the two are known to be equal.
11216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11217 Alan Hayward <alan.hayward@arm.com>
11218 David Sherwood <david.sherwood@arm.com>
11220 * simplify-rtx.c (simplify_const_unary_operation): Use
11221 is_a <scalar_int_mode> instead of checking for a nonzero
11222 precision. Forcibly convert op_mode to a scalar_int_mode
11223 in that case. More clearly differentiate the operand and
11224 result modes and use the former when deciding what the value
11225 of a count-bits operation should be. Use is_int_mode instead
11226 of checking for a MODE_INT. Remove redundant check for whether
11227 this mode has a zero precision.
11229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11230 Alan Hayward <alan.hayward@arm.com>
11231 David Sherwood <david.sherwood@arm.com>
11233 * optabs.c (widen_leading): Change the type of the mode argument
11234 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11235 (widen_bswap): Likewise.
11236 (expand_parity): Likewise.
11237 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11238 (expand_ffs): Likewise.
11239 (epand_unop): Check for scalar integer modes before calling the
11242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11243 Alan Hayward <alan.hayward@arm.com>
11244 David Sherwood <david.sherwood@arm.com>
11246 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11248 (expand_expr_real_1): Update call accordingly.
11250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11251 Alan Hayward <alan.hayward@arm.com>
11252 David Sherwood <david.sherwood@arm.com>
11254 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11255 value_mode arguments. Use scalar_int_mode internally.
11256 (store_bit_field_1): Rename the new integer mode from imode
11257 to op0_mode and use it instead of GET_MODE (op0). Update calls
11258 to store_split_bit_field, store_bit_field_using_insv and
11259 store_fixed_bit_field.
11260 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11261 Use scalar_int_mode internally. Use a bit count rather than a mode
11262 when calculating the largest bit size for get_best_mode.
11263 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11264 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11265 Remove assertion that OP0 has a scalar integer mode.
11266 (store_split_bit_field): Add op0_mode and value_mode arguments.
11267 Update calls to extract_fixed_bit_field.
11268 (extract_bit_field_using_extv): Add an op0_mode argument.
11269 Use scalar_int_mode internally.
11270 (extract_bit_field_1): Rename the new integer mode from imode to
11271 op0_mode and use it instead of GET_MODE (op0). Update calls to
11272 extract_split_bit_field, extract_bit_field_using_extv and
11273 extract_fixed_bit_field.
11274 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11275 to extract_split_bit_field and extract_fixed_bit_field_1.
11276 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11277 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11278 on the target mode.
11279 (extract_split_bit_field): Add an op0_mode argument. Update call
11280 to extract_fixed_bit_field.
11282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11283 Alan Hayward <alan.hayward@arm.com>
11284 David Sherwood <david.sherwood@arm.com>
11286 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11287 * explow.c (hard_function_value): Likewise.
11288 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11289 convert_to_mode call outside the loop.
11290 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11291 for the mode iterator. Require the mode specified by max_pieces
11293 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11295 (copy_blkmode_to_reg): Likewise.
11296 (set_storage_via_setmem): Likewise.
11297 * optabs.c (prepare_cmp_insn): Likewise.
11298 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11299 * stor-layout.c (finish_bitfield_representative): Likewise.
11301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11302 Alan Hayward <alan.hayward@arm.com>
11303 David Sherwood <david.sherwood@arm.com>
11305 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11306 * expr.c (convert_move): Use them.
11307 (convert_modes): Likewise.
11308 (store_expr_with_bounds): Likewise.
11310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11311 Alan Hayward <alan.hayward@arm.com>
11312 David Sherwood <david.sherwood@arm.com>
11314 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11315 parameter for the mode of "x". Remove the "known_x", "known_mode"
11316 and "known_ret" arguments. Change the type of the mode argument
11317 to scalar_int_mode.
11318 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11319 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11320 (reg_num_sign_bit_copies_for_combine): Likewise.
11321 * rtlanal.c (nonzero_bits1): Likewise.
11322 (num_sign_bit_copies1): Likewise.
11323 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11324 (reg_num_sign_bit_copies_general): Likewise.
11325 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11326 (reg_nonzero_bits_general): Likewise.
11328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11329 Alan Hayward <alan.hayward@arm.com>
11330 David Sherwood <david.sherwood@arm.com>
11332 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11333 than in subroutines. Return 1 for non-integer modes.
11334 (cached_num_sign_bit_copies): Change the type of the mode parameter
11335 to scalar_int_mode.
11336 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11337 classes. Handle CONST_INT_P first and then check whether X also
11338 has a scalar integer mode. Check the same thing for inner registers
11339 of a SUBREG and for values that are being extended or truncated.
11341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11346 in subroutines. Return the mode mask for non-integer modes.
11347 (cached_nonzero_bits): Change the type of the mode parameter
11348 to scalar_int_mode.
11349 (nonzero_bits1): Likewise. Remove early exit for other mode
11350 classes. Handle CONST_INT_P first and then check whether X
11351 also has a scalar integer mode.
11353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11354 Alan Hayward <alan.hayward@arm.com>
11355 David Sherwood <david.sherwood@arm.com>
11357 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11358 Return a scalar_int_mode and assert that the size is greater than
11360 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11361 (op_by_pieces_d::op_by_pieces_d): Likewise.
11362 (op_by_pieces_d::run): Likewise.
11363 (can_store_by_pieces): Likewise.
11365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * combine.c (extract_left_shift): Add a mode argument and update
11371 (make_compound_operation_int): Change the type of the mode parameter
11372 to scalar_int_mode and update the call to extract_left_shift.
11374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11375 Alan Hayward <alan.hayward@arm.com>
11376 David Sherwood <david.sherwood@arm.com>
11378 * combine.c (simplify_and_const_int): Change the type of the mode
11379 parameter to scalar_int_mode.
11380 (simplify_and_const_int_1): Likewise. Update recursive call.
11382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11383 Alan Hayward <alan.hayward@arm.com>
11384 David Sherwood <david.sherwood@arm.com>
11386 * combine.c (simplify_compare_const): Check that the mode is a
11387 scalar_int_mode (rather than VOIDmode) before testing its
11389 (simplify_comparison): Move COMPARISON_P handling out of the
11390 loop and restrict the latter part of the loop to scalar_int_modes.
11391 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11392 and when considering SUBREG_REGs. Use is_int_mode instead of
11393 checking GET_MODE_CLASS against MODE_INT.
11395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11396 Alan Hayward <alan.hayward@arm.com>
11397 David Sherwood <david.sherwood@arm.com>
11399 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11400 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11401 shift_unit_mode and for modes involved in scalar shifts.
11403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11404 Alan Hayward <alan.hayward@arm.com>
11405 David Sherwood <david.sherwood@arm.com>
11407 * combine.c (force_int_to_mode): New function, split out from...
11408 (force_to_mode): ...here. Keep xmode up-to-date and use it
11409 instead of GET_MODE (x).
11411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11412 Alan Hayward <alan.hayward@arm.com>
11413 David Sherwood <david.sherwood@arm.com>
11415 * optabs-query.h (extraction_insn::struct_mode): Change type to
11416 opt_scalar_int_mode and update comment.
11417 (extraction_insn::field_mode): Change type to scalar_int_mode.
11418 (extraction_insn::pos_mode): Likewise.
11419 * combine.c (make_extraction): Update accordingly.
11420 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11421 (get_optab_extraction_insn): Likewise.
11422 * recog.c (simplify_while_replacing): Likewise.
11423 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11424 parameter to opt_scalar_int_mode.
11426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11427 Alan Hayward <alan.hayward@arm.com>
11428 David Sherwood <david.sherwood@arm.com>
11430 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11431 to a scalar_int_mode instead of a machine_mode.
11432 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11433 (get_best_mode): Return a boolean and use a pointer argument to store
11434 the selected mode. Replace the limit mode parameter with a bit limit.
11435 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11436 for the values returned by bit_field_mode_iterator::next_mode.
11437 (store_bit_field): Update call to get_best_mode.
11438 (store_fixed_bit_field): Likewise.
11439 (extract_fixed_bit_field): Likewise.
11440 * expr.c (optimize_bitfield_assignment_op): Likewise.
11441 * fold-const.c (optimize_bit_field_compare): Likewise.
11442 (fold_truth_andor_1): Likewise.
11443 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11444 Update for new type of m_mode.
11445 (get_best_mode): As above.
11447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11448 Alan Hayward <alan.hayward@arm.com>
11449 David Sherwood <david.sherwood@arm.com>
11451 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11452 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11453 (store_bit_field): Check is_a <scalar_int_mode> before calling
11454 strict_volatile_bitfield_p.
11455 (extract_bit_field): Likewise.
11457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11458 Alan Hayward <alan.hayward@arm.com>
11459 David Sherwood <david.sherwood@arm.com>
11461 * target.def (cstore_mode): Return a scalar_int_mode.
11462 * doc/tm.texi: Regenerate.
11463 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11464 * targhooks.h (default_cstore_mode): Likewise.
11465 * targhooks.c (default_cstore_mode): Likewise, using a forced
11467 * expmed.c (emit_cstore): Expect the target of the cstore to be
11470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11471 Alan Hayward <alan.hayward@arm.com>
11472 David Sherwood <david.sherwood@arm.com>
11474 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11476 (niter_desc): Likewise mode.
11477 (iv_analyze): Add a mode parameter.
11479 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11480 and change its type to scalar_int_mode.
11481 * loop-iv.c: Update commentary at head of file.
11482 (iv_constant): Pass the mode paraeter before the rtx it describes
11483 and change its type to scalar_int_mode. Remove VOIDmode handling.
11484 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11485 (iv_extend): Likewise.
11486 (shorten_into_mode): Likewise.
11487 (iv_add): Use scalar_int_mode.
11488 (iv_mult): Likewise.
11489 (iv_shift): Likewise.
11490 (canonicalize_iv_subregs): Likewise.
11491 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11492 it describes and change its mode to scalar_int_mode. Also change
11493 the type of the returned inner_mode to scalar_int_mode.
11494 (get_biv_step): Likewise, turning outer_mode from a pointer
11495 into a direct parameter. Update call to get_biv_step_1.
11496 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11497 iv_constant and get_biv_step.
11498 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11499 and change its type to scalar_int_mode. Don't initialise iv->mode
11500 to VOIDmode and remove later checks for its still being VOIDmode.
11501 Update calls to iv_analyze_op and iv_analyze_expr. Check
11502 is_a <scalar_int_mode> when changing the mode under consideration.
11503 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11504 Update call to iv_analyze_expr.
11505 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11506 inner register is not also a scalar_int_mode. Update call to
11508 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11509 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11510 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11511 separate mode class checks. Update calls to iv_analyze. Remove
11512 fix-up of VOIDmodes after iv_analyze_biv.
11513 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11514 don't have a scalar_int_mode. Update call to biv_p.
11516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11517 Alan Hayward <alan.hayward@arm.com>
11518 David Sherwood <david.sherwood@arm.com>
11520 * cfgexpand.c (convert_debug_memory_address): Use
11521 as_a <scalar_int_mode>.
11522 * combine.c (expand_compound_operation): Likewise.
11523 (make_extraction): Likewise.
11524 (change_zero_ext): Likewise.
11525 (simplify_comparison): Likewise.
11526 * cse.c (cse_insn): Likewise.
11527 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11528 (mem_loc_descriptor): Likewise.
11529 (loc_descriptor): Likewise.
11530 * expmed.c (init_expmed_one_mode): Likewise.
11531 (synth_mult): Likewise.
11532 (emit_store_flag_1): Likewise.
11533 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11534 of a comparison with size.
11535 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11536 (reduce_to_bit_field_precision): Likewise.
11537 * function.c (expand_function_end): Likewise.
11538 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11539 * loop-doloop.c (doloop_modify): Likewise.
11540 * optabs.c (expand_binop): Likewise.
11541 (expand_unop): Likewise.
11542 (expand_copysign_absneg): Likewise.
11543 (prepare_cmp_insn): Likewise.
11544 (maybe_legitimize_operand): Likewise.
11545 * recog.c (const_scalar_int_operand): Likewise.
11546 * rtlanal.c (get_address_mode): Likewise.
11547 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11548 (simplify_cond_clz_ctz): Likewise.
11549 * tree-nested.c (get_nl_goto_field): Likewise.
11550 * tree.c (build_vector_type_for_mode): Likewise.
11551 * var-tracking.c (use_narrower_mode): Likewise.
11553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11554 Alan Hayward <alan.hayward@arm.com>
11555 David Sherwood <david.sherwood@arm.com>
11557 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11558 * builtins.c (expand_builtin_signbit): Use it.
11559 * cfgexpand.c (expand_debug_expr): Likewise.
11560 * dojump.c (do_jump): Likewise.
11561 (do_compare_and_jump): Likewise.
11562 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11563 * expmed.c (make_tree): Likewise.
11564 * expr.c (expand_expr_real_2): Likewise.
11565 (expand_expr_real_1): Likewise.
11566 (try_casesi): Likewise.
11567 * fold-const-call.c (fold_const_call_ss): Likewise.
11568 * fold-const.c (unextend): Likewise.
11569 (extract_muldiv_1): Likewise.
11570 (fold_single_bit_test): Likewise.
11571 (native_encode_int): Likewise.
11572 (native_encode_string): Likewise.
11573 (native_interpret_int): Likewise.
11574 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11575 * internal-fn.c (expand_addsub_overflow): Likewise.
11576 (expand_neg_overflow): Likewise.
11577 (expand_mul_overflow): Likewise.
11578 (expand_arith_overflow): Likewise.
11579 * match.pd: Likewise.
11580 * stor-layout.c (layout_type): Likewise.
11581 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11582 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11583 * tree-ssanames.c (get_range_info): Likewise.
11584 * tree-switch-conversion.c (array_value_type) Likewise.
11585 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11586 (vect_recog_divmod_pattern): Likewise.
11587 (vect_recog_mixed_size_cond_pattern): Likewise.
11588 * tree-vrp.c (extract_range_basic): Likewise.
11589 (simplify_float_conversion_using_ranges): Likewise.
11590 * tree.c (int_fits_type_p): Likewise.
11591 * ubsan.c (instrument_bool_enum_load): Likewise.
11592 * varasm.c (mergeable_string_section): Likewise.
11593 (narrowing_initializer_constant_valid_p): Likewise.
11594 (output_constant): Likewise.
11596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11597 Alan Hayward <alan.hayward@arm.com>
11598 David Sherwood <david.sherwood@arm.com>
11600 * machmode.h (NARROWEST_INT_MODE): New macro.
11601 * expr.c (alignment_for_piecewise_move): Use it instead of
11602 GET_CLASS_NARROWEST_MODE (MODE_INT).
11603 (push_block): Likewise.
11604 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11606 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11609 Alan Hayward <alan.hayward@arm.com>
11610 David Sherwood <david.sherwood@arm.com>
11612 * postreload.c (move2add_valid_value_p): Change the type of the
11613 mode parameter to scalar_int_mode.
11614 (move2add_use_add2_insn): Add a mode parameter and use it instead
11616 (move2add_use_add3_insn): Likewise.
11617 (reload_cse_move2add): Update accordingly.
11619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11620 Alan Hayward <alan.hayward@arm.com>
11621 David Sherwood <david.sherwood@arm.com>
11623 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11625 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11626 * optabs.c (expand_unop): Likewise.
11628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11629 Alan Hayward <alan.hayward@arm.com>
11630 David Sherwood <david.sherwood@arm.com>
11632 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11633 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11634 (popcount_loc_descriptor): Likewise.
11635 (bswap_loc_descriptor): Likewise.
11636 (rotate_loc_descriptor): Likewise.
11637 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11638 calling the functions above.
11640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11641 Alan Hayward <alan.hayward@arm.com>
11642 David Sherwood <david.sherwood@arm.com>
11644 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11646 (try_combine): Likewise.
11647 (simplify_if_then_else): Likewise.
11648 * cse.c (cse_insn): Likewise.
11649 * dwarf2out.c (mem_loc_descriptor): Likewise.
11650 * emit-rtl.c (gen_lowpart_common): Likewise.
11651 * simplify-rtx.c (simplify_truncation): Likewise.
11652 (simplify_binary_operation_1): Likewise.
11653 (simplify_const_relational_operation): Likewise.
11654 (simplify_ternary_operation): Likewise.
11655 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11658 Alan Hayward <alan.hayward@arm.com>
11659 David Sherwood <david.sherwood@arm.com>
11661 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11662 * reload.c (push_reload): Likewise.
11663 (find_reloads): Likewise.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11670 (make_compound_operation_int): Likewise.
11671 (change_zero_ext): Likewise.
11672 * expr.c (convert_move): Likewise.
11673 (convert_modes): Likewise.
11674 * fwprop.c (forward_propagate_subreg): Likewise.
11675 * loop-iv.c (get_biv_step_1): Likewise.
11676 * optabs.c (widen_operand): Likewise.
11677 * postreload.c (move2add_valid_value_p): Likewise.
11678 * recog.c (simplify_while_replacing): Likewise.
11679 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11680 (simplify_binary_operation_1): Likewise. Remove redundant
11681 mode equality check.
11683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11684 Alan Hayward <alan.hayward@arm.com>
11685 David Sherwood <david.sherwood@arm.com>
11687 * combine.c (combine_simplify_rtx): Add checks for
11688 is_a <scalar_int_mode>.
11689 (simplify_if_then_else): Likewise.
11690 (make_field_assignment): Likewise.
11691 (simplify_comparison): Likewise.
11692 * ifcvt.c (noce_try_bitop): Likewise.
11693 * loop-invariant.c (canonicalize_address_mult): Likewise.
11694 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11697 Alan Hayward <alan.hayward@arm.com>
11698 David Sherwood <david.sherwood@arm.com>
11700 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11701 is_a <scalar_int_mode> instead of != BLKmode.
11703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11704 Alan Hayward <alan.hayward@arm.com>
11705 David Sherwood <david.sherwood@arm.com>
11707 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11708 instead of != VOIDmode.
11709 * combine.c (if_then_else_cond): Likewise.
11710 (change_zero_ext): Likewise.
11711 * dwarf2out.c (mem_loc_descriptor): Likewise.
11712 (loc_descriptor): Likewise.
11713 * rtlanal.c (canonicalize_condition): Likewise.
11714 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11717 Alan Hayward <alan.hayward@arm.com>
11718 David Sherwood <david.sherwood@arm.com>
11720 * simplify-rtx.c (simplify_binary_operation_1): Use
11721 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11724 Alan Hayward <alan.hayward@arm.com>
11725 David Sherwood <david.sherwood@arm.com>
11727 * wide-int.h (int_traits<unsigned char>) New class.
11728 (int_traits<unsigned short>) Likewise.
11729 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11730 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11732 * combine.c (set_nonzero_bits_and_sign_copies): Use
11733 is_a <scalar_int_mode>.
11734 (find_split_point): Likewise.
11735 (combine_simplify_rtx): Likewise.
11736 (simplify_logical): Likewise.
11737 (expand_compound_operation): Likewise.
11738 (expand_field_assignment): Likewise.
11739 (make_compound_operation): Likewise.
11740 (extended_count): Likewise.
11741 (change_zero_ext): Likewise.
11742 (simplify_comparison): Likewise.
11743 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11744 (ucompare_loc_descriptor): Likewise.
11745 (minmax_loc_descriptor): Likewise.
11746 (mem_loc_descriptor): Likewise.
11747 (loc_descriptor): Likewise.
11748 * expmed.c (init_expmed_one_mode): Likewise.
11749 * lra-constraints.c (lra_constraint_offset): Likewise.
11750 * optabs.c (prepare_libcall_arg): Likewise.
11751 * postreload.c (move2add_note_store): Likewise.
11752 * reload.c (operands_match_p): Likewise.
11753 * rtl.h (load_extend_op): Likewise.
11754 * rtlhooks.c (gen_lowpart_general): Likewise.
11755 * simplify-rtx.c (simplify_truncation): Likewise.
11756 (simplify_unary_operation_1): Likewise.
11757 (simplify_binary_operation_1): Likewise.
11758 (simplify_const_binary_operation): Likewise.
11759 (simplify_const_relational_operation): Likewise.
11760 (simplify_subreg): Likewise.
11761 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11762 * var-tracking.c (adjust_mems): Likewise.
11763 (prepare_call_arguments): Likewise.
11765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11766 Alan Hayward <alan.hayward@arm.com>
11767 David Sherwood <david.sherwood@arm.com>
11769 * machmode.h (is_int_mode): New fuction.
11770 * combine.c (find_split_point): Use it.
11771 (combine_simplify_rtx): Likewise.
11772 (simplify_if_then_else): Likewise.
11773 (simplify_set): Likewise.
11774 (simplify_shift_const_1): Likewise.
11775 (simplify_comparison): Likewise.
11776 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11777 * cse.c (notreg_cost): Likewise.
11778 (cse_insn): Likewise.
11779 * cselib.c (cselib_lookup_1): Likewise.
11780 * dojump.c (do_jump_1): Likewise.
11781 (do_compare_rtx_and_jump): Likewise.
11782 * dse.c (get_call_args): Likewise.
11783 * dwarf2out.c (rtl_for_decl_init): Likewise.
11784 (native_encode_initializer): Likewise.
11785 * expmed.c (emit_store_flag_1): Likewise.
11786 (emit_store_flag): Likewise.
11787 * expr.c (convert_modes): Likewise.
11788 (store_field): Likewise.
11789 (expand_expr_real_1): Likewise.
11790 * fold-const.c (fold_read_from_constant_string): Likewise.
11791 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11792 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11793 * optabs.c (expand_binop): Likewise.
11794 (expand_unop): Likewise.
11795 (expand_abs_nojump): Likewise.
11796 (expand_one_cmpl_abs_nojump): Likewise.
11797 * simplify-rtx.c (mode_signbit_p): Likewise.
11798 (val_signbit_p): Likewise.
11799 (val_signbit_known_set_p): Likewise.
11800 (val_signbit_known_clear_p): Likewise.
11801 (simplify_relational_operation_1): Likewise.
11802 * tree.c (vector_type_mode): Likewise.
11804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11805 Alan Hayward <alan.hayward@arm.com>
11806 David Sherwood <david.sherwood@arm.com>
11808 * machmode.h (smallest_mode_for_size): Fix formatting.
11809 (smallest_int_mode_for_size): New function.
11810 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11811 instead of smallest_mode_for_size.
11812 * combine.c (make_extraction): Likewise.
11813 * config/arc/arc.c (arc_expand_movmem): Likewise.
11814 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11815 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11816 * config/s390/s390.c (s390_expand_insv): Likewise.
11817 * config/sparc/sparc.c (assign_int_registers): Likewise.
11818 * config/spu/spu.c (spu_function_value): Likewise.
11819 (spu_function_arg): Likewise.
11820 * coverage.c (get_gcov_type): Likewise.
11821 (get_gcov_unsigned_t): Likewise.
11822 * dse.c (find_shift_sequence): Likewise.
11823 * expmed.c (store_bit_field_1): Likewise.
11824 * expr.c (convert_move): Likewise.
11825 (store_field): Likewise.
11826 * internal-fn.c (expand_arith_overflow): Likewise.
11827 * optabs-query.c (get_best_extraction_insn): Likewise.
11828 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11829 * stor-layout.c (layout_type): Likewise.
11830 (initialize_sizetypes): Likewise.
11831 * targhooks.c (default_get_mask_mode): Likewise.
11832 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11835 Alan Hayward <alan.hayward@arm.com>
11836 David Sherwood <david.sherwood@arm.com>
11838 * machmode.h (opt_mode::else_blk): New function.
11839 (int_mode_for_mode): Declare.
11840 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11841 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11843 * cfgexpand.c (expand_debug_expr): Likewise.
11844 * combine.c (gen_lowpart_or_truncate): Likewise.
11845 (gen_lowpart_for_combine): Likewise.
11846 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11847 * config/avr/avr.c (avr_to_int_mode): Likewise.
11848 (avr_out_plus_1): Likewise.
11849 (avr_out_plus): Likewise.
11850 (avr_out_round): Likewise.
11851 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11852 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11853 (s390_expand_vcond): Likewise.
11854 * config/spu/spu.c (spu_split_immediate): Likewise.
11855 (spu_expand_mov): Likewise.
11856 * dse.c (get_stored_val): Likewise.
11857 * expmed.c (store_bit_field_1): Likewise.
11858 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11860 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11861 (extract_low_bits): Likewise.
11862 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11863 handling rather than repeating the check.
11864 (emit_group_store): Likewise.
11865 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11866 * optabs.c (expand_absneg_bit): Likewise.
11867 (expand_copysign_absneg): Likewise.
11868 (expand_copysign_bit): Likewise.
11869 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11870 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11871 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11872 * var-tracking.c (prepare_call_arguments): Likewise.
11873 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11874 int_mode_for_mode instead of mode_for_size.
11875 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11878 Alan Hayward <alan.hayward@arm.com>
11879 David Sherwood <david.sherwood@arm.com>
11881 * machmode.h (int_mode_for_size): New function.
11882 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11883 instead of mode_for_size.
11884 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11886 * combine.c (expand_field_assignment): Use int_mode_for_size
11887 instead of mode_for_size.
11888 (make_extraction): Likewise.
11889 (simplify_shift_const_1): Likewise.
11890 (simplify_comparison): Likewise.
11891 * dojump.c (do_jump): Likewise.
11892 * dwarf2out.c (mem_loc_descriptor): Likewise.
11893 * emit-rtl.c (init_derived_machine_modes): Likewise.
11894 * expmed.c (flip_storage_order): Likewise.
11895 (convert_extracted_bit_field): Likewise.
11896 * expr.c (copy_blkmode_from_reg): Likewise.
11897 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11898 * internal-fn.c (expand_mul_overflow): Likewise.
11899 * lower-subreg.c (simple_move): Likewise.
11900 * optabs-libfuncs.c (init_optabs): Likewise.
11901 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11902 * tree.c (vector_type_mode): Likewise.
11903 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11904 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11905 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11906 * tree-vect-stmts.c (vectorizable_load): Likewise.
11907 (vectorizable_store): Likewise.
11909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11910 Alan Hayward <alan.hayward@arm.com>
11911 David Sherwood <david.sherwood@arm.com>
11913 * coretypes.h (pod_mode): New type.
11914 (scalar_int_mode_pod): New typedef.
11915 * machmode.h (pod_mode): New class.
11916 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11917 * genmodes.c (emit_mode_int_n): Update accordingly.
11918 * lower-subreg.h (target_lower_subreg): Change type to
11919 scalar_int_mode_pod.
11920 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11921 scalar_int_mode_pod.
11923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11924 Alan Hayward <alan.hayward@arm.com>
11925 David Sherwood <david.sherwood@arm.com>
11927 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11928 machine_mode to scalar_int_mode.
11929 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11930 (rs6000_option_override_internal): Remove cast to int.
11931 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11932 machine_mode to scalar_int_mode.
11933 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11934 (rs6000_option_override_internal): Remove cast to int.
11935 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11936 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11938 * config/s390/s390.c (s390_expand_builtin): Likewise.
11939 * coretypes.h (scalar_int_mode): New type.
11940 (opt_scalar_int_mode): New typedef.
11941 * machmode.h (scalar_int_mode): New class.
11942 (scalar_int_mode::includes_p): New function.
11943 (byte_mode): Change type to scalar_int_mode.
11944 (word_mode): Likewise.
11945 (ptr_mode): Likewise.
11946 * emit-rtl.c (byte_mode): Likewise.
11947 (word_mode): Likewise.
11948 (ptr_mode): Likewise.
11949 (init_derived_machine_modes): Update accordingly.
11950 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11951 and MODE_PARTIAL_INT.
11952 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11953 opt_scalar_int_mode.
11955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11956 Alan Hayward <alan.hayward@arm.com>
11957 David Sherwood <david.sherwood@arm.com>
11959 * target.def (libgcc_floating_mode_supported_p): Take a
11961 * doc/tm.texi: Regenerate.
11962 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11964 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11965 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11969 Alan Hayward <alan.hayward@arm.com>
11970 David Sherwood <david.sherwood@arm.com>
11972 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11973 * doc/tm.texi: Regenerate.
11974 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11975 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11976 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11977 * targhooks.h (default_floatn_mode): Likewise.
11978 * targhooks.c (default_floatn_mode): Likewise.
11979 * tree.c (build_common_tree_nodes): Update accordingly.
11981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11982 Alan Hayward <alan.hayward@arm.com>
11983 David Sherwood <david.sherwood@arm.com>
11985 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11986 (mode_iterator::iterate_p): Likewise.
11987 (mode_iterator::get_wider): Likewise.
11988 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11991 Alan Hayward <alan.hayward@arm.com>
11992 David Sherwood <david.sherwood@arm.com>
11994 * coretypes.h (opt_scalar_float_mode): New typedef.
11995 * machmode.h (float_mode_for_size): New function.
11996 * emit-rtl.c (double_mode): Delete.
11997 (init_emit_once): Use float_mode_for_size.
11998 * stor-layout.c (layout_type): Likewise.
11999 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12002 Alan Hayward <alan.hayward@arm.com>
12003 David Sherwood <david.sherwood@arm.com>
12005 * output.h (assemble_real): Take a scalar_float_mode.
12006 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12007 * config/arm/arm.md (consttable_4): Likewise.
12008 (consttable_8): Likewise.
12009 (consttable_16): Likewise.
12010 * config/mips/mips.md (consttable_float): Likewise.
12011 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12012 * varasm.c (assemble_real): Take a scalar_float_mode.
12013 (output_constant_pool_2): Update accordingly.
12014 (output_constant): Likewise.
12016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12017 Alan Hayward <alan.hayward@arm.com>
12018 David Sherwood <david.sherwood@arm.com>
12020 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12021 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12022 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12023 (native_encode_real): Likewise.
12024 (native_interpret_real): Likewise.
12025 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12026 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12029 Alan Hayward <alan.hayward@arm.com>
12030 David Sherwood <david.sherwood@arm.com>
12032 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12033 <scalar_float_mode>. Simplify.
12034 (gen_extend_conv_libfunc): Likewise.
12036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12037 Alan Hayward <alan.hayward@arm.com>
12038 David Sherwood <david.sherwood@arm.com>
12040 * coretypes.h (scalar_float_mode): New type.
12041 * machmode.h (mode_traits::from_int): Use machine_mode if
12042 USE_ENUM_MODES is defined.
12043 (is_a): New function.
12045 (dyn_cast): Likewise.
12046 (scalar_float_mode): New class.
12047 (scalar_float_mode::includes_p): New function.
12048 (is_float_mode): Likewise.
12049 * gdbhooks.py (MachineModePrinter): New class.
12050 (build_pretty_printer): Use it for scalar_float_mode.
12051 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12052 (format_helper::format_helper): Turn into a template.
12053 * genmodes.c (get_mode_class): New function.
12054 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12055 or machine_mode if none.
12056 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12057 as_a <scalar_float_mode>.
12058 * dwarf2out.c (mem_loc_descriptor): Likewise.
12059 (insert_float): Likewise.
12060 (add_const_value_attribute): Likewise.
12061 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12062 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12063 (expand_unop): Update accordingly.
12064 (expand_abs_nojump): Likewise.
12065 (expand_copysign_absneg): Take a scalar_float_mode.
12066 (expand_copysign_bit): Likewise.
12067 (expand_copysign): Update accordingly.
12069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12070 Alan Hayward <alan.hayward@arm.com>
12071 David Sherwood <david.sherwood@arm.com>
12073 * coretypes.h (opt_mode): New class.
12074 * machmode.h (opt_mode): Likewise.
12075 (opt_mode::else_void): New function.
12076 (opt_mode::require): Likewise.
12077 (opt_mode::exists): Likewise.
12078 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12079 (GET_MODE_2XWIDER_MODE): Likewise.
12080 (mode_iterator::get_wider): Update accordingly.
12081 (mode_iterator::get_2xwider): Likewise.
12082 (mode_iterator::get_known_wider): Likewise, turning into a template.
12083 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12084 forcing a wider mode to exist.
12085 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12086 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12087 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12088 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12089 * lower-subreg.c (init_lower_subreg): Likewise.
12090 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12091 on the final iteration.
12092 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12093 a wider mode exists before asking for a move pattern.
12094 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12095 forcing a wider mode to exist.
12096 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12097 returning false if no such mode exists.
12098 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12099 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12100 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12101 Avoid checking for a MODE_INT if we already know the mode is not a
12103 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12104 forcing a wider mode to exist.
12105 (expmed_mult_highpart_optab): Likewise.
12106 (expmed_mult_highpart): Likewise.
12107 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12109 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12110 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12111 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12112 * internal-fn.c (expand_mul_overflow): Update use of
12113 GET_MODE_2XWIDER_MODE.
12114 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12115 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12116 GET_MODE_WIDER_MODE.
12117 (convert_plusminus_to_widen): Likewise.
12118 * tree-switch-conversion.c (array_value_type): Likewise.
12119 * var-tracking.c (emit_note_insn_var_location): Likewise.
12120 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12121 Return false inside rather than outside the loop if no wider mode
12123 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12124 and GET_MODE_2XWIDER_MODE
12125 (can_compare_p): Use else_void.
12126 * gdbhooks.py (OptMachineModePrinter): New class.
12127 (build_pretty_printer): Use it for opt_mode.
12129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12134 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12137 Alan Hayward <alan.hayward@arm.com>
12138 David Sherwood <david.sherwood@arm.com>
12140 * machmode.h (mode_traits): New structure.
12141 (get_narrowest_mode): New function.
12142 (mode_iterator::start): Likewise.
12143 (mode_iterator::iterate_p): Likewise.
12144 (mode_iterator::get_wider): Likewise.
12145 (mode_iterator::get_known_wider): Likewise.
12146 (mode_iterator::get_2xwider): Likewise.
12147 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12148 (FOR_EACH_MODE): Likewise.
12149 (FOR_EACH_MODE_FROM): Likewise.
12150 (FOR_EACH_MODE_UNTIL): Likewise.
12151 (FOR_EACH_WIDER_MODE): Likewise.
12152 (FOR_EACH_2XWIDER_MODE): Likewise.
12153 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12154 * combine.c (simplify_comparison): Likewise
12155 * config/i386/i386.c (type_natural_mode): Likewise.
12156 * cse.c (cse_insn): Likewise.
12157 * dse.c (find_shift_sequence): Likewise.
12158 * emit-rtl.c (init_derived_machine_modes): Likewise.
12159 (init_emit_once): Likewise.
12160 * explow.c (hard_function_value): Likewise.
12161 * expmed.c (extract_fixed_bit_field_1): Likewise.
12162 (extract_bit_field_1): Likewise.
12163 (expand_divmod): Likewise.
12164 (emit_store_flag_1): Likewise.
12165 * expr.c (init_expr_target): Likewise.
12166 (convert_move): Likewise.
12167 (alignment_for_piecewise_move): Likewise.
12168 (widest_int_mode_for_size): Likewise.
12169 (emit_block_move_via_movmem): Likewise.
12170 (copy_blkmode_to_reg): Likewise.
12171 (set_storage_via_setmem): Likewise.
12172 (compress_float_constant): Likewise.
12173 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12174 * optabs-query.c (get_best_extraction_insn): Likewise.
12175 * optabs.c (expand_binop): Likewise.
12176 (expand_twoval_unop): Likewise.
12177 (expand_twoval_binop): Likewise.
12178 (widen_leading): Likewise.
12179 (widen_bswap): Likewise.
12180 (expand_parity): Likewise.
12181 (expand_unop): Likewise.
12182 (prepare_cmp_insn): Likewise.
12183 (prepare_float_lib_cmp): Likewise.
12184 (expand_float): Likewise.
12185 (expand_fix): Likewise.
12186 (expand_sfix_optab): Likewise.
12187 * postreload.c (move2add_use_add2_insn): Likewise.
12188 * reg-stack.c (reg_to_stack): Likewise.
12189 * reginfo.c (choose_hard_reg_mode): Likewise.
12190 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12191 * stor-layout.c (mode_for_size): Likewise.
12192 (smallest_mode_for_size): Likewise.
12193 (mode_for_vector): Likewise.
12194 (finish_bitfield_representative): Likewise.
12195 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12196 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12197 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12198 * var-tracking.c (prepare_call_arguments): Likewise.
12200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12201 Alan Hayward <alan.hayward@arm.com>
12202 David Sherwood <david.sherwood@arm.com>
12204 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12205 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12206 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12207 * machmode.h (mode_size): Move earlier in file.
12208 (mode_precision): Likewise.
12209 (mode_inner): Likewise.
12210 (mode_nunits): Likewise.
12211 (mode_unit_size): Likewise.
12212 (unit_unit_precision): Likewise.
12213 (mode_wider): Likewise.
12214 (mode_2xwider): Likewise.
12215 (machine_mode): New class.
12216 (mode_to_bytes): New function.
12217 (mode_to_bits): Likewise.
12218 (mode_to_precision): Likewise.
12219 (mode_to_inner): Likewise.
12220 (mode_to_unit_size): Likewise.
12221 (mode_to_unit_precision): Likewise.
12222 (mode_to_nunits): Likewise.
12223 (GET_MODE_SIZE): Use mode_to_bytes.
12224 (GET_MODE_BITSIZE): Use mode_to_bits.
12225 (GET_MODE_PRECISION): Use mode_to_precision.
12226 (GET_MODE_INNER): Use mode_to_inner.
12227 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12228 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12229 (GET_MODE_NUNITS): Use mode_to_nunits.
12230 * system.h (ALWAYS_INLINE): New macro.
12231 * config/powerpcspe/powerpcspe-c.c
12232 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12233 int for arg1_mode and arg2_mode.
12235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12236 Alan Hayward <alan.hayward@arm.com>
12237 David Sherwood <david.sherwood@arm.com>
12239 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12240 Prefix mode names with E_ in case statements.
12241 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12242 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12243 (aarch64_split_simd_move): Likewise.
12244 (aarch64_gen_storewb_pair): Likewise.
12245 (aarch64_gen_loadwb_pair): Likewise.
12246 (aarch64_gen_store_pair): Likewise.
12247 (aarch64_gen_load_pair): Likewise.
12248 (aarch64_get_condition_code_1): Likewise.
12249 (aarch64_constant_pool_reload_icode): Likewise.
12250 (get_rsqrte_type): Likewise.
12251 (get_rsqrts_type): Likewise.
12252 (get_recpe_type): Likewise.
12253 (get_recps_type): Likewise.
12254 (aarch64_gimplify_va_arg_expr): Likewise.
12255 (aarch64_simd_container_mode): Likewise.
12256 (aarch64_emit_load_exclusive): Likewise.
12257 (aarch64_emit_store_exclusive): Likewise.
12258 (aarch64_expand_compare_and_swap): Likewise.
12259 (aarch64_gen_atomic_cas): Likewise.
12260 (aarch64_emit_bic): Likewise.
12261 (aarch64_emit_atomic_swap): Likewise.
12262 (aarch64_emit_atomic_load_op): Likewise.
12263 (aarch64_evpc_trn): Likewise.
12264 (aarch64_evpc_uzp): Likewise.
12265 (aarch64_evpc_zip): Likewise.
12266 (aarch64_evpc_ext): Likewise.
12267 (aarch64_evpc_rev): Likewise.
12268 (aarch64_evpc_dup): Likewise.
12269 (aarch64_gen_ccmp_first): Likewise.
12270 (aarch64_gen_ccmp_next): Likewise.
12271 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12272 (alpha_emit_xfloating_libcall): Likewise.
12273 (emit_insxl): Likewise.
12274 (alpha_arg_type): Likewise.
12275 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12276 (arc_preferred_simd_mode): Likewise.
12277 (arc_secondary_reload): Likewise.
12278 (get_arc_condition_code): Likewise.
12279 (arc_print_operand): Likewise.
12280 (arc_legitimate_constant_p): Likewise.
12281 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12282 * config/arc/arc.md (casesi_load): Likewise.
12283 (casesi_compact_jump): Likewise.
12284 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12285 (cc_use_register): Likewise.
12286 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12287 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12288 (arm_init_iwmmxt_builtins): Likewise.
12289 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12290 (neon_expand_vector_init): Likewise.
12291 (arm_attr_length_move_neon): Likewise.
12292 (maybe_get_arm_condition_code): Likewise.
12293 (arm_emit_vector_const): Likewise.
12294 (arm_preferred_simd_mode): Likewise.
12295 (arm_output_iwmmxt_tinsr): Likewise.
12296 (thumb1_output_casesi): Likewise.
12297 (thumb2_output_casesi): Likewise.
12298 (arm_emit_load_exclusive): Likewise.
12299 (arm_emit_store_exclusive): Likewise.
12300 (arm_expand_compare_and_swap): Likewise.
12301 (arm_evpc_neon_vuzp): Likewise.
12302 (arm_evpc_neon_vzip): Likewise.
12303 (arm_evpc_neon_vrev): Likewise.
12304 (arm_evpc_neon_vtrn): Likewise.
12305 (arm_evpc_neon_vext): Likewise.
12306 (arm_validize_comparison): Likewise.
12307 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12308 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12309 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12310 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12311 (c6x_preferred_simd_mode): Likewise.
12312 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12313 (epiphany_rtx_costs): Likewise.
12314 * config/epiphany/predicates.md (proper_comparison_operator):
12316 * config/frv/frv.c (condexec_memory_operand): Likewise.
12317 (frv_emit_move): Likewise.
12318 (output_move_single): Likewise.
12319 (output_condmove_single): Likewise.
12320 (frv_hard_regno_mode_ok): Likewise.
12321 (frv_matching_accg_mode): Likewise.
12322 * config/h8300/h8300.c (split_adds_subs): Likewise.
12323 (h8300_rtx_costs): Likewise.
12324 (h8300_print_operand): Likewise.
12325 (compute_mov_length): Likewise.
12326 (output_logical_op): Likewise.
12327 (compute_logical_op_length): Likewise.
12328 (compute_logical_op_cc): Likewise.
12329 (h8300_shift_needs_scratch_p): Likewise.
12330 (output_a_shift): Likewise.
12331 (compute_a_shift_length): Likewise.
12332 (compute_a_shift_cc): Likewise.
12333 (expand_a_rotate): Likewise.
12334 (output_a_rotate): Likewise.
12335 * config/i386/i386.c (classify_argument): Likewise.
12336 (function_arg_advance_32): Likewise.
12337 (function_arg_32): Likewise.
12338 (function_arg_64): Likewise.
12339 (function_value_64): Likewise.
12340 (ix86_gimplify_va_arg): Likewise.
12341 (ix86_legitimate_constant_p): Likewise.
12342 (put_condition_code): Likewise.
12343 (split_double_mode): Likewise.
12344 (ix86_avx256_split_vector_move_misalign): Likewise.
12345 (ix86_expand_vector_logical_operator): Likewise.
12346 (ix86_split_idivmod): Likewise.
12347 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12348 (ix86_build_const_vector): Likewise.
12349 (ix86_build_signbit_mask): Likewise.
12350 (ix86_match_ccmode): Likewise.
12351 (ix86_cc_modes_compatible): Likewise.
12352 (ix86_expand_branch): Likewise.
12353 (ix86_expand_sse_cmp): Likewise.
12354 (ix86_expand_sse_movcc): Likewise.
12355 (ix86_expand_int_sse_cmp): Likewise.
12356 (ix86_expand_vec_perm_vpermi2): Likewise.
12357 (ix86_expand_vec_perm): Likewise.
12358 (ix86_expand_sse_unpack): Likewise.
12359 (ix86_expand_int_addcc): Likewise.
12360 (ix86_split_to_parts): Likewise.
12361 (ix86_vectorize_builtin_gather): Likewise.
12362 (ix86_vectorize_builtin_scatter): Likewise.
12363 (avx_vpermilp_parallel): Likewise.
12364 (inline_memory_move_cost): Likewise.
12365 (ix86_tieable_integer_mode_p): Likewise.
12366 (x86_maybe_negate_const_int): Likewise.
12367 (ix86_expand_vector_init_duplicate): Likewise.
12368 (ix86_expand_vector_init_one_nonzero): Likewise.
12369 (ix86_expand_vector_init_one_var): Likewise.
12370 (ix86_expand_vector_init_concat): Likewise.
12371 (ix86_expand_vector_init_interleave): Likewise.
12372 (ix86_expand_vector_init_general): Likewise.
12373 (ix86_expand_vector_set): Likewise.
12374 (ix86_expand_vector_extract): Likewise.
12375 (emit_reduc_half): Likewise.
12376 (ix86_emit_i387_round): Likewise.
12377 (ix86_mangle_type): Likewise.
12378 (ix86_expand_round_sse4): Likewise.
12379 (expand_vec_perm_blend): Likewise.
12380 (canonicalize_vector_int_perm): Likewise.
12381 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12382 (expand_vec_perm_1): Likewise.
12383 (expand_vec_perm_interleave3): Likewise.
12384 (expand_vec_perm_even_odd_pack): Likewise.
12385 (expand_vec_perm_even_odd_1): Likewise.
12386 (expand_vec_perm_broadcast_1): Likewise.
12387 (ix86_vectorize_vec_perm_const_ok): Likewise.
12388 (ix86_expand_vecop_qihi): Likewise.
12389 (ix86_expand_mul_widen_hilo): Likewise.
12390 (ix86_expand_sse2_abs): Likewise.
12391 (ix86_expand_pextr): Likewise.
12392 (ix86_expand_pinsr): Likewise.
12393 (ix86_preferred_simd_mode): Likewise.
12394 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12395 * config/i386/sse.md (*andnot<mode>3): Likewise.
12396 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12397 (*<code><mode>3): Likewise.
12398 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12399 (ia64_expand_atomic_op): Likewise.
12400 (ia64_arg_type): Likewise.
12401 (ia64_mode_to_int): Likewise.
12402 (ia64_scalar_mode_supported_p): Likewise.
12403 (ia64_vector_mode_supported_p): Likewise.
12404 (expand_vec_perm_broadcast): Likewise.
12405 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12406 (iq2000_function_arg_advance): Likewise.
12407 (iq2000_function_arg): Likewise.
12408 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12409 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12410 (m68k_libcall_value): Likewise.
12411 (m68k_function_value): Likewise.
12412 (sched_attr_op_type): Likewise.
12413 * config/mcore/mcore.c (mcore_output_move): Likewise.
12414 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12416 (microblaze_function_arg): Likewise.
12417 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12418 (mips_print_operand): Likewise.
12419 (mips_mode_ok_for_mov_fmt_p): Likewise.
12420 (mips_vector_mode_supported_p): Likewise.
12421 (mips_preferred_simd_mode): Likewise.
12422 (mips_expand_vpc_loongson_even_odd): Likewise.
12423 (mips_expand_vec_unpack): Likewise.
12424 (mips_expand_vi_broadcast): Likewise.
12425 (mips_expand_vector_init): Likewise.
12426 (mips_expand_vec_reduc): Likewise.
12427 (mips_expand_msa_cmp): Likewise.
12428 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12429 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12430 (cc_flags_for_mode): Likewise.
12431 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12432 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12433 (nds32_output_casesi_pc_relative): Likewise.
12434 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12435 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12436 (nvptx_gen_unpack): Likewise.
12437 (nvptx_gen_pack): Likewise.
12438 (nvptx_gen_shuffle): Likewise.
12439 (nvptx_gen_wcast): Likewise.
12440 (nvptx_preferred_simd_mode): Likewise.
12441 * config/pa/pa.c (pa_secondary_reload): Likewise.
12442 * config/pa/predicates.md (base14_operand): Likewise.
12443 * config/powerpcspe/powerpcspe-c.c
12444 (altivec_resolve_overloaded_builtin): Likewise.
12445 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12447 (rs6000_preferred_simd_mode): Likewise.
12448 (output_vec_const_move): Likewise.
12449 (rs6000_expand_vector_extract): Likewise.
12450 (rs6000_split_vec_extract_var): Likewise.
12451 (reg_offset_addressing_ok_p): Likewise.
12452 (rs6000_legitimate_offset_address_p): Likewise.
12453 (rs6000_legitimize_address): Likewise.
12454 (rs6000_emit_set_const): Likewise.
12455 (rs6000_const_vec): Likewise.
12456 (rs6000_emit_move): Likewise.
12457 (spe_build_register_parallel): Likewise.
12458 (rs6000_darwin64_record_arg_recurse): Likewise.
12459 (swap_selector_for_mode): Likewise.
12460 (spe_init_builtins): Likewise.
12461 (paired_init_builtins): Likewise.
12462 (altivec_init_builtins): Likewise.
12463 (do_load_for_compare): Likewise.
12464 (rs6000_generate_compare): Likewise.
12465 (rs6000_expand_float128_convert): Likewise.
12466 (emit_load_locked): Likewise.
12467 (emit_store_conditional): Likewise.
12468 (rs6000_output_function_epilogue): Likewise.
12469 (rs6000_handle_altivec_attribute): Likewise.
12470 (rs6000_function_value): Likewise.
12471 (emit_fusion_gpr_load): Likewise.
12472 (emit_fusion_p9_load): Likewise.
12473 (emit_fusion_p9_store): Likewise.
12474 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12475 (fusion_gpr_mem_load): Likewise.
12476 (fusion_addis_mem_combo_load): Likewise.
12477 (fusion_addis_mem_combo_store): Likewise.
12478 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12479 (fusion_gpr_mem_load): Likewise.
12480 (fusion_addis_mem_combo_load): Likewise.
12481 (fusion_addis_mem_combo_store): Likewise.
12482 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12484 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12485 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12486 (rs6000_preferred_simd_mode): Likewise.
12487 (output_vec_const_move): Likewise.
12488 (rs6000_expand_vector_extract): Likewise.
12489 (rs6000_split_vec_extract_var): Likewise.
12490 (reg_offset_addressing_ok_p): Likewise.
12491 (rs6000_legitimate_offset_address_p): Likewise.
12492 (rs6000_legitimize_address): Likewise.
12493 (rs6000_emit_set_const): Likewise.
12494 (rs6000_const_vec): Likewise.
12495 (rs6000_emit_move): Likewise.
12496 (rs6000_darwin64_record_arg_recurse): Likewise.
12497 (swap_selector_for_mode): Likewise.
12498 (paired_init_builtins): Likewise.
12499 (altivec_init_builtins): Likewise.
12500 (rs6000_expand_float128_convert): Likewise.
12501 (emit_load_locked): Likewise.
12502 (emit_store_conditional): Likewise.
12503 (rs6000_output_function_epilogue): Likewise.
12504 (rs6000_handle_altivec_attribute): Likewise.
12505 (rs6000_function_value): Likewise.
12506 (emit_fusion_gpr_load): Likewise.
12507 (emit_fusion_p9_load): Likewise.
12508 (emit_fusion_p9_store): Likewise.
12509 * config/rx/rx.c (rx_gen_move_template): Likewise.
12510 (flags_from_mode): Likewise.
12511 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12512 (s390_slb_comparison): Likewise.
12513 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12514 (s390_vector_mode_supported_p): Likewise.
12515 (s390_cc_modes_compatible): Likewise.
12516 (s390_match_ccmode_set): Likewise.
12517 (s390_canonicalize_comparison): Likewise.
12518 (s390_emit_compare_and_swap): Likewise.
12519 (s390_branch_condition_mask): Likewise.
12520 (s390_rtx_costs): Likewise.
12521 (s390_secondary_reload): Likewise.
12522 (__SECONDARY_RELOAD_CASE): Likewise.
12523 (s390_expand_cs): Likewise.
12524 (s390_preferred_simd_mode): Likewise.
12525 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12526 * config/sh/sh.c (sh_print_operand): Likewise.
12527 (dump_table): Likewise.
12528 (sh_secondary_reload): Likewise.
12529 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12530 * config/sh/sh.md (casesi_worker_1): Likewise.
12531 (casesi_worker_2): Likewise.
12532 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12533 (fcc_comparison_operator): Likewise.
12534 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12535 (emit_soft_tfmode_cvt): Likewise.
12536 (sparc_preferred_simd_mode): Likewise.
12537 (output_cbranch): Likewise.
12538 (sparc_print_operand): Likewise.
12539 (sparc_expand_vec_perm_bmask): Likewise.
12540 (vector_init_bshuffle): Likewise.
12541 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12542 (spu_vector_mode_supported_p): Likewise.
12543 (spu_expand_insv): Likewise.
12544 (spu_emit_branch_or_set): Likewise.
12545 (spu_handle_vector_attribute): Likewise.
12546 (spu_builtin_splats): Likewise.
12547 (spu_builtin_extract): Likewise.
12548 (spu_builtin_promote): Likewise.
12549 (spu_expand_sign_extend): Likewise.
12550 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12551 (tilegx_simd_int): Likewise.
12552 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12553 (tilepro_simd_int): Likewise.
12554 * config/v850/v850.c (const_double_split): Likewise.
12555 (v850_print_operand): Likewise.
12556 (ep_memory_offset): Likewise.
12557 * config/vax/vax.c (vax_rtx_costs): Likewise.
12558 (vax_output_int_move): Likewise.
12559 (vax_output_int_add): Likewise.
12560 (vax_output_int_subtract): Likewise.
12561 * config/visium/predicates.md (visium_branch_operator): Likewise.
12562 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12563 (visium_print_operand_address): Likewise.
12564 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12565 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12566 (xtensa_expand_conditional_branch): Likewise.
12567 (xtensa_copy_incoming_a7): Likewise.
12568 (xtensa_output_literal): Likewise.
12569 * dfp.c (decimal_real_maxval): Likewise.
12570 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12573 Alan Hayward <alan.hayward@arm.com>
12574 David Sherwood <david.sherwood@arm.com>
12576 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12577 (mode_nunits_inline): Likewise.
12578 (mode_inner_inline): Likewise.
12579 (mode_unit_size_inline): Likewise.
12580 (mode_unit_precision_inline): Likewise.
12581 (emit_insn_modes_h): Likewise. Also emit a #define of the
12583 (emit_mode_wider): Add an E_ prefix to mode names.
12584 (emit_mode_complex): Likewise.
12585 (emit_mode_inner): Likewise.
12586 (emit_mode_adjustments): Likewise.
12587 (emit_mode_int_n): Likewise.
12588 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12589 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12590 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12591 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12592 (CRC32_BUILTIN, ENTRY): Likewise.
12593 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12594 (aarch64_pop_regs): Likewise.
12595 (aarch64_process_components): Likewise.
12596 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12597 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12598 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12599 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12600 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12601 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12603 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12604 (pcmpistr): Likewise.
12605 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12606 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12607 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12609 * config/rl78/rl78.c (mduc_regs): Likewise.
12610 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12611 (htm_expand_builtin): Likewise.
12612 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12613 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12614 * config/xtensa/xtensa.c (print_operand): Likewise.
12615 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12616 (NUM_MODE_VECTOR_INT): Likewise.
12617 * genoutput.c (null_operand): Likewise.
12618 (output_operand_data): Likewise.
12619 * genrecog.c (print_parameter_value): Likewise.
12620 * lra.c (debug_operand_data): Likewise.
12622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12624 * dwarf2out.c (scompare_loc_descriptor_wide)
12625 (scompare_loc_descriptor_narrow): New functions, split out from...
12626 (scompare_loc_descriptor): ...here.
12627 * expmed.c (emit_store_flag_int): New function, split out from...
12628 (emit_store_flag): ...here.
12630 2017-08-30 Richard Biener <rguenther@suse.de>
12632 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12633 (dwarf2out_early_finish): Move setting of AT_pubnames from
12634 early debug output to early finish.
12636 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12638 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12639 and -mdata-region to the assembler.
12641 2017-08-30 Richard Biener <rguenther@suse.de>
12643 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12645 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12646 (dwarf2out_early_global_decl): Only generate a DIE for the
12647 abstract origin if it doesn't already exist or is a declaration DIE.
12648 (resolve_addr): Do not add the linkage name twice when
12649 generating a stub DIE for the DW_TAG_GNU_call_site target.
12651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12654 Use machine_mode rather than int for arg1_mode.
12656 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12659 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12660 that the second argument of the built-in functions to unpack
12661 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12662 switch statement instead a lot of if statements.
12663 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12664 Allow 64-bit values to be in Altivec registers as well as
12665 traditional floating point registers.
12666 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12668 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12670 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12671 MAX_REGS_PER_ADDRESS == 1.
12673 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12675 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12676 * config/i386/i386.c (ix86_option_override_internal): Simplify
12677 setting of opts->x_flag_entry.
12679 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12680 Jakub Jelinek <jakub@redhat.com>
12681 Richard Biener <rguenther@suse.de>
12683 PR tree-optimization/81503
12684 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12685 folded constant fits in the target type; reorder tests for clarity.
12687 2017-08-29 Martin Liska <mliska@suse.cz>
12689 * passes.def: Include pass_lower_switch.
12690 * stmt.c (dump_case_nodes): Remove and move to
12691 tree-switch-conversion.
12692 (case_values_threshold): Likewise.
12693 (expand_switch_as_decision_tree_p): Likewise.
12694 (emit_case_decision_tree): Likewise.
12695 (expand_case): Likewise.
12696 (balance_case_nodes): Likewise.
12697 (node_has_low_bound): Likewise.
12698 (node_has_high_bound): Likewise.
12699 (node_is_bounded): Likewise.
12700 (emit_case_nodes): Likewise.
12701 (struct simple_case_node): New struct.
12702 (add_case_node): Remove.
12703 (emit_case_dispatch_table): Use vector instead of case_list.
12704 (reset_out_edges_aux): Remove.
12705 (compute_cases_per_edge): Likewise.
12706 (expand_case): Build list of simple_case_node.
12707 (expand_sjlj_dispatch_table): Use it.
12708 * tree-switch-conversion.c (struct case_node): Moved from
12709 stmt.c and adjusted.
12710 (emit_case_nodes): Likewise.
12711 (node_has_low_bound): Likewise.
12712 (node_has_high_bound): Likewise.
12713 (node_is_bounded): Likewise.
12714 (case_values_threshold): Likewise.
12715 (reset_out_edges_aux): Likewise.
12716 (compute_cases_per_edge): Likewise.
12717 (add_case_node): Likewise.
12718 (dump_case_nodes): Likewise.
12719 (balance_case_nodes): Likewise.
12720 (expand_switch_as_decision_tree_p): Likewise.
12721 (emit_jump): Likewise.
12722 (emit_case_decision_tree): Likewise.
12723 (try_switch_expansion): Likewise.
12724 (do_jump_if_equal): Likewise.
12725 (emit_cmp_and_jump_insns): Likewise.
12726 (fix_phi_operands_for_edge): New function.
12727 (record_phi_operand_mapping): Likewise.
12728 (class pass_lower_switch): New pass.
12729 (pass_lower_switch::execute): New function.
12730 (make_pass_lower_switch): Likewise.
12731 (conditional_probability):
12732 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12733 * tree-pass.h: Add make_pass_lower_switch.
12735 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12738 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12741 2017-08-29 Richard Biener <rguenther@suse.de>
12743 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12744 we do not add a DW_AT_inline attribute twice.
12745 (gen_subprogram_die): Remove code setting DW_AT_inline on
12746 DECL_ABSTRACT_P nodes.
12748 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12750 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12751 calls to internal functions.
12752 (gimplify_modify_expr): Likewise.
12753 * tree-call-cdce.c (use_internal_fn): Likewise.
12754 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12755 (convert_to_divmod): Set the nothrow flag.
12756 * tree-if-conv.c (predicate_mem_writes): Likewise.
12757 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12758 (vectorizable_call): Likewise.
12759 (vectorizable_store): Likewise.
12760 (vectorizable_load): Likewise.
12761 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12762 (vect_recog_mask_conversion_pattern): Likewise.
12764 2017-08-29 Martin Liska <mliska@suse.cz>
12767 * gcc.c (driver_handle_option): Add new argument.
12768 * opts-common.c (handle_option): Pass
12769 target_option_override_hook.
12770 * opts-global.c (lang_handle_option): Add new option.
12771 (set_default_handlers): Add new argument.
12772 (decode_options): Likewise.
12773 * opts.c (target_handle_option): Likewise.
12774 (common_handle_option): Call target_option_override_hook.
12775 * opts.h (struct cl_option_handler_func): Add hook for
12776 target option override.
12777 (struct cl_option_handlers): Likewise.
12778 (set_default_handlers): Add new argument.
12779 (decode_options): Likewise.
12780 (common_handle_option): Likewise.
12781 (target_handle_option): Likewise.
12782 * toplev.c (toplev::main): Pass targetm.target_option.override
12785 2017-08-29 Richard Biener <rguenther@suse.de>
12786 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12788 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12789 life to the active subtree.
12791 2017-08-28 Jeff Law <law@redhat.com>
12793 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12794 derive_equivalences.
12795 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12796 Code moved into....
12797 (edge_info::derive_equivalences): New private member function
12799 * tree-ssa-dom.c (class edge_info): Changed from a struct
12800 to a class. Add ctor/dtor, methods and data members.
12801 (edge_info::edge_info): Renamed from allocate_edge_info.
12802 Initialize additional members.
12803 (edge_info::~edge_info): New.
12804 (free_dom_edge_info): Delete the edge info.
12805 (record_edge_info): Use new class & associated member functions.
12806 Tighten forms for testing for edge equivalences.
12807 (record_temporary_equivalences): Iterate over the simple
12808 equivalences rather than assuming there's only one per edge.
12809 (cprop_into_successor_phis): Iterate over the simple
12810 equivalences rather than assuming there's only one per edge.
12811 (optimize_stmt): Use operand_equal_p rather than pointer
12812 equality for mini-DSE code.
12814 2017-08-28 Nathan Sidwell <nathan@acm.org>
12816 * gcc.c (execute): Fold SIGPIPE handling into switch
12817 statement. Adjust internal error message.
12819 2017-08-28 Richard Biener <rguenther@suse.de>
12822 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12823 Do nothing for removed DIEs.
12825 2017-08-28 Richard Biener <rguenther@suse.de>
12827 PR tree-optimization/81977
12828 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12831 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12834 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12835 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12836 using targetm.gen_mem_thread_fence.
12838 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12841 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12842 predicate to register_operand. Reorder operands.
12843 (*btr<mode>): Ditto.
12844 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12845 (*btr<mode>_mask): Ditto.
12847 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12849 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12850 * config/rs6000/xmmintrin.h: New file.
12851 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12853 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12856 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12857 parameter and_insn and return it.
12858 (recombine_lvx_pattern): Insert a copy to ensure availability of
12859 the base register of the copied masking operation at the point of
12860 the instruction replacement.
12861 (recombine_stvx_pattern): Likewise.
12863 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12865 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12866 undocumented switches.
12867 (-mpower9-dform-vector): Likewise.
12868 (-mpower9-dform): Likewise.
12869 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12870 comments to delete references to -mpower9-dform* switches.
12871 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12872 Delete reference to -mpower9-dform* switches, test for
12873 -mpower9-vector instead.
12874 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12875 (OTHER_P9_VECTOR_MASKS): Likewise.
12876 (POWERPC_MASKS): Likewise.
12877 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12878 tests against -mpower9-dform* to -mpower9-vector. Delete code
12879 that checked for -mpower9-dform* consistancy with other options.
12880 Add test for -mpower9-misc to enable other power9 switches.
12881 (rs6000_init_hard_regno_mode_ok): Likewise.
12882 (rs6000_option_override_internal): Likewise.
12883 (rs6000_emit_prologue): Likewise.
12884 (rs6000_emit_epilogue): Likewise.
12885 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12886 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12887 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12889 (emit_fusion_p9_store): Likewise.
12890 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12891 resetting these macros if the assembler does not support ISA 3.0
12893 (TARGET_P9_DFORM_VECTOR): Likewise.
12894 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12895 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12897 2017-08-25 Alan Modra <amodra@gmail.com>
12899 PR rtl-optimization/81747
12900 * cse.c (cse_extended_basic_block): Don't attempt to record
12901 equivalences for degenerate conditional branches that branch
12902 to their fall-through.
12904 2017-08-24 Martin Sebor <msebor@redhat.com>
12906 PR middle-end/81908
12907 * gimple-fold.c (size_must_be_zero_p): New function.
12908 (gimple_fold_builtin_memory_op): Call it.
12910 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12912 * config/rs6000/mm_malloc.h: New file.
12914 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12916 PR tree-optimization/81913
12917 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12918 analysis when either IVs in condition can wrap.
12920 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12922 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12923 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12925 2017-08-24 Richard Biener <rguenther@suse.de>
12928 * targhooks.c (default_target_can_inline_p): Properly
12929 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12930 is present and always compare.
12931 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12932 infer -mfpmath=sse from TARGET_SSE_P.
12933 (ix86_can_inline_p): Properly use target_option_default_node when
12934 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12936 2017-08-24 Richard Biener <rguenther@suse.de>
12939 * dwarf2out.c (output_die): Handle flag_generate_offload like
12941 (output_comp_unit): Likewise.
12942 (gen_array_type_die): Likewise.
12943 (dwarf2out_early_finish): Likewise.
12944 (note_variable_value_in_expr): Likewise.
12945 (dwarf2out_finish): Likewise. Adjust assert.
12946 * cgraphunit.c (symbol_table::compile): Move setting of
12947 flag_generate_offload earlier ...
12948 (symbol_table::finalize_compilation_unit): ... here, before
12949 early debug finalization.
12951 2017-08-24 Richard Biener <rguenther@suse.de>
12953 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12954 and ipa-fnsummary.h.
12955 (ix86_can_inline_p): When ix86_fpmath flags do not match
12956 check whether the callee uses FP math at all.
12958 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12960 PR middle-end/81931
12961 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12962 instead of TYPE_PRECISION.
12964 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12965 Alan Hayward <alan.hayward@arm.com>
12966 David Sherwood <david.sherwood@arm.com>
12968 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12970 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12972 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12974 (ix86_frame::stack_realign_allocate): New field.
12975 (struct machine_frame_state): Modify comments.
12976 (machine_frame_state::sp_realigned_fp_end): New field.
12977 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12978 layout calculation.
12979 (sp_valid_at): Add assertion to assure no attempt to access invalid
12980 offset of a realigned stack.
12981 (fp_valid_at): Likewise.
12982 (choose_baseaddr): Modify comments.
12983 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12984 ix86_expand_prologue.
12985 (ix86_expand_prologue): Modify stack realignment and allocation.
12986 (ix86_expand_epilogue): Modify comments.
12987 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12988 avx2_runtime, avx512f, and avx512f_runtime.
12990 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12992 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12993 (mstackrealign): Do not init to -1.
12994 * config/i386/i386.c (ix86_option_override_internal):
12995 Check opts_set, not opts when setting default value of
12996 opts->x_ix86_force_align_arg_pointer.
12998 2017-08-23 Richard Biener <rguenther@suse.de>
13000 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13001 lang_hooks.decl_printable_name.
13002 * print-rtl-function.c (print_rtx_function): Likewise.
13003 * tree-pretty-print.c (dump_function_header): Likewise.
13005 2017-08-23 Richard Biener <rguenther@suse.de>
13008 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13009 -g0 at compile-time.
13011 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13013 PR middle-end/19706
13014 * doc/sourcebuild.texi (Other hardware attributes):
13017 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13019 PR middle-end/19706
13020 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13021 Fix single-use check.
13023 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13025 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13027 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13029 * doc/install.texi: Modify to add more details on running selected
13032 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13034 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13035 is combined with -mabi=ms.
13036 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13039 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13041 PR tree-optimization/81488
13042 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13043 and cached_basis fields.
13044 (MAX_SPREAD): New constant.
13045 (alloc_cand_and_find_basis): Initialize new fields.
13046 (clear_visited): New function.
13047 (create_phi_basis_1): Rename from create_phi_basis, set visited
13048 and cached_basis fields.
13049 (create_phi_basis): New wrapper function.
13050 (phi_add_costs_1): Rename from phi_add_costs, add spread
13051 parameter, set visited field, short-circuit when limits reached.
13052 (phi_add_costs): New wrapper function.
13053 (record_phi_increments_1): Rename from record_phi_increments, set
13055 (record_phi_increments): New wrapper function.
13056 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13057 (phi_incr_cost): New wrapper function.
13058 (all_phi_incrs_profitable_1): Rename from
13059 all_phi_incrs_profitable, set visited field.
13060 (all_phi_incrs_profitable): New wrapper function.
13062 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13063 Alan Hayward <alan.hayward@arm.com>
13064 David Sherwood <david.sherwood@arm.com>
13066 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13067 that takes the outer and inner modes.
13068 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13069 comparison as the canonical test for a paradoxical subreg.
13070 * combine.c (simplify_set): Use paradoxical_subreg_p.
13071 (make_extraction): Likewise.
13072 (force_to_mode): Likewise.
13073 (rtx_equal_for_field_assignment_p): Likewise.
13074 (gen_lowpart_for_combine): Likewise.
13075 (simplify_comparison): Likewise.
13076 * cse.c (equiv_constant): Likewise.
13077 * expmed.c (store_bit_field_1): Likewise.
13078 * final.c (alter_subreg): Likewise.
13079 * fwprop.c (propagate_rtx): Likewise.
13080 (forward_propagate_subreg): Likewise.
13081 * ira-conflicts.c (ira_build_conflicts): Likewise.
13082 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13083 * lra-constraints.c (curr_insn_transform): Likewise.
13084 (split_reg): Likewise.
13085 * lra-eliminations.c (move_plus_up): Likewise.
13086 (lra_eliminate_regs_1): Likewise.
13087 * recog.c (general_operand): Likewise.
13088 * ree.c (combine_reaching_defs): Likewise.
13089 * reload.c (push_reload): Likewise.
13090 (find_reloads): Likewise.
13091 * reload1.c (elimination_effects): Likewise.
13092 (compute_reload_subreg_offset): Likewise.
13093 (choose_reload_regs): Likewise.
13094 * rtlanal.c (subreg_lsb_1): Likewise.
13095 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13096 (simplify_subreg): Likewise.
13097 * var-tracking.c (track_loc_p): Likewise.
13098 * emit-rtl.c (byte_lowpart_offset): Likewise.
13099 (paradoxical_subreg_p): Delete out-of-line definition.
13101 2017-08-22 Jeff Law <law@redhat.com>
13103 PR tree-optimization/81741
13104 PR tree-optimization/71947
13105 * tree-ssa-dom.c: Include tree-inline.h.
13106 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13107 equivalences if one is more expensive to compute than the other.
13108 * tree-ssa-scopedtables.h (class const_or_copies): Make
13109 record_const_or_copy_raw method private.
13110 (class avail_exprs_stack): New method simplify_binary_operation.
13111 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13112 avail_exprs_stack::simplify_binary_operation as needed.
13113 (avail_exprs_stack::simplify_binary_operation): New function.
13115 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13117 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13118 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13119 (DOT_SYMBOLS): Likewise.
13120 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13121 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13122 (RS6000_ABI_NAME): Likewise.
13123 (TARGET_CMODEL): Likewise.
13124 (TOC_SECTION_ASM_OP): Likewise.
13125 (SET_CMODEL): New macro.
13126 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13128 2017-08-22 Richard Biener <rguenther@suse.de>
13130 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13131 to allow for free-lang-data replacements similar to verify_type_variant.
13133 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13134 Alan Hayward <alan.hayward@arm.com>
13135 David Sherwood <david.sherwood@arm.com>
13137 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13138 VOIDmode for the LABEL_REF.
13140 2017-08-22 Richard Biener <rguenther@suse.de>
13142 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13144 2017-08-22 Richard Biener <rguenther@suse.de>
13146 * common.opt (feliminate-dwarf2-dups): Ignore.
13147 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13148 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13149 same_die_p_wrap, compute_section_prefix,
13150 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13151 (comdat_symbol_id, comdat_symbol_number): Likewise.
13152 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13154 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13155 (output_die): Mark unreachable path unreachable.
13156 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13157 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13158 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13159 (dwarf2out_early_finish): Likewise.
13161 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13163 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13165 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13168 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13169 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13170 (avr_attribute_table) <io, io_low, address>: Initialize
13171 .decl_required with true.
13173 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13175 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13176 undocumented debugging options.
13177 (-mvsx-scalar-double): Likewise.
13178 (-mallow-df-permute): Likewise.
13179 (-mvectorize-builtins): Likewise.
13180 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13181 (rs6000_builtin_vectorized_function): Likewise.
13182 (rs6000_builtin_md_vectorized_function): Likewise.
13183 (rs6000_opt_vars): Likewise.
13185 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13188 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13189 (*btrq_imm): Rename from *btrq.
13190 (*btcq_imm): Rename from *btcq.
13191 (btsc): New code attribute.
13192 (*<btsc><mode>): New insn pattern.
13193 (*btr<mode>): Ditto.
13194 (*<btsc><mode>_mask): New insn_and_split pattern.
13195 (*btr<mode>_mask): Ditto.
13197 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13198 Alan Hayward <alan.hayward@arm.com>
13199 David Sherwood <david.sherwood@arm.com>
13201 * function.c (pad_below): Simplify padding calculation.
13203 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13204 Alan Hayward <alan.hayward@arm.com>
13205 David Sherwood <david.sherwood@arm.com>
13207 * target.def (function_prologue): Remove frame size argument.
13208 (function_epilogue): Likewise.
13209 * doc/tm.texi: Regenerate.
13210 * final.c (final_start_function): Update call to function_prologue.
13211 (final_end_function): Update call to function_epilogue.
13212 (default_function_pro_epilogue): Remove frame size argument.
13213 * output.h (default_function_pro_epilogue): Likewise.
13214 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13215 (arm_output_function_prologue): Likewise.
13216 * config/frv/frv.c (frv_function_prologue): Likewise.
13217 (frv_function_epilogue): Likewise.
13218 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13219 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13220 (ia64_output_function_epilogue): Likewise.
13221 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13222 (m32r_output_function_epilogue): Likewise.
13223 * config/microblaze/microblaze.c (microblaze_function_prologue)
13224 (microblaze_function_epilogue): Likewise.
13225 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13226 (mips_output_function_epilogue): Likewise.
13227 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13228 (mmix_target_asm_function_epilogue): Likewise.
13229 * config/msp430/msp430.c (msp430_start_function): Likewise.
13230 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13231 (nds32_asm_function_epilogue): Likewise.
13232 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13233 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13234 (pa_output_function_epilogue): Likewise.
13235 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13236 (rs6000_output_function_epilogue): Likewise.
13237 * config/rl78/rl78.c (rl78_start_function): Likewise.
13238 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13239 (rs6000_output_function_epilogue): Likewise.
13240 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13241 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13242 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13243 (sparc_asm_function_epilogue): Likewise.
13245 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13247 * tree.h (type_has_mode_precision_p): New function.
13248 * convert.c (convert_to_integer_1): Use it.
13249 * expr.c (expand_expr_real_2): Likewise.
13250 (expand_expr_real_1): Likewise.
13251 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13252 * match.pd: Likewise.
13253 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13254 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13255 * tree-tailcall.c (process_assignment): Likewise.
13256 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13257 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13258 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13259 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13260 (vectorizable_assignment): Likewise.
13261 (vectorizable_shift): Likewise.
13262 (vectorizable_operation): Likewise.
13263 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13265 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13267 * match.pd: Add pow (C, x) simplification.
13269 2017-08-21 Richard Biener <rguenther@suse.de>
13271 PR tree-optimization/81900
13272 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13273 for blocks with abnormal predecessors.
13274 (compute_antic): Do not set visited flag prematurely.
13276 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13279 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13281 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13283 * stor-layout.h (vector_type_mode): Move to...
13284 * tree.h (vector_type_mode): ...here.
13285 * stor-layout.c (vector_type_mode): Move to...
13286 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13288 2017-08-21 Richard Biener <rguenther@suse.de>
13290 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13291 register_external_die hooks.
13292 (debug_false_tree_charstarstar_uhwistar): Declare.
13293 (debug_nothing_tree_charstar_uhwi): Likewise.
13294 * debug.c (do_nothing_debug_hooks): Adjust.
13295 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13296 (debug_nothing_tree_charstar_uhwi): Likewise.
13297 * dbxout.c (dbx_debug_hooks): Adjust.
13298 (xcoff_debug_hooks): Likewise.
13299 * sdbout.c (sdb_debug_hooks): Likewise.
13300 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13301 * dwarf2out.c (macinfo_label_base): New global.
13302 (dwarf2out_register_external_die): New function for the
13303 register_external_die hook.
13304 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13305 (dwarf2_debug_hooks): Use them.
13306 (dwarf2_lineno_debug_hooks): Adjust.
13307 (struct die_struct): Add with_offset flag.
13308 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13309 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13310 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13311 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13312 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13313 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13314 defining section names for the early LTO debug variants.
13315 (reset_indirect_string): New helper.
13316 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13317 (print_dw_val): Add support for offsetted symbol references.
13318 (get_ultimate_context): Split out from is_cxx.
13319 (is_cxx): Use get_ultimate_context.
13320 (is_fortran): Add decl overload.
13321 (compute_comp_unit_symbol): Split out worker from
13322 compute_section_prefix.
13323 (compute_section_prefix): Call compute_comp_unit_symbol and
13324 set comdat_type_p here.
13325 (output_die): Skip DIE symbol output for the LTO added one.
13326 Handle DIE symbol references with offset.
13327 (output_comp_unit): Guard section name mangling properly.
13328 For LTO debug sections emit a symbol at the section beginning
13329 which we use to refer to its DIEs.
13330 (add_abstract_origin_attribute): For DIEs registered via
13331 dwarf2out_register_external_die directly refer to the early
13332 DIE rather than indirectly through the shadow one we created.
13333 Remove obsolete call to dwarf2out_abstract_function for
13334 non-function/block origins.
13335 (gen_array_type_die): When generating early LTO debug do
13336 not emit DW_AT_string_length.
13337 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13338 late when in LTO. As suggested place a gcc_unreachable for
13339 the DECL_ABSTRACT_P case.
13340 (gen_subprogram_die): Avoid another specification DIE
13341 for early built declarations/definitions for the late LTO case.
13342 (gen_variable_die): Add type references for late duplicated VLA dies
13344 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13345 we have the abstract instance already.
13346 (process_scope_var): Adjust decl DIE contexts in LTO which
13347 first puts them in limbo.
13348 (gen_decl_die): Do not generate type DIEs late apart from
13349 types for VLAs or for decls we do not yet have a DIE. Do not
13350 call dwarf2out_abstract_function late.
13351 (dwarf2out_early_global_decl): Make sure to create DIEs
13352 for abstract instances of a decl first.
13353 (dwarf2out_late_global_decl): Adjust comment.
13354 (output_macinfo_op): With multiple macro sections use
13355 macinfo_label_base to distinguish labels.
13356 (output_macinfo): Likewise. Update macinfo_label_base.
13357 Pass in the line info label.
13358 (note_variable_value_in_expr): When generating LTO resolve
13359 all variable values here by generating DIEs as needed.
13360 (init_sections_and_labels): Add early LTO debug flag parameter
13361 and generate different sections and names if set. Add generation
13362 counter for the labels so we can have multiple of them.
13363 (reset_dies): Helper to allow DIEs to be output multiple times.
13364 (dwarf2out_finish): When outputting DIEs to the fat part of an
13365 LTO object first reset DIEs.
13366 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13367 (modified_type_die): Check for decl_ultimate_origin being self
13369 (gen_type_die_with_usage): Likewise.
13370 (gen_typedef_die): Allow decl_ultimate_origin being self.
13371 (set_decl_abstract_flags): Remove.
13372 (set_block_abstract_flags): Likewise.
13373 (dwarf2out_abstract_function): Treat the early generated DIEs
13374 as the abstract copy and only add DW_AT_inline and
13375 DW_AT_artificial here and call set_decl_origin_self.
13376 If the DIE has an abstract origin don't do anything.
13377 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13378 if we have none yet (Go fails to build one, PR78628).
13379 (variably_modified_type_p): Prevent endless recursion for Ada
13380 cyclic pointer types.
13381 * lto-streamer-in.c: Include debug.h.
13382 (dref_queue): New global.
13383 (lto_read_tree_1): Stream in DIE references.
13384 (lto_input_tree): Register DIE references.
13385 (input_function): Stream DECL_DEBUG_ARGS.
13386 * lto-streamer-out.c: Include debug.h.
13387 (lto_write_tree_1): Output DIE references.
13388 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13389 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13390 (output_function): Stream DECL_DEBUG_ARGS.
13391 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13392 Stream DECL_ABSTRACT_ORIGIN.
13393 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13394 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13395 DECL_CONTEXT for file-scope decls.
13396 * lto-streamer.h (struct dref_entry): Declare.
13397 (dref_queue): Likewise.
13398 * cfgexpand.c (pass_expand::execute): Do not call the
13399 outlining_inline_function hook here.
13400 * lto-wrapper.c (debug_obj): New global.
13401 (tool_cleanup): Unlink it if required.
13402 (debug_objcopy): New function.
13403 (run_gcc): Handle early debug sections in the IL files by
13404 extracting them to separate files, partially linkin them and
13405 feeding the result back as result to the linker.
13406 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13407 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13408 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13409 sections into a separate segment.
13410 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13412 (darwin_asm_dwarf_section): Likewise.
13413 (darwin_asm_output_dwarf_offset): Likewise.
13414 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13416 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13417 Alan Hayward <alan.hayward@arm.com>
13418 David Sherwood <david.sherwood@arm.com>
13420 * read-md.h (md_reader::record_potential_iterator_use): Replace
13421 pointer argument with an rtx and an index.
13422 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13423 (apply_mode_iterator): Likewise.
13424 (apply_code_iterator): Likewise.
13425 (apply_int_iterator): Likewise.
13426 (apply_subst_iterator): Likewise.
13427 (record_iterator_use): Likewise.
13428 (record_attribute_use): Likewise.
13429 (md_reader::record_potential_iterator_use): Likewise. Update calls
13430 to record_iterator_use and apply_iterator.
13431 (iterator_use): Replace ptr with x and index.
13432 (attribute_use): Likewise.
13433 (apply_attribute_uses): Update calls to apply_iterator.
13434 (apply_iterators): Likewise. Update initialization of iterator_use.
13435 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13436 and record_potential_iterator_use.
13437 (rtx_reader::read_rtx_operand): Likewise.
13439 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13440 Alan Hayward <alan.hayward@arm.com>
13441 David Sherwood <david.sherwood@arm.com>
13443 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13446 2017-08-21 Richard Biener <rguenther@suse.de>
13448 PR middle-end/81884
13449 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13450 at struct end conservatively when comparing common bases.
13452 2017-08-21 Richard Biener <rguenther@suse.de>
13454 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13455 (mem_ref_in_stmt): Remove.
13456 (determine_max_movement): Use ref index to get at the reference.
13457 (invariantness_dom_walker::before_dom_children): Deal with
13458 lim data already initialized.
13459 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13461 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13463 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13464 (TARGET_ISA_ROUND): Ditto.
13465 (TARGET_ROUND): Ditto.
13466 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13467 * config/i386/i386.md: Ditto.
13468 * config/i386/sse.md: Ditto.
13469 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13470 with OPTION_MASK_ISA_SSE4_1.
13472 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13475 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13476 __builtin_ia32_lzcnt_u16.
13478 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13481 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13482 (rs6000_set_current_function): Rewrite function to use it.
13484 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13487 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13488 and TYPE_WARN_IF_NOT_ALIGN.
13489 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13490 (handle_warn_if_not_align): New.
13491 (place_union_field): Call handle_warn_if_not_align.
13492 (place_field): Call handle_warn_if_not_align.
13493 Copy TYPE_WARN_IF_NOT_ALIGN.
13494 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13495 (layout_type): Likewise.
13496 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13498 (tree_decl_common): Add warn_if_not_align.
13499 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13500 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13501 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13502 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13503 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13504 * doc/extend.texi: Document warn_if_not_aligned attribute.
13505 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13507 2017-08-17 Martin Liska <mliska@suse.cz>
13510 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13511 (get_data_dependence): Use it as pointer type.
13512 (distribute_loop): Likewise.
13514 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13516 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13517 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13518 (p8_vmrgow_<mode>_direct): New define_insn.
13519 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13520 handle endianness for vmrgew and vmrgow permute patterns.
13522 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13524 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13525 * config/rs6000/rs6000-cpus.def: Remove comment.
13526 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13527 (POWERPC_MASKS): Likewise.
13528 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13529 use of TARGET_VSX_TIMODE.
13530 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13531 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13532 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13533 (rs6000_option_override_internal): Remove dead code.
13534 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13535 (rs6000_legitimize_reload_address): Likewise.
13536 (rs6000_legitimate_address_p): Likewise.
13537 (rs6000_opt_masks): Delete "vsx-timode".
13538 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13539 from function comment.
13540 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13541 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13542 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13544 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13545 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13546 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13547 (VSX_TI): Likewise.
13549 (define_peephole2): Likewise.
13551 2017-08-17 Martin Sebor <msebor@redhat.com>
13554 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13555 past the end of an array.
13556 (test_pp_format): Add test cases.
13558 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13560 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13561 missing ECF_NOTHROW flags.
13563 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13566 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13567 operands residing in integer registers.
13568 (*vsx_le_perm_load_<mode>): Likewise.
13569 (*vsx_le_perm_store_<mode>): Likewise.
13570 (define_peephole2): Add peepholes to optimize the above.
13572 2017-08-17 Marek Polacek <polacek@redhat.com>
13574 PR middle-end/81814
13575 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13576 to mimic what shorten_compare did. Change the return type to bool.
13577 (fold_cond_expr_with_comparison): Update call to
13578 operand_equal_for_comparison_p.
13579 (fold_ternary_loc): Likewise.
13581 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13583 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13585 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13587 2017-08-17 Richard Biener <rguenther@suse.de>
13589 * tree-ssa-structalias.c (solve_graph): When propagating
13590 to successors update the graphs succ edges and avoid duplicate work.
13592 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13595 * config/i386/i386.c (ix86_option_override_internal): Save target
13596 specific options after ix86_stack_protector_guard_reg was changed.
13598 2017-08-17 Richard Biener <rguenther@suse.de>
13600 PR tree-optimization/81827
13601 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13602 (new_var_info): Initialize it conservatively.
13603 (get_call_vi): Mark register vars.
13604 (new_scalar_tmp_constraint_exp): Likewise.
13605 (handle_rhs_call): Likewise.
13606 (handle_const_call): Likewise.
13607 (create_function_info_for): Likewise.
13608 (solve_constraints): Sort varinfos to separate register from
13609 non-register vars to pack points-to solution bitmaps during
13612 2017-08-17 Marek Polacek <polacek@redhat.com>
13614 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13616 2017-08-17 Richard Biener <rguenther@suse.de>
13618 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13619 to true when overflow is undefined and we saturated the result.
13621 2017-08-17 Alan Modra <amodra@gmail.com>
13624 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13625 Don't use store multiple if only one reg needs saving.
13626 (interesting_frame_related_regno): New function.
13627 (rs6000_frame_related): Don't emit frame info for regs that
13629 (rs6000_emit_epilogue): Likewise.
13631 2017-08-16 Nathan Sidwell <nathan@acm.org>
13633 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13634 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13635 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13636 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13637 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13638 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13639 (verify_type): Adjust for TYPE_BINFO move.
13640 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13641 process TYPE_BINFO directly.
13642 (hash_tree): Likewise.
13643 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13645 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13648 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13650 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13652 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13655 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13656 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13657 Add ix86_binary_operator_ok to insn constraint.
13658 (*iordi_1_bts): Ditto.
13659 (*xordi_1_btc): Ditto.
13660 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13661 Update corresponding peephole2 pattern.
13665 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13667 PR tree-optimization/81832
13668 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13669 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13671 2017-08-16 Marek Polacek <polacek@redhat.com>
13674 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13675 regarding size_zero_node.
13677 2017-08-16 Martin Liska <mliska@suse.cz>
13680 * config.gcc: Respect previously set extra_objs in case
13683 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13685 PR tree-optimization/81835
13686 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13687 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13688 not depend on the phi.
13690 2017-08-16 Alan Modra <amodra@gmail.com>
13692 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13695 2017-08-16 Alan Modra <amodra@gmail.com>
13697 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13698 (save_reg_p): ..into this. Update all callers.
13699 (first_reg_to_save): Simplify.
13701 2017-08-16 Alan Modra <amodra@gmail.com>
13703 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13706 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13710 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13711 constant count if that count is less than 32.
13713 2017-08-15 Nathan Sidwell <nathan@acm.org>
13715 * gcc.c (execute): Emit friendlier message if inferior is killed
13716 by an external cause.
13718 2017-08-15 Richard Biener <rguenther@suse.de>
13720 PR tree-optimization/81790
13721 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13722 CONSTRUCTORs from simplifying and VN.
13724 2017-08-14 Martin Sebor <msebor@redhat.com>
13726 * builtin-attrs.def: Add comments.
13728 2017-08-14 Martin Sebor <msebor@redhat.com>
13731 * doc/extend.texi (attribute nonstring): Document new attribute.
13733 2017-08-14 Martin Sebor <msebor@redhat.com>
13736 * tree-diagnostic.c (default_tree_printer): Handle %G.
13737 * gimple-pretty-print.h (percent_G_format): Declare new function.
13738 * gimple-pretty-print.c (percent_G_format): Define.
13739 * tree-pretty-print.c (percent_K_format): Add argument.
13741 2017-08-14 Martin Sebor <msebor@redhat.com>
13743 PR translation/79998
13744 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13745 Remove a stray space.
13747 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13750 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13751 (*iordi_1_bts): Ditto.
13752 (*xordi_1_btc): Ditto.
13754 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13757 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13758 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13760 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13761 quoted strings, and make more translator-friendly.
13762 (darwin_rs6000_override_options): Likewise.
13763 (rs6000_option_override_internal): Likewise.
13764 (rs6000_return_in_memory): Fix overlong line.
13765 (init_cmulative_args): Use quoted strings, and make more
13766 translator-friendly.
13767 (rs6000_pass_by_reference): Fix overlong line.
13768 (def_builtin): Use quoted strings.
13769 (altivec_expand_predicate_builtin): Use quoted strings, and make
13770 more translator-friendly.
13771 (htm_expand_builtin): Use quoted strings.
13772 (cpu_expand_builtin): Use quoted strings, and make more
13773 translator-friendly.
13774 (altivec_expand_builtin): Likewise.
13775 (paired_expand_predicate_builtin): Likewise.
13776 (rs6000_invalid_builtin): Likewise.
13777 (builtin_function_type): Use quoted strings.
13778 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13779 more translator-friendly.
13780 (rs6000_trampoline_init): Likewise.
13781 (rs6000_handle_altivec_attribute): Likewise.
13782 (rs6000_inner_target_options): Use quoted strings.
13783 (rs6000_disable_incompatible_switches): Likewise.
13784 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13785 strings, and make more translator-friendly.
13786 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13788 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13790 PR tree-optimization/81799
13791 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13792 cond_expr to simple gimple operand.
13794 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13796 PR middle-end/46932
13797 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13799 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13803 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13804 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13805 TARGET_GASISR_PROLOGUES.
13806 * config/avr/avr.c (avr_option_override): Same.
13807 (avr_pass_pre_proep::execute): Same.
13809 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13812 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13813 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13815 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13817 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13818 define_insn_and_split. Split to a memory load from 0 in
13819 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13820 using PTR mode iterator.
13821 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13822 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13823 (*add_tp_<mode>): Redefine as define_insn_and_split.
13824 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13825 address space. Merge with *add_tp_x32 using PTR mode iterator.
13826 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13827 Split to an add with a memory load from 0 in
13828 DEFAULT_TLS_SEG_REG address space.
13830 2017-08-12 Andrew Pinski <apinski@cavium.com>
13832 * config/aarch64/aarch64-option-extensions.def (rdma):
13833 Fix feature string to what Linux prints out in /proc/cpuinfo.
13835 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13838 * dwarf2out.c (modified_type_die): For C typedef types that have
13839 an ultimate origin, process the ultimate origin instead of the
13841 (gen_typedef_die): Assert that input DECLs have no ultimate
13843 (gen_type_die_with_usage): For typedef variants that have an
13844 ultimate origin, just call gen_decl_die on the original DECL.
13845 (process_scope_var): Avoid creating DIEs for local typedefs and
13846 concrete static variables.
13848 2017-08-12 Alan Modra <amodra@gmail.com>
13852 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13853 match gnu-user.h startfile.
13854 (ENDFILE_LINUX_SPEC): Similarly.
13856 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13859 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13861 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13863 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13864 * config/aarch64/aarch64.md (mov<mode>): Change.
13865 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13866 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13867 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13869 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13871 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13872 for storage order barriers.
13874 2017-08-11 Martin Liska <mliska@suse.cz>
13877 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13878 variables of void type.
13880 2017-08-11 Martin Liska <mliska@suse.cz>
13882 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13883 TARGET_SUPPORTS_ALIASES.
13884 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13885 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13886 (optimize_weakref): Likewise.
13887 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13888 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13889 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13891 2017-08-11 Martin Liska <mliska@suse.cz>
13894 * config/i386/i386.c (make_resolver_func): Do complete
13895 refactoring of the function.
13897 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13900 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13901 * config/i386/i386.c (ix86_stack_protect_guard): Use
13902 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13903 * doc/invoke.texi (x86 Options): Document
13904 -mstack-protector-guard-symbol= option.
13906 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13908 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13909 * config/i386/i386.c (ix86_split_stack_guard): New function.
13910 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13911 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13912 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13913 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13914 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13915 (split_stack_space_check): Call ix86_split_stack_guard.
13917 2017-08-10 Martin Sebor <msebor@redhat.com>
13919 * print-tree.c (print_node): Print location using the established
13921 Replace spaces with colons.
13922 (debug_raw, debug): Ditto.
13924 2017-08-10 Martin Sebor <msebor@redhat.com>
13927 * pretty-print.c (pp_format): Correct the handling of %s precision.
13929 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13932 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13934 (ix86_finalize_stack_frame_flags): This. Also clear
13935 frame_pointer_needed if -fno-omit-frame-pointer is used without
13937 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13938 with ix86_finalize_stack_frame_flags.
13939 (ix86_expand_epilogue): Likewise.
13940 (ix86_expand_split_stack_prologue): Likewise.
13941 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13943 2017-08-10 Martin Liska <mliska@suse.cz>
13946 * c-attribs.c (handle_target_attribute):
13947 Report warning for an empty string argument of target attribute.
13949 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13952 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13954 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13955 or DECL_NONLOCAL labels.
13956 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13957 or DECL_NONLOCAL labels here.
13959 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13961 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13962 to indicate when early gimple folding has been disabled.
13963 (rs6000_gimple_fold_builtin): Add debug content.
13964 (rs6000_invalid_builtin): Fix whitespace.
13965 (rs6000_expand_builtin): Fix whitespace.
13966 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13968 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13971 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13972 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13974 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13976 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13977 (qdf24xx): Likewise.
13978 * config/aarch64/aarch64.md: Include falkor.md.
13979 * config/aarch64/falkor.md: New.
13981 2017-08-09 Marek Polacek <polacek@redhat.com>
13984 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13985 * diagnostic.c (emit_diagnostic): Add a comment.
13986 (emit_diagnostic_valist): New function.
13988 2017-08-09 Marek Polacek <polacek@redhat.com>
13991 * input.c (make_location): New overload.
13992 * input.h (make_location): Declare.
13994 2017-08-08 Alan Modra <amodra@gmail.com>
13995 H.J. Lu <hongjiu.lu@intel.com>
13998 * gcc.c (NO_PIE_SPEC): Delete.
13999 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14001 (LINK_PIE_SPEC): ..to here.
14002 (LINK_COMMAND_SPEC): Support -no-pie.
14003 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14004 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14006 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14007 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14008 (ENDFILE_CRTEND_SPEC): Similarly.
14010 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14013 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14014 (mstack-protector-guard-offset=): Ditto.
14015 * config/i386/i386.c (ix86_option_override): Handle
14016 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14018 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14019 ix86_stack_protect_guard_offset variables.
14020 (TARGET_STACK_PROTECT_GUARD): Always define.
14021 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14022 and -mstack-protector-guard-offset= options.
14024 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14026 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14027 boundary case for the last candidate.
14029 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14031 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14032 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14034 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14036 PR middle-end/19706
14037 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14038 * config/aarch64/aarch64-builtins.c
14039 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14040 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14041 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14043 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14044 Andrew Pinski <pinskia@gmail.com>
14046 PR middle-end/19706
14047 * internal-fn.def (XORSIGN): New.
14048 * optabs.def (xorsign_optab): New.
14049 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14050 (convert_expand_mult_copysign): New.
14051 (pass_optimize_widening_mul::execute): Call
14052 convert_expand_mult_copysign.
14054 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14056 PR tree-optimization/81354
14057 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14058 Insert on edges rather than explicitly creating landing pads.
14059 (analyze_candidates_and_replace): Commit edge inserts.
14061 2017-08-08 Richard Biener <rguenther@suse.de>
14063 PR middle-end/81719
14064 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14065 (expand_simple_operations): Also look through ADDR_EXPRs with
14066 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14068 2017-08-08 Richard Biener <rguenther@suse.de>
14070 PR tree-optimization/81723
14071 * tree-vect-slp.c (struct bst_traits): New hash traits.
14072 (bst_fail): New global.
14073 (vect_build_slp_tree_2): New worker, split out from ...
14074 (vect_build_slp_tree): ... this now wrapping it with using
14075 bst_fail set to cache SLP tree build fails. Properly handle
14077 (vect_analyze_slp_instance): Allocate and free bst_fail.
14079 2017-08-08 Martin Liska <mliska@suse.cz>
14082 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14083 LABEL_DECLs that can be from a different function.
14085 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14087 PR tree-optimization/81744
14088 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14089 loop's number of iterations.
14091 2017-08-08 Martin Liska <mliska@suse.cz>
14093 * asan.c: Include header files.
14094 * attribs.c (build_decl_attribute_variant): New function moved
14096 (build_type_attribute_qual_variant): Likewise.
14097 (cmp_attrib_identifiers): Likewise.
14098 (simple_cst_list_equal): Likewise.
14099 (omp_declare_simd_clauses_equal): Likewise.
14100 (attribute_value_equal): Likewise.
14101 (comp_type_attributes): Likewise.
14102 (build_type_attribute_variant): Likewise.
14103 (lookup_ident_attribute): Likewise.
14104 (remove_attribute): Likewise.
14105 (merge_attributes): Likewise.
14106 (merge_type_attributes): Likewise.
14107 (merge_decl_attributes): Likewise.
14108 (merge_dllimport_decl_attributes): Likewise.
14109 (handle_dll_attribute): Likewise.
14110 (attribute_list_equal): Likewise.
14111 (attribute_list_contained): Likewise.
14112 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14113 (lookup_attribute_by_prefix): Likewise.
14114 * bb-reorder.c: Include header files.
14115 * builtins.c: Likewise.
14116 * calls.c: Likewise.
14117 * cfgexpand.c: Likewise.
14118 * cgraph.c: Likewise.
14119 * cgraphunit.c: Likewise.
14120 * convert.c: Likewise.
14121 * dwarf2out.c: Likewise.
14122 * final.c: Likewise.
14123 * fold-const.c: Likewise.
14124 * function.c: Likewise.
14125 * gimple-expr.c: Likewise.
14126 * gimple-fold.c: Likewise.
14127 * gimple-pretty-print.c: Likewise.
14128 * gimple.c: Likewise.
14129 * gimplify.c: Likewise.
14130 * hsa-common.c: Likewise.
14131 * hsa-gen.c: Likewise.
14132 * internal-fn.c: Likewise.
14133 * ipa-chkp.c: Likewise.
14134 * ipa-cp.c: Likewise.
14135 * ipa-devirt.c: Likewise.
14136 * ipa-fnsummary.c: Likewise.
14137 * ipa-inline.c: Likewise.
14138 * ipa-visibility.c: Likewise.
14140 * lto-cgraph.c: Likewise.
14141 * omp-expand.c: Likewise.
14142 * omp-general.c: Likewise.
14143 * omp-low.c: Likewise.
14144 * omp-offload.c: Likewise.
14145 * omp-simd-clone.c: Likewise.
14146 * opts-global.c: Likewise.
14147 * passes.c: Likewise.
14148 * predict.c: Likewise.
14149 * sancov.c: Likewise.
14150 * sanopt.c: Likewise.
14151 * symtab.c: Likewise.
14152 * toplev.c: Likewise.
14153 * trans-mem.c: Likewise.
14154 * tree-chkp.c: Likewise.
14155 * tree-eh.c: Likewise.
14156 * tree-into-ssa.c: Likewise.
14157 * tree-object-size.c: Likewise.
14158 * tree-parloops.c: Likewise.
14159 * tree-profile.c: Likewise.
14160 * tree-ssa-ccp.c: Likewise.
14161 * tree-ssa-live.c: Likewise.
14162 * tree-ssa-loop.c: Likewise.
14163 * tree-ssa-sccvn.c: Likewise.
14164 * tree-ssa-structalias.c: Likewise.
14165 * tree-ssa.c: Likewise.
14166 * tree-streamer-in.c: Likewise.
14167 * tree-vectorizer.c: Likewise.
14168 * tree-vrp.c: Likewise.
14169 * tsan.c: Likewise.
14170 * ubsan.c: Likewise.
14171 * varasm.c: Likewise.
14172 * varpool.c: Likewise.
14173 * tree.c: Remove functions moved to attribs.[ch].
14174 * tree.h: Likewise.
14175 * config/aarch64/aarch64.c: Add attrs.h header file.
14176 * config/alpha/alpha.c: Likewise.
14177 * config/arc/arc.c: Likewise.
14178 * config/arm/arm.c: Likewise.
14179 * config/avr/avr.c: Likewise.
14180 * config/bfin/bfin.c: Likewise.
14181 * config/c6x/c6x.c: Likewise.
14182 * config/cr16/cr16.c: Likewise.
14183 * config/cris/cris.c: Likewise.
14184 * config/darwin.c: Likewise.
14185 * config/epiphany/epiphany.c: Likewise.
14186 * config/fr30/fr30.c: Likewise.
14187 * config/frv/frv.c: Likewise.
14188 * config/ft32/ft32.c: Likewise.
14189 * config/h8300/h8300.c: Likewise.
14190 * config/i386/winnt.c: Likewise.
14191 * config/ia64/ia64.c: Likewise.
14192 * config/iq2000/iq2000.c: Likewise.
14193 * config/lm32/lm32.c: Likewise.
14194 * config/m32c/m32c.c: Likewise.
14195 * config/m32r/m32r.c: Likewise.
14196 * config/m68k/m68k.c: Likewise.
14197 * config/mcore/mcore.c: Likewise.
14198 * config/microblaze/microblaze.c: Likewise.
14199 * config/mips/mips.c: Likewise.
14200 * config/mmix/mmix.c: Likewise.
14201 * config/mn10300/mn10300.c: Likewise.
14202 * config/moxie/moxie.c: Likewise.
14203 * config/msp430/msp430.c: Likewise.
14204 * config/nds32/nds32-isr.c: Likewise.
14205 * config/nds32/nds32.c: Likewise.
14206 * config/nios2/nios2.c: Likewise.
14207 * config/nvptx/nvptx.c: Likewise.
14208 * config/pa/pa.c: Likewise.
14209 * config/pdp11/pdp11.c: Likewise.
14210 * config/powerpcspe/powerpcspe.c: Likewise.
14211 * config/riscv/riscv.c: Likewise.
14212 * config/rl78/rl78.c: Likewise.
14213 * config/rx/rx.c: Likewise.
14214 * config/s390/s390.c: Likewise.
14215 * config/sh/sh.c: Likewise.
14216 * config/sol2.c: Likewise.
14217 * config/sparc/sparc.c: Likewise.
14218 * config/spu/spu.c: Likewise.
14219 * config/stormy16/stormy16.c: Likewise.
14220 * config/tilegx/tilegx.c: Likewise.
14221 * config/tilepro/tilepro.c: Likewise.
14222 * config/v850/v850.c: Likewise.
14223 * config/vax/vax.c: Likewise.
14224 * config/visium/visium.c: Likewise.
14225 * config/xtensa/xtensa.c: Likewise.
14227 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14230 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14231 constraints since the -mupper-regs-* switches have been
14233 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14234 into a vector from a double word element that was extracted from
14235 another vector, and eliminate extra XXPERMDI instructions.
14236 (vsx_concat_<mode>_2): Likewise.
14237 (vsx_concat_<mode>_3): Likewise.
14238 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14239 concat to allow optimizing inserts from previous extracts.
14241 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14243 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14244 memory reference to a SSP offset in TLS address space.
14245 (ix86_print_operand) <case '@'>: Remove.
14246 (ix86_print_operand_punct_valid_p): Remove '@' code.
14247 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14248 UNSPEC_SP_TLS_TEST.
14249 (stack_tls_protect_set_<mode>): Remove.
14250 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14251 (stack_tls_protect_test_<mode>): Remove.
14252 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14254 2017-08-07 Olivier Hainque <hainque@adacore.com>
14257 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14259 2017-08-07 Douglas Rupp <rupp@adacore.com>
14261 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14262 variable was referenced as multidir in command.
14264 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14267 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14270 2017-08-07 Martin Liska <mliska@suse.cz>
14272 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14273 * config/rl78/rl78.c: Add include of attribs.h.
14274 * config/sh/sh.c: Likewise.
14275 * config/v850/v850.c: Likewise.
14277 2017-08-07 Tom de Vries <tom@codesourcery.com>
14279 PR middle-end/78266
14280 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14282 2017-08-07 Martin Liska <mliska@suse.cz>
14284 * config/mips/mips.c: Include attribs.h.
14286 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14289 * doc/invoke.texi: Document change in behvaior for -Ofast for
14292 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14294 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14296 (aarch64_pop_regs): Likewise.
14297 (aarch64_gen_load_pair): Likewise.
14298 (aarch64_save_callee_saves): Likewise.
14299 (aarch64_restore_callee_saves): Likewise.
14301 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14303 * config/i386/i386.c: Revert the last change.
14305 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14308 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14310 (ix86_finalize_stack_frame_flags): This. Also clear
14311 frame_pointer_needed if -fno-omit-frame-pointer is used without
14313 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14314 with ix86_finalize_stack_frame_flags.
14315 (ix86_expand_epilogue): Likewise.
14316 (ix86_expand_split_stack_prologue): Likewise.
14318 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14321 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14322 to P_AES for Westmere.
14324 2017-08-07 Jonathan Yong <10walls@gmail.com>
14326 * config/i386/mingw.opt (fset-stack-executable): Removed.
14327 * config/i386/cygming.opt (fset-stack-executable): Moved
14329 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14331 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14333 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14335 2017-08-07 Marek Polacek <polacek@redhat.com>
14337 PR middle-end/81737
14338 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14340 2017-08-07 Martin Liska <mliska@suse.cz>
14342 * attribs.h (canonicalize_attr_name): New function.
14343 (cmp_attribs): Move from c-format.c and adjusted.
14344 (is_attribute_p): Moved from tree.h.
14345 * tree-inline.c: Add new includes.
14346 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14347 (private_is_attribute_p): Remove.
14348 (private_lookup_attribute): Likewise.
14349 (private_lookup_attribute_by_prefix): Simplify.
14350 (remove_attribute): Use is_attribute_p.
14351 * tree.h: Remove removed declarations.
14353 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14355 PR middle-end/81698
14356 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14357 instead of computing it in the function. Formatting fix.
14358 (expand_case): Don't rely on default_edge being the first edge,
14359 clear it if removing it, pass default_edge to
14360 emit_case_dispatch_table.
14361 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14364 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14366 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14367 insn in the function, emit NOP after the insn.
14369 2017-08-06 Tom de Vries <tom@codesourcery.com>
14371 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14374 2017-08-06 Tom de Vries <tom@codesourcery.com>
14376 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14379 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14381 PR tree-optimization/57371
14382 * match.pd: New pattern.
14384 2017-08-04 Marek Polacek <polacek@redhat.com>
14386 PR middle-end/81695
14387 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14388 perform the computation in offset_int.
14390 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14392 PR tree-optimization/81136
14393 * tree-vectorizer.h: Include tree-hash-traits.h.
14394 (vec_base_alignments): New typedef.
14395 (vec_info): Add a base_alignments field.
14396 (vect_record_base_alignments): Declare.
14397 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14399 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14400 (create_data_ref): Add an is_conditional_in_stmt argument.
14401 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14402 the is_conditional_in_stmt field.
14403 (data_ref_loc): Add an is_conditional_in_stmt field.
14404 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14405 (find_data_references_in_stmt): Update call to create_data_ref.
14406 (graphite_find_data_references_in_stmt): Likewise.
14407 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14408 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14409 (vect_record_base_alignment): New function.
14410 (vect_record_base_alignments): Likewise.
14411 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14412 for nested statements even if we fail to compute a misalignment.
14413 Use pooled base alignments for unconditional references.
14414 (vect_find_same_alignment_drs): Compare base addresses instead
14416 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14417 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14419 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14421 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14422 Add an explicit name for the enum. Use auto_vec for slp_instances
14423 and grouped_stores.
14424 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14426 (_bb_vec_info): Add a constructor and destructor.
14427 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14428 (destroy_loop_vec_info): Delete.
14429 (vect_destroy_datarefs): Likewise.
14430 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14431 (vec_info::vec_info): New function.
14432 (vec_info::~vec_info): Likewise.
14433 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14434 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14435 destroy_loop_vec_info.
14436 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14437 (_loop_vec_info::_loop_vec_info): ...this.
14438 (destroy_loop_vec_info): Replace with...
14439 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14440 the stmt_vec_infos. Leave handling of vec_info information to its
14441 destructor. Remove explicit vector releases.
14442 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14443 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14444 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14445 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14446 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14447 (destroy_bb_vec_info): Replace with...
14448 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14449 information to its destructor.
14450 (vect_slp_analyze_bb_1): Use new and delete instead of
14451 new_bb_vec_info and destroy_bb_vec_info.
14452 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14455 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14457 * tree-data-ref.h (subscript): Add access_fn field.
14458 (data_dependence_relation): Add could_be_independent_p.
14459 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14460 (same_access_functions): Move to tree-data-ref.c.
14461 * tree-data-ref.c (ref_contains_union_access_p): New function.
14462 (access_fn_component_p): Likewise.
14463 (access_fn_components_comparable_p): Likewise.
14464 (dr_analyze_indices): Add a reference to access_fn_component_p.
14465 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14467 (constant_access_functions): Likewise.
14468 (add_other_self_distances): Likewise.
14469 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14470 (initialize_data_dependence_relation): Use XCNEW and remove
14471 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14472 of access functions that have the same type. Allow the
14473 subsequence to end with different bases in some circumstances.
14474 Record the chosen access functions in SUB_ACCESS_FN.
14475 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14476 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14477 (subscript_dependence_tester_1): Likewise dra and drb.
14478 (build_classic_dist_vector): Update calls accordingly.
14479 (subscript_dependence_tester): Likewise.
14480 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14481 DDR_COULD_BE_INDEPENDENT_P.
14482 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14483 comp_alias_ddrs instead of may_alias_ddrs.
14484 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14486 (vect_analyze_data_ref_dependence): Use it if
14487 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14488 distance vectors if that fails.
14489 (dependence_distance_ge_vf): New function.
14490 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14491 LOOP_VINFO_MAY_ALIAS_DDRS.
14493 2017-08-04 Richard Biener <rguenther@suse.de>
14495 PR middle-end/81705
14496 * fold-const.c (fold_binary_loc): Properly restrict
14497 minus_var0 && minus_var1 case when associating undefined overflow
14500 2017-08-04 Tom de Vries <tom@codesourcery.com>
14502 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14504 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14506 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14507 Don't start diagnostic messages with a capital letter.
14508 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14510 (rs6000_invalid_builtin): Likewise.
14511 (rs6000_trampoline_init): Likewise.
14513 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14516 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14517 after setting changeable df flags.
14519 2017-08-03 Richard Biener <rguenther@suse.de>
14521 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14522 up if the use is in USE - X.
14524 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14526 * toplev.c (dumpfile.h): New include.
14527 (internal_error_reentered): New static function. Use it...
14528 (internal_error_function): ...here to handle reentered internal_error.
14530 2017-08-03 Richard Biener <rguenther@suse.de>
14532 PR middle-end/81148
14533 * fold-const.c (split_tree): Add minus_var and minus_con
14534 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14535 here but always use minus_*.
14536 (associate_trees): Assert we never associate with MINUS_EXPR
14537 and NULL first operand. Do not recurse for PLUS_EXPR operands
14538 when associating as MINUS_EXPR either.
14539 (fold_binary_loc): Track minus_var and minus_con.
14541 2017-08-03 Tom de Vries <tom@codesourcery.com>
14544 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14545 ACCEL_COMPILER, apply finish_options on
14546 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14548 2017-08-03 Tom de Vries <tom@codesourcery.com>
14551 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14552 function_entry_patch_area_size > 0.
14554 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14557 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14558 instead of 10??LU, perform unit multiplication in wide_int,
14559 don't change alloc_object_size_limit if the limit is larger
14562 PR tree-optimization/81655
14563 PR tree-optimization/81588
14564 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14565 the case when ranges[i].low and high are 1 for unsigned type with
14568 PR middle-end/81052
14569 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14570 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14572 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14574 * tree-vrp.h: Add include guard.
14576 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14579 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14580 (ud2): New insn pattern.
14581 * config/i386/i386.c (ix86_expand_epilogue):
14582 For naked functions, generate ud2 instead of trap insn.
14584 2017-08-02 Marek Polacek <polacek@redhat.com>
14587 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14589 2017-08-02 Tom de Vries <tom@codesourcery.com>
14590 Cesar Philippidis <cesar@codesourcery.com>
14592 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14593 Add missing edge probabilities.
14595 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14597 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14598 Correct endianness.
14600 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14602 PR middle-end/79499
14603 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14604 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14605 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14606 prologue_seq sequences - if any.
14608 2017-08-02 Richard Biener <rguenther@suse.de>
14610 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14611 via vectors if supported, integer extracts via punning if supported
14612 or otherwise vector extracts.
14614 2017-08-02 Richard Biener <rguenther@suse.de>
14616 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14618 (bitmap_insert_into_set): ... this.
14620 2017-08-02 Richard Biener <rguenther@suse.de>
14622 PR tree-optimization/81633
14624 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14626 PR tree-optimization/71752
14627 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14629 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14631 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14632 (machine_function::call_ms2sysv_pad_out): Remove field.
14633 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14634 (ix86_compute_frame_layout): Likewise.
14636 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14639 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14640 attribute with interrupt attribute.
14642 2017-08-01 Andrew Pinski <apinski@cavium.com>
14644 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14645 BIT_INSERT_EXPR's operand 1
14646 to see if the types precision matches.
14648 2017-08-01 Martin Liska <mliska@suse.cz>
14650 PR middle-end/70140
14651 * builtins.c (expand_builtin_memcpy_args): Remove.
14652 (expand_builtin_memcpy): Call newly added function
14653 expand_builtin_memory_copy_args.
14654 (expand_builtin_memcpy_with_bounds): Likewise.
14655 (expand_builtin_mempcpy): Remove last argument.
14656 (expand_builtin_mempcpy_with_bounds): Likewise.
14657 (expand_builtin_memory_copy_args): New function created from
14658 expand_builtin_mempcpy_args with small modifications.
14659 (expand_builtin_mempcpy_args): Remove.
14660 (expand_builtin_stpcpy): Remove unused argument.
14661 (expand_builtin): Likewise.
14662 (expand_builtin_with_bounds): Likewise.
14664 2017-08-01 Martin Liska <mliska@suse.cz>
14667 Make mempcpy more optimal (PR middle-end/70140).
14669 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14672 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14673 __builtin_vec_cmpne verify both arguments are compatible vectors
14674 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14675 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14676 move computation of aligned to after checking the argument types.
14680 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14681 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14684 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14685 Doug Rupp <rupp@adacore.com>
14686 Olivier Hainque <hainque@adacore.com>
14688 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14689 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14691 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14692 for TARGET_OS_CPP_BUILTIN.
14693 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14694 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14696 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14697 passing required abi options to the assembler for EABI configurations.
14698 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14699 of .text.hot and .text.unlikely sections for kernel modules when
14700 using ARM style exceptions.
14701 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14702 options. Add EXTRA_CC1_SPEC.
14703 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14705 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14707 (ARM_TARGET2_DWARF_FORMAT): Define.
14708 * config/arm/t-vxworks: Adjust multilib control to removal of the
14709 Diab command line options.
14711 2017-08-01 Martin Liska <mliska@suse.cz>
14713 PR gcov-profile/81561
14714 * gcov.c (unblock): Make unblocking safe as we need to preserve
14715 index correspondence of blocks and block_lists.
14717 2017-08-01 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/81181
14720 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14721 (compute_antic): ... end of iteration here.
14723 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14725 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14726 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14727 (ftree-slp-vectorize): Likewise.
14728 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14729 can no longer be set independent of flag_tree_loop_vectorize.
14730 * omp-general.c (emp_max_vf): Likewise.
14731 * opts.c (enable_fdo_optimizations): Remove references to
14732 flag_tree_vectorize, these are now implicit.
14733 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14734 and leave it for the options machinery.
14736 2017-08-01 Martin Liska <mliska@suse.cz>
14738 PR middle-end/70140
14739 * builtins.c (expand_builtin_memcpy_args): Remove.
14740 (expand_builtin_memcpy): Call newly added function
14741 expand_builtin_memory_copy_args.
14742 (expand_builtin_memcpy_with_bounds): Likewise.
14743 (expand_builtin_mempcpy): Remove last argument.
14744 (expand_builtin_mempcpy_with_bounds): Likewise.
14745 (expand_builtin_memory_copy_args): New function created from
14746 expand_builtin_mempcpy_args with small modifications.
14747 (expand_builtin_mempcpy_args): Remove.
14748 (expand_builtin_stpcpy): Remove unused argument.
14749 (expand_builtin): Likewise.
14750 (expand_builtin_with_bounds): Likewise.
14752 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14755 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14756 print "ds:" only for immediates in generic address space.
14758 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14761 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14762 (ix86_function_ok_for_sibcall): Return false for naked functions.
14764 2017-08-01 Richard Biener <rguenther@suse.de>
14766 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14767 (compute_antic): Seed worklist with exit block predecessors.
14768 * cfganal.c (dfs_find_deadend): For a cycle return the source
14769 of the edge closing it.
14771 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14773 * config/aarch64/aarch64.c
14774 (aarch64_can_const_movi_rtx_p): Move 0 check.
14776 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14778 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14779 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14781 * match.pd: Ditto in address comparison pattern.
14783 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14785 PR tree-optimization/81627
14786 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14787 closed ssa form for store-store chain.
14789 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14791 PR tree-optimization/81620
14792 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14793 for store-store chain.
14795 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14797 PR tree-optimization/81588
14798 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14799 ranges[i].in_p, invert comparison code ccode. For >/>=,
14800 swap rhs1 and rhs2 and comparison code unconditionally,
14801 for </<= don't do that. Don't swap rhs1/rhs2 again if
14802 ranges[i].in_p, instead invert comparison code ccode if
14803 opcode or oe->rank is BIT_IOR_EXPR.
14806 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14807 a direct optab to conversion optab.
14808 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14809 with GET_MODE_INNER as last argument instead of optab_handler.
14810 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14811 vector extraction if possible and optab is available.
14812 * expr.c (store_constructor): Use convert_optab_handler instead
14813 of optab_handler. Use vector initialization from smaller
14814 vectors if possible and optab is available.
14815 * tree-vect-stmts.c (vectorizable_load): Likewise.
14816 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14817 now have two modes.
14818 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14819 of vec_init from half-sized vectors with the same element mode.
14820 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14821 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14822 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14823 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14824 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14825 after mode in gen_vec_extract* calls.
14826 (vec_extract<mode>): Renamed to ...
14827 (vec_extract<mode><ssescalarmodelower>): ... this.
14828 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14829 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14830 element mode after mode in gen_vec_init* calls.
14831 (VEC_INIT_HALF_MODE): New mode iterator.
14832 (vec_init<mode>): Renamed to ...
14833 (vec_init<mode><ssescalarmodelower>): ... this.
14834 (vec_init<mode><ssehalfvecmodelower>): New expander.
14835 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14836 (vec_extractv2sfsf): ... this.
14837 (vec_initv2sf): Renamed to ...
14838 (vec_initv2sfsf): ... this.
14839 (vec_extractv2si): Renamed to ...
14840 (vec_extractv2sisi): ... this.
14841 (vec_initv2si): Renamed to ...
14842 (vec_initv2sisi): ... this.
14843 (vec_extractv4hi): Renamed to ...
14844 (vec_extractv4hihi): ... this.
14845 (vec_initv4hi): Renamed to ...
14846 (vec_initv4hihi): ... this.
14847 (vec_extractv8qi): Renamed to ...
14848 (vec_extractv8qiqi): ... this.
14849 (vec_initv8qi): Renamed to ...
14850 (vec_initv8qiqi): ... this.
14851 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14852 (vec_init<mode>): Renamed to ...
14853 (vec_init<mode><VEC_base_l>): ... this.
14854 (vec_extract<mode>): Renamed to ...
14855 (vec_extract<mode><VEC_base_l>): ... this.
14856 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14857 (vec_initv2sfsf): ... this.
14858 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14859 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14860 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14861 element mode after mode in gen_vec_init* calls.
14862 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14863 (vec_init<mode><Vel>): ... this.
14864 (vec_extract<mode>): Renamed to ...
14865 (vec_extract<mode><Vel>): ... this.
14866 * config/aarch64/iterators.md (Vel): New mode attribute.
14867 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14868 Add element mode after mode in gen_vec_extract* calls.
14869 * config/s390/vector.md (non_vec_l): New mode attribute.
14870 (vec_extract<mode>): Renamed to ...
14871 (vec_extract<mode><non_vec_l>): ... this.
14872 (vec_init<mode>): Renamed to ...
14873 (vec_init<mode><non_vec_l>): ... this.
14874 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14875 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14877 * config/arm/iterators.md (V_elem_l): New mode attribute.
14878 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14879 (vec_extract<mode><V_elem_l>): ... this.
14880 (vec_extractv2di): Renamed to ...
14881 (vec_extractv2didi): ... this.
14882 (vec_init<mode>): Renamed to ...
14883 (vec_init<mode><V_elem_l>): ... this.
14884 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14885 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14886 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14887 Add element mode after gen_vec_extract* calls.
14888 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14889 (vec_init<mode><unitmode>): ... this.
14890 (vec_extract<mode>): Renamed to ...
14891 (vec_extract<mode><unitmode>): ... this.
14892 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14893 (vec_init<mode><unitmode>): ... this.
14894 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14895 (vec_initv2sfsf): ... this.
14896 (vec_extractv2sf): Renamed to ...
14897 (vec_extractv2sfsf): ... this.
14898 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14899 Add element mode after gen_vec_extract* calls.
14900 * config/mips/mips.md (unitmode): New mode iterator.
14901 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14902 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14903 * config/spu/spu.md (inner_l): New mode attribute.
14904 (vec_init<mode>): Renamed to ...
14905 (vec_init<mode><inner_l>): ... this.
14906 (vec_extract<mode>): Renamed to ...
14907 (vec_extract<mode><inner_l>): ... this.
14908 * config/sparc/sparc.md (veltmode): New mode iterator.
14909 (vec_init<VMALL:mode>): Renamed to ...
14910 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14911 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14912 (vec_initv2sisi): ... this.
14913 (vec_initv2sf): Renamed to ...
14914 (vec_initv2sfsf): ... this.
14915 (vec_extractv2sf): Renamed to ...
14916 (vec_extractv2sfsf): ... this.
14917 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14918 (vec_init<mode>): Renamed to ...
14919 (vec_init<mode><VEC_base_l>): ... this.
14920 (vec_extract<mode>): Renamed to ...
14921 (vec_extract<mode><VEC_base_l>): ... this.
14922 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14923 (vec_initv2sfsf): ... this.
14924 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14925 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14926 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14927 gen_vec_init* calls.
14929 2017-08-01 Richard Biener <rguenther@suse.de>
14931 PR tree-optimization/81297
14932 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14933 TREE_OVERFLOW from INTEGER_CSTs.
14935 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14937 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14939 2017-07-31 Carl Love <cel@us.ibm.com>
14941 * config/rs6000/rs6000-c: Add support for built-in functions
14942 vector signed char vec_xl_be (signed long long, signed char *);
14943 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14944 vector signed int vec_xl_be (signed long long, signed int *);
14945 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14946 vector signed long long vec_xl_be (signed long long, signed long long *);
14947 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14948 vector signed short vec_xl_be (signed long long, signed short *);
14949 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14950 vector double vec_xl_be (signed long long, double *);
14951 vector float vec_xl_be (signed long long, float *);
14952 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14953 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14954 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14956 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14957 (altivec_expand_builtin): Add switch statement to call
14958 altivec_expand_xl_be for each builtin.
14959 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14960 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14961 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14962 __builtin_vsx_le_be_v16qi.
14963 * doc/extend.texi: Update the built-in documentation file for the
14964 new built-in functions.
14966 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14969 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14971 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14973 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14975 * config.gcc: Add z14.
14976 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14977 CPU model numbers for z13s and z14.
14978 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14980 * config/s390/s390-opts.h (enum processor_type): Rename
14981 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14982 * config/s390/s390.c (processor_table): Add field for CPU name to
14983 be passed to Binutils.
14984 (s390_asm_output_machine_for_arch): Use the new field in
14985 processor_table for Binutils.
14986 (s390_expand_builtin): Replace arch12 with z14.
14987 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14988 (s390_get_sched_attrmask): Likewise.
14989 (s390_get_unit_mask): Likewise.
14990 * config/s390/s390.opt: Add z14 to processor_type enum.
14992 2017-07-31 Martin Jambor <mjambor@suse.cz>
14995 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14996 regardless of optimization level.
14998 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14999 Martin Liska <mliska@suse.cz>
15001 * predict.def: Remove old comment and adjust probability.
15002 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15003 PREDICT statements.
15005 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15008 * config/i386/i386.c (ix86_function_naked): New function.
15009 (ix86_can_use_return_insn_p): Return false for naked functions.
15010 (ix86_expand_prologue): Skip prologue for naked functions.
15011 (ix86_expand_epilogue): Skip epilogue for naked functions
15012 and emit trap instruction.
15013 (ix86_warn_func_return): New function.
15014 (ix86_attribute_table): Add "naked" attribute specification.
15015 (TARGET_WARN_FUNC_RETURN): Define.
15016 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15018 2017-07-31 Martin Liska <mliska@suse.cz>
15020 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15021 (dump_gimple_bb_header): Always dump BB info.
15022 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15024 2017-07-31 Martin Liska <mliska@suse.cz>
15027 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15028 also with current_function_decl non-null equality.
15030 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15033 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15034 change type to the element type, instead add eltype variable and
15035 use it where we are interested in the element type.
15037 PR tree-optimization/81603
15038 * ipa-polymorphic-call.c
15039 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15040 offset arithmetic in offset_int, bail out if the resulting bit offset
15041 doesn't fit into shwi.
15043 2017-07-31 Martin Liska <mliska@suse.cz>
15045 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15046 (gimplify_save_expr): Fix comment.
15048 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15051 * config/i386/i386.c (ix86_function_arg): Update arguments for
15053 (ix86_compute_frame_layout): Set the initial stack offset to
15054 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15055 INCOMING_FRAME_SP_OFFSET.
15056 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15057 stack before exception handler returns.
15058 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15059 the 'ERROR_CODE' for exception handler.
15061 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15063 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15064 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15065 (ASM_OUTPUT_REG_POP): Ditto.
15066 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15067 instead of asm_fprintf to output pure string.
15069 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15071 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15072 to imported_module_or_decl hook.
15073 (debug_nothing_tree_tree_tree_bool): Remove.
15074 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15075 * debug.c (do_nothing_debug_hooks): Use
15076 debug_nothing_tree_tree_tree_bool_bool instead of
15077 debug_nothing_tree_tree_tree_bool.
15078 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15079 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15080 * sdbout.c (sdb_debug_hooks): Likewise.
15081 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15082 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15084 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15085 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15086 attribute, don't add anything.
15088 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15090 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15091 (fold_build2_stat_loc): Likewise.
15092 (fold_build3_stat_loc): Likewise.
15093 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15094 (fold_build1_loc): Remove macro.
15095 (fold_build2_loc): Likewise.
15096 (fold_build3_loc): Likewise.
15098 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15100 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15101 (gimple_build_debug_bind_source_stat): Likewise.
15102 * gimple.h (gimple_build_debug_bind): Remove macro.
15103 (gimple_build_debug_bind_source): Likewise.
15105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15107 * bitmap.c (bitmap_alloc): Adjust.
15108 (bitmap_gc_alloc): Likewise.
15109 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15111 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15113 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15114 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15115 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15116 (bitmap_gc_alloc_stat): Likewise.
15117 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15121 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15122 * rtl.h (shallow_copy_rtx): Remove macro.
15124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15126 * emit-rtl.c (gen_raw_REG): Adjust.
15127 * gengenrtl.c (gendef): Likewise.
15128 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15129 * rtl.h (rtx_alloc): Remove macro.
15131 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15133 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15134 (build_tree_list_stat): Likewise.
15135 * tree.h (build_tree_list): Remove macro.
15136 (build_tree_list_vec): Likewise.
15138 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15140 * tree.c (make_vector_stat): Remove _stat from name.
15141 (build_vector_stat): Likewise.
15142 * tree.h (make_vector_stat): Remove macro.
15143 (build_vector_stat): Likewise.
15145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15147 * tree.h (build_var_debug_value): Remove prototype.
15149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15151 * tree.c (tree_cons_stat): Remove _stat from name.
15152 * tree.h (tree_cons): Remove macro.
15154 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15156 * tree.c (build_vl_exp_stat): Remove _stat from name.
15157 * tree.h (build_vl_exp): Remove macro.
15159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15161 * tree.c (build_decl_stat): Remove _stat from name.
15162 * tree.h (build_decl): Remove macro.
15164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15166 * gimple.c (gimple_build_with_ops_stat): Adjust.
15167 (gimple_alloc_stat): Remove _stat from name.
15168 * gimple.h (gimple_alloc): Remove macro.
15170 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15172 * tree.c (make_tree_vec_stat): Remove _stat from name.
15173 (grow_tree_vec_stat): Likewise.
15174 * tree.h (make_tree_vec_stat): Adjust prototype.
15175 (grow_tree_vec_stat): Likewise.
15176 (make_tree_vec): Remove macro.
15177 (grow_tree_vec): Likewise.
15179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15181 * fold-const.c (fold_build1_stat_loc): Adjust.
15182 (fold_build2_stat_loc): Likewise.
15183 (fold_build3_stat_loc): Likewise.
15184 * tree.c (build0_stat): Remove _stat from name.
15185 (build1_stat): Likewise.
15186 (build2_stat): Likewise.
15187 (build3_stat): Likewise.
15188 (build4_stat): Likewise.
15189 (build5_stat): Likewise.
15190 * tree.h (build1_loc): Remove macro, and rename _stat function
15192 (build2_loc): Likewise.
15193 (build3_loc): Likewise.
15194 (build4_loc): Likewise.
15195 (build5_loc): Likewise.
15197 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15199 * tree.c (make_int_cst_stat): Remove _stat from name.
15200 * tree.h (make_int_cst_stat): Adjust prototype.
15201 (make_int_cst): Remove macro.
15203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15205 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15206 * tree.h (make_tree_binfo_stat): Adjust prototype.
15207 (make_tree_binfo): Remove.
15209 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15211 * tree.c (copy_node_stat): Rename to copy_node.
15212 (build_distinct_type_copy): Adjust.
15213 * tree.h (copy_node_stat): Adjust prototype.
15214 (copy_node): Remove macro.
15216 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15218 * tree.c (make_node_stat): rename to make_node.
15219 (build_tree_list_stat): Adjust.
15220 (build0_stat): Likewise.
15221 (build2_stat): Likewise.
15222 (build3_stat): Likewise.
15223 (build4_stat): Likewise.
15224 (build5_stat): Likewise.
15225 (build_decl_stat): Likewise.
15226 * tree.h (make_node_stat): Adjust prototype.
15227 (make_node): remove macro.
15229 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15231 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15232 (PPC_FEATURE2_SCV): Likewise.
15233 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15235 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15237 * config/aarch64/aarch64.c
15238 (aarch64_internal_mov_immediate): Add new special pattern.
15239 * config/aarch64/aarch64.md (*movdi_aarch64):
15240 Add reg/32bit const mov case.
15242 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15243 Richard Sandiford <richard.sandiford@linaro.org>
15245 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15246 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15247 Add integer and movi cases.
15248 (movi-split-hf-df-sf split, fp16): New.
15249 (enabled): Added TARGET_FP_F16INST.
15250 * config/aarch64/iterators.md (GPF_HF): New.
15251 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15253 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15255 * config/aarch64/aarch64.c
15256 (aarch64_simd_container_mode): Add prototype.
15257 (aarch64_expand_mov_immediate): Add HI support.
15258 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15259 (aarch64_can_const_movi_rtx_p): New.
15260 (aarch64_preferred_reload_class):
15261 Remove restrictions of using FP registers for certain SIMD operations.
15262 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15263 (aarch64_valid_floating_const): Add integer move validation.
15264 (aarch64_simd_imm_scalar_p): Remove.
15265 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15266 (aarch64_legitimate_constant_p): Expand list of supported cases.
15267 * config/aarch64/aarch64-protos.h
15268 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15269 (aarch64_reinterpret_float_as_int): New.
15270 (aarch64_simd_imm_scalar_p): Remove.
15271 * config/aarch64/constraints.md (Uvi): New.
15272 (Dd): Split into Ds and new Dd.
15273 * config/aarch64/aarch64.md (*movsi_aarch64):
15275 (*movdi_aarch64): Add SIMD mov case.
15277 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15279 * tree-predcom.c: (struct chain): Handle store-store chain in which
15280 stores for elimination only store loop invariant values.
15281 (execute_pred_commoning_chain): Ditto.
15282 (prepare_initializers_chain_store_elim): Ditto.
15283 (prepare_finalizers): Ditto.
15284 (is_inv_store_elimination_chain): New function.
15285 (initialize_root_vars_store_elim_1): New function.
15287 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15289 * tree-predcom.c: Revise general description of the pass.
15290 (enum chain_type): New enum type for store elimination.
15291 (struct chain): New field supporting store elimination.
15292 (struct component): Ditto.
15293 (dump_chain): Dump store-stores chain.
15294 (release_chain): Release resources.
15295 (split_data_refs_to_components): Compute and create component
15296 contains only stores for elimination.
15297 (get_chain_last_ref_at): New function.
15298 (make_invariant_chain): Initialization.
15299 (make_rooted_chain): Specify chain type in parameter and record it.
15300 (add_looparound_copies): Skip for store-stores chain.
15301 (determine_roots_comp): Compute type of chain and pass it to
15303 (initialize_root_vars_store_elim_2): New function.
15304 (finalize_eliminated_stores): New function.
15305 (remove_stmt): Handle store for elimination.
15306 (execute_pred_commoning_chain): Execute predictive commoning on
15307 store-store chains.
15308 (determine_unroll_factor): Skip unroll for store-stores chain.
15309 (prepare_initializers_chain_store_elim): New function.
15310 (prepare_initializers_chain): Hanlde store-store chain.
15311 (prepare_finalizers_chain, prepare_finalizers): New function.
15312 (tree_predictive_commoning_loop): Return integer value indicating
15313 if loop is unrolled or lcssa form is corrupted.
15314 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15316 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15318 * tree-predcom.c (initialize_root): Delete.
15319 (execute_pred_commoning_chain): Initialize root vars and replace
15320 reference of non-combined chain directly, rather than call above
15323 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15325 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15326 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15328 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15330 * tree-predcom.c (struct chain): New field init_seq.
15331 (release_chain): Release init_seq.
15332 (prepare_initializers_chain): Record intialization stmts in above
15334 (insert_init_seqs): New function.
15335 (tree_predictive_commoning_loop): Call insert_init_seqs.
15337 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15339 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15341 2017-07-28 Richard Biener <rguenther@suse.de>
15343 * match.pd: Remove superfluous :c.
15344 * genmatch.c (simplify::id): Add member.
15345 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15347 (current_id): New global.
15348 (dt_node::parent): Move from ...
15349 (dt_operand::parent): ... here. Add for_id member.
15350 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15351 (decision_tree::find_node): Relax order requirement when
15352 merging DT_TRUE nodes to ones inbetween the current simplify
15353 and the one we try to merge with. Add diagnostic whenever
15354 we need to enforce pattern order by not merging.
15355 (decision_tree::insert): Set current_id.
15356 (decision_tree::print_node): Dump parent node and for_id.
15357 (parser::last_id): Add member.
15358 (parser::push_simplify): Assign unique id.
15359 (parser::parser): Initialize last_id.
15361 2017-07-28 Martin Liska <mliska@suse.cz>
15364 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15365 gimple_build_debug_bind.
15367 2017-07-28 Richard Biener <rguenther@suse.de>
15369 PR tree-optimization/81502
15370 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15372 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15374 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15375 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15376 for BIT_FIELD_REF args.
15377 * fold-const.c (make_bit_field_ref): Likewise.
15378 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15380 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15383 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15384 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15385 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15386 Or it into SANITIZER_UNDEFINED.
15387 * ubsan.c: Include gimple-fold.h and varasm.h.
15388 (ubsan_expand_ptr_ifn): New function.
15389 (instrument_pointer_overflow): New function.
15390 (maybe_instrument_pointer_overflow): New function.
15391 (instrument_object_size): Formatting fix.
15392 (pass_ubsan::execute): Call instrument_pointer_overflow
15393 and maybe_instrument_pointer_overflow.
15394 * internal-fn.c (expand_UBSAN_PTR): New function.
15395 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15396 * sanitizer.def (__ubsan_handle_pointer_overflow,
15397 __ubsan_handle_pointer_overflow_abort): New builtins.
15398 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15399 * internal-fn.def (UBSAN_PTR): New internal function.
15400 * opts.c (sanitizer_opts): Add pointer-overflow.
15401 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15402 * fold-const.c (build_range_check): Compute pointer range check in
15403 integral type if pointer arithmetics would be needed. Formatting
15406 2017-07-28 Martin Liska <mliska@suse.cz>
15409 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15410 parameters that are of a variable-length.
15412 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15414 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15416 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15417 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15418 (CRT_CALL_STATIC_FUNCTION): Likewise.
15419 (ASM_DEFAULT_SPEC): New define.
15420 (ASM_SPEC32): Likewise.
15421 (ASM_SPEC64): Likewise.
15422 (ASM_SPEC_COMMON): Likewise.
15423 (ASM_SPEC): Likewise.
15424 (INVALID_64BIT): Likewise.
15425 (LINK_OS_DEFAULT_SPEC): Likewise.
15426 (LINK_OS_SPEC32): Likewise.
15427 (LINK_OS_SPEC64): Likewise.
15428 (POWERPC_LINUX): Likewise.
15429 (PTRDIFF_TYPE): Likewise.
15430 (RESTORE_FP_PREFIX): Likewise.
15431 (RESTORE_FP_SUFFIX): Likewise.
15432 (SAVE_FP_PREFIX): Likewise.
15433 (SAVE_FP_SUFFIX): Likewise.
15434 (SIZE_TYPE): Likewise.
15435 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15436 (TARGET_64BIT): Likewise.
15437 (TARGET_64BIT): Likewise.
15438 (TARGET_AIX): Likewise.
15439 (WCHAR_TYPE_SIZE): Likewise.
15440 (WCHAR_TYPE): Undefine.
15441 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15442 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15443 (CPP_OS_RTEMS_SPEC): Delete.
15444 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15445 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15447 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15449 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15451 PR tree-optimization/81578
15452 * tree-parloops.c (build_new_reduction): Bail out if
15453 reduction_code isn't one of the standard OpenMP reductions.
15454 Move the details printing after that decision.
15456 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15458 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15459 related to reload_in_progress.
15460 (splat_input_operand): Likewise.
15461 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15463 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15465 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15466 (TARGET_INSTANTIATE_DECLS): Likewise.
15467 (legitimate_indexed_address_p): Delete reload_in_progress code.
15468 (rs6000_debug_legitimate_address_p): Likewise.
15469 (rs6000_eliminate_indexed_memrefs): Likewise.
15470 (rs6000_emit_le_vsx_store): Likewise.
15471 (rs6000_emit_move_si_sf_subreg): Likewise.
15472 (rs6000_emit_move): Likewise.
15473 (register_to_reg_type): Likewise.
15474 (rs6000_pre_atomic_barrier): Likewise.
15475 (rs6000_machopic_legitimize_pic_address): Likewise.
15476 (rs6000_allocate_stack_temp): Likewise.
15477 (rs6000_address_for_fpconvert): Likewise.
15478 (rs6000_address_for_altivec): Likewise.
15479 (rs6000_secondary_memory_needed_rtx): Delete function.
15480 (rs6000_check_sdmode): Likewise.
15481 (rs6000_alloc_sdmode_stack_slot): Likewise.
15482 (rs6000_instantiate_decls): Likewise.
15483 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15484 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15485 Delete reload_in_progress.
15486 (*vec_reload_and_plus_<mptrsize>): Likewise.
15487 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15488 (vsx_div_v2di): Likewise.
15489 (vsx_udiv_v2di): Likewise.
15491 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15493 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15494 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15495 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15496 (rs6000_debug_reg_global): Delete print of LRA status.
15497 (rs6000_option_override_internal): Delete dead LRA related code.
15498 (rs6000_lra_p): Delete function.
15499 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15501 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15503 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15504 * config/riscv/rtems.h: New file.
15506 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15507 Sudakshina Das <sudi.das@arm.com>
15509 * config/aarch64/aarch64.md
15510 (define_split for and<mode>3nr_compare): Move
15511 non aarch64_logical_operand to a register.
15512 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15513 register immediate operand to a register.
15514 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15516 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15518 PR middle-end/81564
15519 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15521 2017-07-27 Richard Biener <rguenther@suse.de>
15523 PR tree-optimization/81573
15524 PR tree-optimization/81494
15525 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15526 multi defuse cycle case.
15528 2017-07-27 Richard Biener <rguenther@suse.de>
15530 PR tree-optimization/81571
15531 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15534 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15536 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15537 earlier and only if MASK_FPU is set. Adjust formatting.
15539 2017-07-27 Martin Liska <mliska@suse.cz>
15541 * opt-functions.awk: Add validation of value of Init.
15542 * optc-gen.awk: Pass new argument.
15544 2017-07-27 Martin Liska <mliska@suse.cz>
15546 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15547 Fix wrong condition.
15549 2017-07-27 Martin Liska <mliska@suse.cz>
15551 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15552 BBs and edges seen by autoFDO.
15554 2017-07-27 Richard Biener <rguenther@suse.de>
15556 PR tree-optimization/81502
15557 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15558 with incompatible but same sized type.
15559 (execute_update_addresses_taken): Likewise.
15561 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15563 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15564 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15566 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15569 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15570 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15571 Change s_operand to memory_operand.
15573 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15575 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15576 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15577 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15578 Emit instructions rather than returning an expression. Handle TFmode
15579 and KFmode by casting to TImode.
15580 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15581 (rs6000_emit_le_vsx_store): Likewise.
15582 * config/rs6000/vsx.md (VSX_TI): New iterator.
15583 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15584 (*vsx_le_undo_permute_<mode>): Likewise.
15585 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15586 emit the split sequence.
15587 (*vsx_le_perm_store_<mode>): Likewise.
15589 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15591 PR tree-optimization/81555
15592 PR tree-optimization/81556
15593 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15594 if true, force CHANGED for the recursive invocation.
15595 (reassociate_bb): Remember original length of ops array, pass
15596 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15598 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15599 attributes for noipa attribute. For naked attribute use
15600 lookup_attribute first before lookup_attribute_spec.
15601 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15603 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15604 for functions with noipa attribute.
15605 (cgraph_externally_visible_p): Return true for functions with noipa
15607 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15608 for functions with noipa attribute.
15609 * doc/extend.texi: Document noipa function attribute.
15610 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15611 also for functions with noipa attribute.
15612 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15614 2017-07-26 Andrew Pinski <apinski@cavium.com>
15616 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15617 vec_unalign_load_cost and vec_unalign_store_cost.
15619 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15621 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15622 -mvsx-small-integer option.
15623 (ISA_3_0_MASKS_IEEE): Likewise.
15624 (OTHER_VSX_VECTOR_MASKS): Likewise.
15625 (POWERPC_MASKS): Likewise.
15626 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15627 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15628 code, only testing for DImode being allowed in non-VSX floating
15630 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15631 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15633 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15634 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15635 TARGET_P8_VECTOR test.
15636 (rs6000_secondary_reload_simple_move): Likewise.
15637 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15638 since TARGET_P9_VECTOR was already tested.
15639 (rs6000_opt_masks): Remove -mvsx-small-integer.
15640 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15641 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15643 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15644 test for TARGET_VEXTRACTUB was used, and that uses
15646 (p9 extract splitter): Likewise.
15647 (vsx_extract_<mode>_di_p9): Likewise.
15648 (vsx_extract_<mode>_store_p9): Likewise.
15649 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15650 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15651 the elimination of TARGET_VSX_SMALL_INTEGER.
15652 (vsx_extract_<mode>_p8): Likewise.
15653 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15654 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15655 (vsx_set_<mode>_p9): Likewise.
15656 (vsx_set_v4sf_p9): Likewise.
15657 (vsx_set_v4sf_p9_zero): Likewise.
15658 (vsx_insert_extract_v4sf_p9): Likewise.
15659 (vsx_insert_extract_v4sf_p9_2): Likewise.
15660 * config/rs6000/rs6000.md (sign extend splitter): Change
15661 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15662 (floatsi<mode>2_lfiwax_mem): Likewise.
15663 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15664 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15665 since a test for TARGET_P9_VECTOR was used.
15666 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15667 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15668 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15669 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15670 TARGET_P8_VECTOR test.
15671 (fix_trunc<mode>si2_stfiwx): Likewise.
15672 (fix_trunc<mode>si2_internal): Likewise.
15673 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15674 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15676 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15677 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15678 TARGET_P8_VECTOR test.
15679 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15680 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15681 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15683 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15684 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15685 since a test for TARGET_P9_VECTOR was used.
15686 (splitter for loading small constants): Likewise.
15688 2017-07-26 Andrew Pinski <apinski@cavium.com>
15690 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15693 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15696 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15697 (fp_valid_at): Likewise.
15699 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15701 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15702 (qdf24xx_addrcost_table): Likewise.
15703 (cortexa57_tunings): Update to use generic_branch_cost.
15704 (cortexa72_tunings): Likewise.
15705 (cortexa73_tunings): Likewise.
15706 (qdf24xx_tunings): Likewise.
15708 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15710 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15711 (thunderx2t99_branch_cost): Likewise.
15712 (cortexa35_tunings): Update to use generic_branch_cost.
15713 (cortexa53_tunings): Likewise.
15714 (cortexa57_tunings): Likewise.
15715 (cortexa72_tunings): Likewise.
15716 (cortexa73_tunings): Likewise.
15717 (thunderx2t99_tunings): Likewise.
15719 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15721 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15722 (sparc_option_override): Honour MASK_FSMULD.
15723 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15724 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15725 * config/sparc/sparc.opt (mfsmuld): New option.
15726 * doc/invoke.texi (mfsmuld): Document option.
15728 2017-07-26 Marek Polacek <polacek@redhat.com>
15730 PR middle-end/70992
15731 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15733 2017-07-26 Richard Biener <rguenther@suse.de>
15735 * gimple-match-head.c (do_valueize): Return OP if valueize
15737 (get_def): New helper to get at the def stmt of a SSA name
15738 if valueize allows.
15739 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15740 do_valueize to get at the def stmt.
15741 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15743 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15745 PR middle-end/46932
15746 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15748 2017-07-26 Martin Liska <mliska@suse.cz>
15751 * function.c (expand_function_start): Make expansion of
15752 nonlocal_goto_save_area after parm_birth_insn.
15754 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15756 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15757 from all CPU target flags enable members.
15759 2017-07-26 Richard Biener <rguenther@suse.de>
15761 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15762 (decision_tree::gen): Make 'type' const.
15763 (write_predicate): Likewise.
15765 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15767 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15768 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15769 (rs6000_option_override_internal): Likewise.
15770 (rs6000_expand_vector_set): Likewise.
15771 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15772 (TARGET_UPPER_REGS_SF): Likewise.
15773 (TARGET_UPPER_REGS_DI): Likewise.
15774 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15775 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15776 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15777 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15778 (Splitters for DI constants in Altivec registers): Likewise.
15779 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15780 (vsx_set_v4sf_p9): Likewise.
15781 (vsx_set_v4sf_p9_zero): Likewise.
15782 (vsx_insert_extract_v4sf_p9): Likewise.
15783 (vsx_insert_extract_v4sf_p9_2): Likewise.
15785 2017-07-25 Carl Love <cel@us.ibm.com>
15787 * doc/extend.texi: Update the built-in documentation file for the
15788 existing built-in functions
15789 vector signed char vec_cnttz (vector signed char);
15790 vector unsigned char vec_cnttz (vector unsigned char);
15791 vector signed short vec_cnttz (vector signed short);
15792 vector unsigned short vec_cnttz (vector unsigned short);
15793 vector signed int vec_cnttz (vector signed int);
15794 vector unsigned int vec_cnttz (vector unsigned int);
15795 vector signed long long vec_cnttz (vector signed long long);
15796 vector unsigned long long vec_cnttz (vector unsigned long long);
15798 2017-07-25 Andrew Pinski <apinski@cavium.com>
15800 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15801 accesses where the use is for the first operand of a BIT_INSERT.
15803 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15806 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15807 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15809 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15811 * config/i386/gstabs.h: Delete.
15812 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15814 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15816 * config/i386/i386.c (ix86_decompose_address): Do not check for
15817 register RTX when looking at index_reg or base_reg.
15818 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15820 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15822 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15823 to update EH info here.
15825 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15827 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15829 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15831 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15833 2017-07-25 Torsten Duwe <duwe@suse.de>
15835 * common.opt: Introduce -fpatchable-function-entry
15836 command line option, and its variables function_entry_patch_area_size
15837 and function_entry_patch_area_start.
15838 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15839 including a two-value parser.
15840 * target.def (print_patchable_function_entry): New target hook.
15841 * targhooks.h (default_print_patchable_function_entry): New function.
15842 * targhooks.c (default_print_patchable_function_entry): Likewise.
15843 * toplev.c (process_options): Switch off IPA-RA if
15844 patchable function entries are being generated.
15845 * varasm.c (assemble_start_function): Look at the
15846 patchable-function-entry command line switch and current
15847 function attributes and maybe generate NOP instructions by
15848 calling the print_patchable_function_entry hook.
15849 * doc/extend.texi: Document patchable_function_entry attribute.
15850 * doc/invoke.texi: Document -fpatchable_function_entry
15851 command line option.
15852 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15854 * doc/tm.texi: Re-generate.
15856 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15859 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15860 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15862 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15864 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15866 2017-07-25 Richard Biener <rguenther@suse.de>
15868 PR tree-optimization/81455
15869 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15870 not walk in cycles when looking for guards.
15872 2017-07-25 Richard Biener <rguenther@suse.de>
15874 PR tree-optimization/81529
15875 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15876 when optimizing backedge uses.
15878 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15880 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15882 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15883 to dl_section_ref. On AIX, append an expression to subtract
15884 the size of the section length to dl_section_ref.
15886 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15888 * configure.ac: If any of the config.* scripts fail, exit 1.
15889 * configure: Regenerate.
15891 2017-07-25 Richard Biener <rguenther@suse.de>
15893 PR middle-end/81546
15894 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15895 of immediate uses, be more verbose on errors.
15897 2017-07-25 Richard Biener <rguenther@suse.de>
15899 PR tree-optimization/81510
15900 * tree-vect-loop.c (vect_is_simple_reduction): When the
15901 reduction stmt is not inside the loop bail out.
15903 2017-07-25 Richard Biener <rguenther@suse.de>
15905 PR tree-optimization/81303
15906 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15907 profitability check against LOOP_VINFO_NITERSM1.
15909 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15911 * domwalk.c (cmp_bb_postorder): Simplify.
15912 (sort_bbs_postorder): New function. Use it...
15913 (dom_walker::walk): ...here to optimize common cases.
15915 2017-07-25 Martin Liska <mliska@suse.cz>
15918 * ipa-visibility.c (function_and_variable_visibility): Make the
15919 redirection just on target that supports aliasing.
15920 Fix GNU coding style.
15922 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15925 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15927 * config/arm/rtems.h: Likewise.
15928 * config/bfin/rtems.h: Likewise.
15929 * config/i386/rtemself.h: Likewise.
15930 * config/lm32/rtems.h: Likewise.
15931 * config/m32c/rtems.h: Likewise.
15932 * config/m68k/rtemself.h: Likewise.
15933 * config/microblaze/rtems.h: Likewise.
15934 * config/mips/rtems.h: Likewise.
15935 * config/moxie/rtems.h: Likewise.
15936 * config/nios2/rtems.h: Likewise.
15937 * config/powerpcspe/rtems.h: Likewise.
15938 * config/rs6000/rtems.h: Likewise.
15939 * config/rtems.h: Likewise.
15940 * config/sh/rtems.h: Likewise.
15941 * config/sh/rtemself.h: Likewise.
15942 * config/sparc/rtemself.h: Likewise.
15944 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15947 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15948 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15949 * tree-ssa-structalias.c (alias_get_name): Same.
15951 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15954 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15955 instructions if no du chain is found.
15957 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15959 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15961 2017-07-25 Richard Biener <rguenther@suse.de>
15963 PR middle-end/81505
15964 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15967 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15969 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15970 upper-regs options.
15971 (ISA_2_7_MASKS_SERVER): Likewise.
15972 (ISA_3_0_MASKS_IEEE): Likewise.
15973 (OTHER_P8_VECTOR_MASKS): Likewise.
15974 (OTHER_VSX_VECTOR_MASKS): Likewise.
15975 (POWERPC_MASKS): Likewise.
15976 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15977 duplicate list of options.
15978 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15979 explicit -mupper-regs options.
15980 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15981 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15982 alias for -mupper-regs-df.
15983 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15984 (rs6000_init_hard_regno_mode_ok): Likewise.
15985 (rs6000_option_override_internal): Likewise.
15986 (rs6000_opt_masks): Likewise.
15987 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15988 options in terms of whether -mvsx or -mpower8-vector was used.
15989 (TARGET_UPPER_REGS_DI): Likewise.
15990 (TARGET_UPPER_REGS_SF): Likewise.
15991 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15992 -mupper-regs-* options.
15994 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15996 * passes.c (emergency_dump_function): Print some empty lines and a
15997 header before the RTL dump.
15999 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16001 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16003 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16006 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16007 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16009 2017-07-24 Carl Love <cel@us.ibm.com>
16011 * config/rs6000/rs6000-c.c: Add support for built-in functions
16012 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16013 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16014 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16015 vec_extract_fp_from_shortl): Add defines for the two builtins.
16016 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16017 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16019 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16020 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16021 * doc/extend.texi: Update the built-in documentation file for the
16022 new built-in function.
16024 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16027 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16029 * doc/generic.texi: Likewise.
16030 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16031 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16033 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16035 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16036 (aarch64_mls_elt_merge<mode>): Likewise.
16038 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16040 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16041 having __cxa_atexit.
16043 2017-07-23 Michael Collison <michael.collison@arm.com>
16045 * config/arm/arm.c (arm_option_override): Deprecate
16046 use of -mstructure-size-boundary.
16047 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16048 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16050 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16053 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16054 Reduce cost estimate for direct moves.
16056 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16059 * config/i386/i386.c (ix86_option_override_internal): Disable
16060 BMI, BMI2 and TBM instructions for -m16.
16062 2017-07-21 Carl Love <cel@us.ibm.com>
16064 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16065 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16066 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16067 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16068 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16069 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16070 VMULOSW): New enum "unspec" values.
16071 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16072 altivec_vmulosw): New patterns.
16073 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16074 VMULOSW): Add definitions.
16076 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16078 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16079 (qdf24xx): Likewise.
16080 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16081 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16082 (AARCH64_FL_V8_1): Renumber.
16083 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16084 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16085 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16086 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16087 rdma to feature modifiers list.
16089 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16091 PR middle-end/56727
16092 * ipa-visibility (function_and_variable_visibility): Convert
16093 recursive PLT call to direct call if appropriate.
16095 2017-07-21 Andrew Pinski <apinski@cavium.com>
16097 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16098 operand 1 to see if the types precision matches.
16099 * fold-const.c (operand_equal_p): Likewise.
16101 2017-07-21 Richard Biener <rguenther@suse.de>
16103 PR tree-optimization/81303
16104 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16105 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16106 (vect_peeling_hash_get_lowest_cost): Adjust.
16107 (vect_enhance_data_refs_alignment): Likewise. Use
16108 vect_get_peeling_costs_all_drs to compute the penalty for no
16109 peeling to match up costs.
16111 2017-07-21 Richard Biener <rguenther@suse.de>
16113 PR tree-optimization/81500
16114 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16115 we didn't identify a reduction path.
16117 2017-07-21 Tom de Vries <tom@codesourcery.com>
16118 Cesar Philippidis <cesar@codesourcery.com>
16120 PR gcov-profile/81442
16121 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16124 2017-07-21 Tom de Vries <tom@codesourcery.com>
16127 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16129 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16130 nvptx_override_options_after_change.
16132 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16134 * dwarf2out.c (output_file_names): Avoid double testing for
16135 dwarf_version >= 5.
16137 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16139 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16141 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16143 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16145 (try_crossjump_to_edge): Do not punt on partitioned functions.
16147 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16149 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16150 Put all BBs reachable only via paths crossing cold region to cold
16152 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16154 2016-07-21 Richard Biener <rguenther@suse.de>
16156 PR tree-optimization/81303
16157 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16158 into account prologue and epilogue iterations when raising
16159 min_profitable_iters to sth at least covering one vector iteration.
16161 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16163 * config/arm/arm.c (arm_test_cpu_arch_dat):
16166 2017-07-20 Nathan Sidwell <nathan@acm.org>
16168 Remove TYPE_METHODS.
16169 * tree.h (TYPE_METHODS): Delete.
16170 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16171 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16172 (dbxout_type_methods): Scan TYPE_FIELDS.
16173 (dbxout_type): Don't check TYPE_METHODS here.
16174 * function.c (use_register_for_decl): Always ignore register for
16175 class types when not optimizing.
16176 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16177 * tree.c (free_lang_data_in_type): Stitch out member functions and
16178 templates from TYPE_FIELDS.
16179 (build_distinct_type_copy, verify_type_variant,
16180 verify_type): Member fns are on TYPE_FIELDS.
16181 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16182 * tree-pretty-print.c (dump_generic_node): Likewise.
16184 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16187 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16188 V2TImode and V4TImode.
16189 (ix86_expand_vector_extract): Likewise.
16190 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16191 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16192 (ssescalarmode): Handle V4TImode and V2TImode.
16193 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16194 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16195 (VEXTRACTI128_MODE): New mode iterator.
16196 (splitter for *vec_extractv?ti first element): New.
16197 (VEC_INIT_MODE): New mode iterator.
16198 (vec_init<mode>): Consolidate 3 expanders into one using
16199 VEC_INIT_MODE mode iterator.
16201 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16203 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16204 non_spilled_static_chain_regno_p.
16206 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16208 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16210 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16212 * bb-reorder.c (connect_traces): Allow copying of blocks within
16215 2017-07-20 Richard Biener <rguenther@suse.de>
16217 * gimple.h (gimple_phi_result): Add gphi * overload.
16218 (gimple_phi_result_ptr): Likewise.
16219 (gimple_phi_arg): Likewise. Adjust index assert to only
16220 allow actual argument accesses rather than all slots available
16222 (gimple_phi_arg_def): Add gphi * overload.
16223 * tree-phinodes.c (make_phi_node): Initialize only actual
16225 (resize_phi_node): Clear memory not covered by old node,
16226 do not initialize excess argument slots.
16227 (reserve_phi_args_for_new_edge): Initialize new argument slot
16230 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16232 PR tree-optimization/81388
16234 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16236 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16237 by removing computation of may_be_zero.
16239 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16240 Tom de Vries <tom@codesourcery.com>
16242 PR middle-end/81030
16243 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16244 when gimple level profile disagrees with what RTL expander did.
16246 2017-07-20 Richard Biener <rguenther@suse.de>
16248 PR tree-optimization/61171
16249 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16250 (vect_analyze_stmt): Add slp instance parameter.
16251 (vectorizable_reduction): Likewise.
16252 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16253 (vect_is_simple_reduction): Deal with chains not detected
16254 as SLP reduction chain, specifically not properly associated
16255 chains containing a mix of plus/minus.
16256 (get_reduction_op): Remove.
16257 (get_initial_defs_for_reduction): Simplify, pass in whether
16258 this is a reduction chain, pass in the SLP node for the PHIs.
16259 (vect_create_epilog_for_reduction): Get the SLP instance as
16261 (vectorizable_reduction): Get the SLP instance as arg.
16262 During analysis remember the SLP node with the PHIs in the
16263 instance. Simplify getting at the vectorized reduction PHIs.
16264 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16265 through SLP instance.
16266 (vect_slp_analyze_operations): Likewise.
16267 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16268 (vect_transform_stmt): Likewise.
16270 2017-07-20 Tom de Vries <tom@codesourcery.com>
16272 PR tree-optimization/81489
16273 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16274 read of phi arg location to before loop that modifies phi.
16276 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16278 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16281 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16283 PR middle-end/81331
16284 * except.c (execute): Fix ordering issue.
16286 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16288 PR rtl-optimization/81423
16289 * combine.c (make_compound_operation_int): Don't try to optimize
16290 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16292 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16294 PR rtl-optimization/81423
16295 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16296 with a constant that is -1 in the truncated to mode.
16298 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16300 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16301 (determine_unlikely_bbs): ... here.
16302 * predict.h (propagate_unlikely_bbs_forward): Declare.
16303 * cfgexpand.c (pass_expand::execute): Use it.
16304 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16306 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16307 propagate_unlikely_bbs_forward.
16309 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16311 PR middle-end/81331
16312 * except.c (maybe_add_nop_after_section_switch): New function.
16315 2017-07-19 Tom de Vries <tom@codesourcery.com>
16317 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16319 2017-07-19 Tom de Vries <tom@codesourcery.com>
16321 * gimple.h (gimple_phi_arg): Make assert more strict.
16323 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16325 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16326 * config/rs6000/mmintrin.h: New file.
16327 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16329 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16331 PR tree-optimization/81346
16332 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16334 2017-07-19 Tom de Vries <tom@codesourcery.com>
16336 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16338 2017-07-19 Tom de Vries <tom@codesourcery.com>
16340 * config/nvptx/nvptx-modes.def: Add V2DImode.
16341 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16342 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16343 (nvptx_output_mov_insn): Handle lack of mov.b128.
16344 (nvptx_print_operand): Handle 'H' and 'L' codes.
16345 (nvptx_vector_mode_supported): Allow V2DImode.
16346 (nvptx_preferred_simd_mode): New function.
16347 (nvptx_data_alignment): New function.
16348 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16349 nvptx_preferred_simd_mode.
16350 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16352 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16354 2017-07-19 Tom de Vries <tom@codesourcery.com>
16356 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16357 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16358 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16359 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16360 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16361 (mov<VECIM>_insn): New define_insn.
16362 (define_expand "mov<VECIM>): New define_expand.
16364 2017-07-19 Tom de Vries <tom@codesourcery.com>
16366 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16368 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16370 PR tree-optimization/81346
16371 * fold-const.h (fold_div_compare, range_check_type): Declare.
16372 * fold-const.c (range_check_type): New function.
16373 (build_range_check): Use range_check_type.
16374 (fold_div_compare): No longer static, rewritten into
16375 a match.pd helper function.
16376 (fold_comparison): Don't call fold_div_compare here.
16377 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16378 as helper function.
16380 2017-07-19 Nathan Sidwell <nathan@acm.org>
16382 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16383 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16384 * tree.c (find_decls_types_r, verify_type): Use
16385 TYPE_{MIN,MAX}_VALUE_RAW.
16386 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16387 (hash_tree): Likewise.
16388 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16390 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16393 2017-07-18 Tom de Vries <tom@codesourcery.com>
16395 PR middle-end/81464
16396 * omp-expand.c (expand_omp_for_static_chunk): Handle
16397 equal-argument loop exit phi.
16399 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16402 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16403 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16404 operand 2 predicate.
16405 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16406 operand 2 predicate.
16407 (ror,rol -> rorx splitters): Use const_int_operand as
16408 operand 2 predicate.
16410 2017-06-18 Richard Biener <rguenther@suse.de>
16412 PR tree-optimization/81410
16413 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16414 the gap in the ! slp_perm SLP case after each group.
16416 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16418 PR middle-end/81463
16419 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16422 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16424 PR middle-end/81462
16425 * predict.c (set_even_probabilities): Cleanup; do not affect
16426 probabilities that are already known.
16427 (combine_predictions_for_bb): Call even when count is set.
16429 2017-07-18 Nathan Sidwell <nathan@acm.org>
16431 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16434 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16437 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16438 optimization for loop niter analysis.
16440 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16443 * config/avr/avr.c (avr_optimize_casesi): Don't use
16444 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16446 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16448 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16449 body_cost_vec from _vect_peel_extended_info.
16450 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16451 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16454 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16456 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16458 2017-07-18 Richard Biener <rguenther@suse.de>
16460 PR tree-optimization/80620
16461 PR tree-optimization/81403
16462 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16463 info when re-using a VN table entry.
16465 2017-07-18 Richard Biener <rguenther@suse.de>
16467 PR tree-optimization/81418
16468 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16469 vectype_in. Verify that with lane-reducing reduction operations
16470 we have a single def-use cycle.
16472 2017-07-17 Carl Love <cel@us.ibm.com>
16474 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16476 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16477 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16478 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16479 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16480 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16481 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16482 VMULOSW): New enum "unspec" values.
16483 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16484 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16485 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16486 altivec_vmulosw): New patterns.
16487 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16488 VMULOSW): Add definitions.
16489 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16491 * config/alpha/alpha.c: Include predict.h.
16493 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16495 * tree-vrp.c (compare_assert_loc): Fix comparison function
16496 to return predictable results.
16498 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16500 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16502 (subdi3): Likewise.
16503 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16504 * doc/invoke.texi (mexpand-adddi): Update text.
16506 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16508 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16509 that also clobbers the CC register. The old expand code is moved
16511 (*arc_clzsi2): ... here.
16512 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16513 the CC register. The old expand code is moved to ...
16514 (arc_ctzsi2): ... here.
16516 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16518 * config/arc/arc.opt (mindexed-loads): Use initial value
16519 TARGET_INDEXED_LOADS_DEFAULT.
16520 (mauto-modify-reg): Use initial value
16521 TARGET_AUTO_MODIFY_REG_DEFAULT.
16522 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16523 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16524 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16525 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16527 2017-07-17 Martin Liska <mliska@suse.cz>
16530 * opts.c (finish_options): Do not allow -fgnu-tm
16531 w/ -fsanitize={kernel-,}address. Say sorry.
16533 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16536 * tree-loop-distribution.c (classify_partition): Only assert on
16537 numer of iterations.
16538 (merge_dep_scc_partitions): Delete prameter. Update function call.
16539 (distribute_loop): Remove code handling loop with unknown niters.
16540 (pass_loop_distribution::execute): Skip loop with unknown niters.
16542 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16545 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16546 function sort_partitions_by_post_order.
16548 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16550 PR tree-optimization/81374
16551 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16552 the max index of basic blocks, rather than number of basic blocks.
16554 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16556 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16558 (arc_legitimate_pic_operand_p): Likewise.
16559 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16561 (arc_needs_pcl_p): Likewise.
16562 (arc_legitimate_pc_offset_p): Likewise.
16563 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16564 function is also used in constrains.md.
16565 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16566 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16567 PLUS. Only return true/false in known cases, otherwise assert.
16568 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16569 is already called in arc_legitimate_constant_p.
16570 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16572 (LEGITIMATE_PIC_OPERAND_P): Use
16573 arc_raw_symbolic_reference_mentioned_p function.
16574 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16579 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16580 Andrew Burgess <andrew.burgess@embecosm.com>
16582 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16583 (arc_return_address_register): New function.
16584 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16585 (arc_handle_fndecl_attribute): Add naked attribute.
16586 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16587 (TARGET_WARN_FUNC_RETURN): Likewise.
16588 (arc_allocate_stack_slots_for_args): New function.
16589 (arc_warn_func_return): Likewise.
16590 (machine_function): Change type fn_type.
16591 (arc_compute_function_type): Consider new naked function type,
16592 change function return type.
16593 (arc_must_save_register): Adapt to handle new
16594 arc_compute_function_type's return type.
16595 (arc_expand_prologue): Likewise.
16596 (arc_expand_epilogue): Likewise.
16597 (arc_return_address_regs): Delete.
16598 (arc_return_address_register): New function.
16599 (arc_epilogue_uses): Use above function.
16600 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16601 (arc_function_type): Change encoding, add naked type.
16602 (ARC_INTERRUPT_P): Change to handle the new encoding.
16603 (ARC_FAST_INTERRUPT_P): Likewise.
16604 (ARC_NORMAL_P): Define.
16605 (ARC_NAKED_P): Likewise.
16606 (arc_compute_function_type): Delete prototype.
16607 * config/arc/arc.md (in_ret_delay_slot): Use
16608 arc_return_address_register function.
16609 (simple_return): Likewise.
16610 (p_return_i): Likewise.
16612 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16614 PR tree-optimization/81428
16615 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16616 can't be built for those types.
16618 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16620 Remove stuff dead since r239246.
16622 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16623 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16624 (avr_inform_devices): Remove dead stuff.
16626 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16628 * config/arm/arm_neon.h: Fix softp typo.
16630 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16632 PR tree-optimization/81365
16633 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16634 aggregate moves onto bb predecessor edges, make sure there are no
16635 loads that could alias the lhs in between the start of bb and the
16638 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16641 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16642 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16643 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16645 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16647 PR tree-optimization/81396
16648 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16649 (init_symbolic_number): Initialize it to 1.
16650 (perform_symbolic_merge): Add n_ops from both operands into the new
16652 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16653 without base_addr as useless if they need more than one operation.
16654 (bswap_replace): Handle !bswap case for NULL base_addr.
16656 2017-07-17 Tom de Vries <tom@codesourcery.com>
16659 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16662 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16664 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16665 conditional builtin define __FIX_LEON3FT_B2BST.
16667 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16669 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16670 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16673 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16675 PR rtl-optimization/81424
16676 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16677 to remove potential trapping from operands if -fnon-call-exceptions.
16679 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16681 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16682 profile_proability for scalling.
16683 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16685 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16687 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16689 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16691 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16692 fixpoint arithmetics.
16694 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16696 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16697 fixpoint arithmetics.
16699 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16701 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16702 fixpoint arithmetics.
16704 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16706 * profile-count.h (profile_probability::from_reg_br_prob_note,
16707 profile_probability::to_reg_br_prob_note): New functions.
16708 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16709 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16710 * predict.c (probability_reliable_p): Update.
16711 (edge_probability_reliable_p): Update.
16712 (br_prob_note_reliable_p): Update.
16713 (invert_br_probabilities): Update.
16714 (add_reg_br_prob_note): New function.
16715 (combine_predictions_for_insn): Update.
16716 * asan.c (asan_clear_shadow): Update.
16717 * cfgbuild.c (compute_outgoing_frequencies): Update.
16718 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16719 (update_br_prob_note): Update.
16720 (rtl_verify_edges): Update.
16721 (purge_dead_edges): Update.
16722 (fixup_reorder_chain): Update.
16723 * emit-rtl.c (try_split): Update.
16724 * ifcvt.c (cond_exec_process_insns): Update.
16725 (cond_exec_process_if_block): Update.
16726 (dead_or_predicable): Update.
16727 * internal-fn.c (expand_addsub_overflow): Update.
16728 (expand_neg_overflow): Update.
16729 (expand_mul_overflow): Update.
16730 * loop-doloop.c (doloop_modify): Update.
16731 * loop-unroll.c (compare_and_jump_seq): Update.
16732 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16733 * predict.h: Update.
16734 * reorg.c (mostly_true_jump): Update.
16736 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16737 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16738 * config/arc/arc.c: (emit_unlikely_jump): Update.
16739 * config/arm/arm.c: (emit_unlikely_jump): Update.
16740 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16741 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16742 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16743 (ix86_print_operand): Update.
16744 (ix86_split_fp_branch): Update.
16745 (predict_jump): Update.
16746 * config/ia64/ia64.c (ia64_print_operand): Update.
16747 * config/mmix/mmix.c (mmix_print_operand): Update.
16748 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16749 (rs6000_expand_split_stack_prologue): Update.
16750 * config/rs6000/rs6000.c: Update.
16751 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16752 (s390_expand_vec_movstr): Update.
16753 (s390_expand_cs_tdsi): Update.
16754 (s390_expand_split_stack_prologue): Update.
16755 * config/sh/sh.c (sh_print_operand): Update.
16756 (expand_cbranchsi4): Update.
16757 (expand_cbranchdi4): Update.
16758 * config/sparc/sparc.c (output_v9branch): Update.
16759 * config/spu/spu.c (get_branch_target): Update.
16760 (ea_load_store_inline): Update.
16761 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16762 * config/tilepro/tilepro.c: Update.
16764 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16766 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16767 (gimplify_save_expr): Likewise.
16769 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16771 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16773 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16775 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16777 * timevar.def (TV_IPA_FNSUMMARY): Define.
16779 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16781 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16782 to back store errata sensitive sequence from being generated.
16783 (sqrtdf2_fix): Likewise.
16785 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16787 * tree-ssa-threadupdate.c (compute_path_counts,
16788 update_joiner_offpath_counts): Use profile_probability.
16790 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16793 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16795 * config/arm/arm-c.c (arm_cpu_builtins): Define
16796 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16798 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16800 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16801 array entries to represent __ieee128 versions of the
16802 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16803 scalar_extract_sig, and scalar_insert_exp built-in functions.
16804 (altivec_resolve_overloaded_builtin): Add special case handling
16805 for the __builtin_scalar_insert_exp function, as represented by
16806 the P9V_BUILTIN_VEC_VSIEDP constant.
16807 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16808 exponent support for __ieee128 argument.
16809 (VSESQP): Add scalar extract signature support for __ieee128
16811 (VSTDCNQP): Add scalar test negative support for __ieee128
16813 (VSIEQP): Add scalar insert exponent support for __int128 argument
16814 with __ieee128 result.
16815 (VSIEQPF): Add scalar insert exponent support for __ieee128
16816 argument with __ieee128 result.
16817 (VSTDCQP): Add scalar test data class support for __ieee128
16819 (VSTDCNQP): Add overload support for scalar test negative with
16820 __ieee128 argument.
16821 (VSTDCQP): Add overload support for scalar test data class
16822 __ieee128 argument.
16823 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16824 UNSPEC_VSX_SXSIGDP.
16825 (UNSPEC_VSX_SIEXPQP): New constant.
16826 (xsxexpqp): New insn for VSX scalar extract exponent quad
16828 (xsxsigqp): New insn for VSX scalar extract significand quad
16830 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16831 precision with floating point argument.
16832 (xststdcqp): New expand for VSX scalar test data class quad
16834 (xststdcnegqp): New expand for VSX scalar test negative quad
16836 (xststdcqp): New insn to match expansions for VSX scalar test data
16837 class quad precision and VSX scalar test negative quad precision.
16838 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16839 special case operand checking to enforce that second operand of
16840 VSX scalar test data class with quad precision argument is a 7-bit
16842 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16843 prototypes and descriptions of __ieee128 versions of
16844 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16845 scalar_test_data_class, and scalar_test_neg built-in functions.
16847 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16849 PR tree-optimization/81162
16850 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16851 replace a negate with an add.
16853 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16855 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16857 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16859 * config/arm/arm-c.c (arm_cpu_builtins): Define
16860 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16862 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16864 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16865 (armv8-r): Set ARM Cortex-R52 as default CPU.
16866 * config/arm/arm-tables.opt: Regenerate.
16867 * config/arm/arm-tune.md: Regenerate.
16868 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16870 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16871 extension for -mcpu=cortex-r52.
16873 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16875 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16876 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16877 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16878 (fp-armv8): Define it as FP_ARMv8 only.
16879 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16880 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16882 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16883 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16884 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16885 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16886 than TARGET_FPU_ARMV8.
16887 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16888 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16889 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16891 * config/arm/neon.md (neon_vrint): Likewise.
16892 (neon_vcvt): Likewise.
16893 (neon_<fmaxmin_op><mode>): Likewise.
16894 (<fmaxmin><mode>3): Likewise.
16895 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16896 * config/arm/predicates.md (arm_cond_move_operator): Check against
16897 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16899 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16901 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16902 to top of function.
16904 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16906 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16907 loop in comment with memset.
16909 2017-07-14 Martin Liska <mliska@suse.cz>
16911 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16912 * dwarf2out.c (is_java): Remove the function.
16913 (output_pubname): Remove usage of the function.
16914 (lower_bound_default): Remove usage of DW_LANG_Java.
16915 (gen_compile_unit_die): Likewise.
16916 * gcc.c: Remove compiler defaults for .java and .zip files.
16917 * gimple-expr.c (remove_suffix): Change as there's no longer
16918 extension than 4-letter one.
16919 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16920 (gimplify_save_expr): Likewise.
16921 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16922 as it's possible even for other languages than Java.
16923 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16924 * lto-opts.c (lto_write_options): Remove reference to Java.
16925 * opts.c (strip_off_ending): Update file extension handling.
16926 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16927 * tree-eh.c (lower_resx): Likewise.
16928 * tree.c (free_lang_data_in_type): Remove dead code.
16929 (find_decls_types_r): Likewise.
16930 (build_common_builtin_nodes): Remove Java from a comment.
16931 (verify_type): Remove dead code.
16932 * varasm.c (assemble_external): Remove Java from a comment.
16934 2017-07-14 Martin Liska <mliska@suse.cz>
16936 * opts.c (finish_options): Add quotes.
16937 (common_handle_option): Likewise.
16939 2017-07-14 Martin Liska <mliska@suse.cz>
16941 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16942 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16943 Remove N_SO_PASCAL.
16944 * dwarf2out.c (lower_bound_default): Do not handle
16946 (gen_compile_unit_die): Likewise.
16947 * gcc.c: Remove default extension binding for GNU Pascal.
16948 * stmt.c: Remove Pascal language from a comment.
16949 * xcoffout.c: Likewise.
16951 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16954 * diagnostic-show-locus.c (fixit_cmp): New function.
16955 (layout::layout): Sort m_fixit_hints.
16956 (column_range::column_range): Assert that the values are valid.
16957 (struct char_span): New struct.
16958 (correction::overwrite): New method.
16959 (struct source_line): New struct.
16960 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16961 calls in terms of classes source_line and char_span, and
16962 correction::overwrite.
16963 (selftest::test_overlapped_fixit_printing_2): New function.
16964 (selftest::diagnostic_show_locus_c_tests): Call it.
16966 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16968 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16969 early if there is no lhs.
16971 2017-07-13 Martin Liska <mliska@suse.cz>
16973 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16974 (gen_reference_type_die): Likewise.
16975 * stor-layout.c: Remove Pascal-related comment.
16977 2017-07-13 Martin Liska <mliska@suse.cz>
16979 * opts.c (finish_options): Add quotes to error messages.
16980 (parse_sanitizer_options): Likewise.
16982 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16984 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16986 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16988 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16990 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16992 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16994 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16996 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16999 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17000 provides the hardware capability bits, define the macro
17001 __BUILTIN_CPU_SUPPORTS__.
17002 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17003 if GLIBC does not provide the hardware capability bits. Add a
17004 gcc_unreachable call if the built-in cpu function is neither
17005 __builtin_cpu_is nor __builtin_cpu_supports.
17006 (rs6000_get_function_versions_dispatcher): Change the warning
17007 that an old GLIBC is used which does not export the capability
17008 bits to be an error.
17009 * doc/extend.texi (target_clones attribute): Document the
17010 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17011 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17012 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17013 the macros defined by GCC if the newer GLIBC is available.
17015 2017-07-12 Jeff Law <law@redhat.com>
17017 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17018 remaining includes slightly.
17019 * config/riscv/riscv-builtins.c: Include profile-count.h.
17021 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17024 * config/avr/avr.c (avr_set_current_function): In diagnostic
17025 messages: Quote keywords and (parts of) identifiers.
17026 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17029 2017-07-12 Carl Love <cel@us.ibm.com>
17031 * config/rs6000/rs6000-c.c: Add support for built-in functions
17032 vector bool char vec_revb (vector bool char);
17033 vector bool short vec_revb (vector short char);
17034 vector bool int vec_revb (vector bool int);
17035 vector bool long long vec_revb (vector bool long long);
17036 * doc/extend.texi: Update the built-in documentation file for the
17037 new built-in functions.
17039 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17041 * config/s390/s390.md: Remove movcc splitter.
17043 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17045 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17046 load/store on condition.
17048 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17051 * config/avr/avr.c (avr_encode_section_info)
17052 [progmem && !TREE_READONLY]: Error if progmem object needs
17055 2017-07-11 Michael Collison <michael.collison@arm.com>
17057 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17060 2017-07-11 Carl Love <cel@us.ibm.com>
17062 * config/rs6000/rs6000-c.c: Add support for builtins
17063 vector unsigned int vec_parity_lsbb (vector signed int);
17064 vector unsigned int vec_parity_lsbb (vector unsigned int);
17065 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17066 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17067 vector unsigned long long vec_parity_lsbb (vector signed long long);
17068 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17069 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17070 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17071 * doc/extend.texi: Update the built-in documentation file for the
17072 new built-in functions.
17074 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17076 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17077 (layout::m_primary_loc): New field.
17078 (layout::layout): Initialize new field. Move location filtering
17079 logic from here to...
17080 (layout::maybe_add_location_range): ...this new method. Add
17081 support for filtering to just the lines already specified by other
17083 (layout::will_show_line_p): New method.
17084 (gcc_rich_location::add_location_if_nearby): New method.
17085 (selftest::test_add_location_if_nearby): New test function.
17086 (selftest::diagnostic_show_locus_c_tests): Call it.
17087 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17090 2017-07-11 Tom de Vries <tom@codesourcery.com>
17092 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17093 (bb_first_real_insn): New function.
17094 (nvptx_single): Add extra initialization of broadcasted condition
17097 2017-07-11 Nathan Sidwell <nathan@acm.org>
17099 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17101 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17103 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17104 Binutils doc as TEXI will mess them up.
17105 * doc/invoke.texi (AVR Options): Same here.
17107 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17109 * config/sparc/sparc.opt (mfix-ut700): New option.
17110 (mfix-gr712rc): Likewise.
17111 (sparc_fix_b2bst): New variable.
17112 * doc/invoke.texi (SPARC options): Document them.
17113 (ARM options): Fix warnings.
17114 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17115 instructions to prevent sequences that can trigger the store-store
17116 errata for certain LEON3FT processors.
17117 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17118 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17119 * config/sparc/sparc.md (fix_b2bst): New attribute.
17120 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17122 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17125 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17127 (*rsqrtsf2_sse): Ditto.
17129 (div<mode>3): Macroize insn from divdf3 and divsf3
17130 using MODEF mode iterator.
17132 2017-07-10 Martin Sebor <msebor@redhat.com>
17134 PR tree-optimization/80397
17135 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17136 instead of testing for equality to INTEGER_TYPE.
17138 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17140 * config.gcc: Remove uclibc from arc target spec.
17142 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17144 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17146 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17149 * lto-wrapper.c (remove_option): New function.
17150 (merge_and_complain): Merge PIC/PIE options more realistically.
17152 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17154 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17158 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17159 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17160 * config.in: Regenerate.
17161 * configure: Regenerate.
17162 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17163 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17164 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17165 (TARGET_GASISR_PROLOGUES): ...target mask.
17166 * common/config/avr/avr-common.c
17167 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17168 Set -mgas-isr-prologues.
17169 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17170 INSERT_PASS_BEFORE for it.
17171 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17172 * config/avr/avr.c (avr_option_override)
17173 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17174 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17175 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17176 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17177 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17179 (make_avr_pass_pre_proep): New function.
17180 (emit_push_sfr) <treg>: Add argument to function and use it
17181 instead of TMP_REG.
17182 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17183 and set machine->gasisr.yes.
17184 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17185 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17186 __gcc_isr.n_pushed to .L__stack_usage.
17187 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17188 (avr_asm_final_postscan_insn): ...this new static function.
17189 * config/avr/avr.h (machine_function)
17190 <is_no_gccisr, use_L__stack_usage>: New fields.
17191 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17192 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17193 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17194 (gasisr, *gasisr): New expander and insn.
17195 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17196 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17197 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17199 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17201 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17204 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17206 Move jump-tables out of .text again.
17209 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17210 (ASM_OUTPUT_ADDR_VEC): New function.
17211 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17212 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17213 INSN_ADDRESSes as asm comment.
17214 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17215 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17216 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17217 * config/avr/avr.md (*tablejump): Adjust comment.
17218 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17219 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17221 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17222 (avr_output_addr_vec): New proto.
17223 (avr_log_t) <insn_addresses>: New field.
17225 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17228 * config/i386/i386.c (ix86_function_arg_advance): Set
17229 outgoing_args_on_stack to true if there are outgoing arguments
17231 (ix86_function_arg): Likewise.
17232 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17233 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17234 * config/i386/i386.h (machine_function): Add
17235 outgoing_args_on_stack.
17237 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17239 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17240 supporting pthreds.
17241 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17243 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17245 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17246 (REAL_H): Remove $(MACHMODE_H).
17247 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17249 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17250 $(MACHMODE_H) and double-int.h.
17251 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17253 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17256 2017-07-07 Andrew Pinski <apinski@cavium.com>
17258 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17259 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17261 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17263 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17264 Add warning if GCC was not configured to link against a GLIBC that
17265 exports the hardware capability bits.
17266 (make_resolver_func): Make resolver function private and not a
17267 COMDAT function. Create the name with clone_function_name instead
17268 of make_unique_name.
17271 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17272 correct operand in doing the split.
17274 2017-07-07 Carl Love <cel@us.ibm.com>
17276 * config/rs6000/rs6000-c: Add support for built-in function
17277 vector unsigned short vec_pack_to_short_fp32 (vector float,
17279 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17280 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17281 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17282 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17283 (convert_4f32_8i16): Add define_expand.
17284 * doc/extend.texi: Update the built-in documentation file for the
17285 new built-in function.
17287 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17289 * config/sparc/m8.md: New file.
17290 * config/sparc/sparc.md: Include m8.md.
17292 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17294 * config/sparc/sparc.opt: New option -mvis4b.
17295 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17296 (sparc_option_override): Handle VIS4B.
17297 (enum sparc_builtins): Define
17298 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17299 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17300 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17301 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17302 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17303 (check_constant_argument): New function.
17304 (sparc_vis_init_builtins): Define builtins
17305 __builtin_vis_dictunpack{8,16,32},
17306 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17307 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17308 __builtin_vis_fpcmpde{8,16,32}shl and
17309 __builtin_vis_fpcmpur{8,16,32}shl.
17310 (sparc_expand_builtin): Check that the constant operands to
17311 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17312 constant and in range.
17313 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17315 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17316 (SPARC_IMM5_P): Likewise.
17317 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17318 (enabled): Handle vis4b.
17319 (UNSPEC_DICTUNPACK): New unspec.
17320 (UNSPEC_FPCMPSHL): Likewise.
17321 (UNSPEC_FPUCMPSHL): Likewise.
17322 (UNSPEC_FPCMPDESHL): Likewise.
17323 (UNSPEC_FPCMPURSHL): Likewise.
17324 (cpu_feature): New CPU feature `vis4b'.
17325 (dictunpack{8,16,32}): New insns.
17326 (FPCSMODE): New mode iterator.
17327 (fpcscond): New code iterator.
17328 (fpcsucond): Likewise.
17329 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17330 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17331 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17332 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17333 * config/sparc/constraints.md: Define constraints `q' for unsigned
17334 2-bit integer constants and `t' for unsigned 5-bit integer
17336 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17338 (imm5_operand_dictunpack16): Likewise.
17339 (imm5_operand_dictunpack32): Likewise.
17340 (imm2_operand): Likewise.
17341 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17342 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17343 ditunpack* and fpcmp*shl builtins.
17345 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17347 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17348 * config.in: Add HAVE_AS_SPARC6 define.
17349 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17351 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17353 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17354 (CPP_CPU_SPEC): Handle m8.
17355 (ASM_CPU_SPEC): Likewise.
17356 * config/sparc/sparc-opts.h (enum processor_type): Add
17358 * config/sparc/sparc.c (m8_costs): New struct.
17359 (sparc_option_override): Handle TARGET_CPU_m8.
17360 (sparc32_initialize_trampoline): Likewise.
17361 (sparc64_initialize_trampoline): Likewise.
17362 (sparc_issue_rate): Likewise.
17363 (sparc_register_move_cost): Likewise.
17364 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17365 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17366 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17367 (CPP_CPU_SPEC): Handle M8.
17368 (ASM_CPU_SPEC): Likewise.
17369 (AS_M8_FLAG): Define.
17370 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17371 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17372 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17374 * configure: Regenerate.
17375 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17378 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17380 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17382 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17383 ("*movdi_insn_sp32"): Do not set v3pipe.
17384 ("*movsi_insn"): Likewise.
17385 ("*movdi_insn_sp64"): Likewise.
17386 ("*movsf_insn"): Likewise.
17387 ("*movdf_insn_sp32"): Likewise.
17388 ("*movdf_insn_sp64"): Likewise.
17389 ("*zero_extendsidi2_insn_sp64"): Likewise.
17390 ("*sign_extendsidi2_insn"): Likewise.
17391 ("*mov<VM32:mode>_insn"): Likewise.
17392 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17393 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17394 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17395 ("<vlop:code><VL:mode>3"): Likewise.
17396 ("*not_<vlop:code><VL:mode>3"): Likewise.
17397 ("*nand<VL:mode>_vis"): Likewise.
17398 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17399 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17400 ("one_cmpl<VL:mode>2"): Likewise.
17401 ("faligndata<VM64:mode>_vis"): Likewise.
17402 ("alignaddrsi_vis"): Likewise.
17403 ("alignaddrdi_vis"): Likweise.
17404 ("alignaddrlsi_vis"): Likewise.
17405 ("alignaddrldi_vis"): Likewise.
17406 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17407 ("bmaskdi_vis"): Likewise.
17408 ("bmasksi_vis"): Likewise.
17409 ("bshuffle<VM64:mode>_vis"): Likewise.
17410 ("cmask8<P:mode>_vis"): Likewise.
17411 ("cmask16<P:mode>_vis"): Likewise.
17412 ("cmask32<P:mode>_vis"): Likewise.
17413 ("pdistn<P:mode>_vis"): Likewise.
17414 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17416 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17418 * config/sparc/sparc.md ("subtype"): New insn attribute.
17419 ("*wrgsr_sp64"): Set insn subtype.
17420 ("*rdgsr_sp64"): Likewise.
17421 ("alignaddrsi_vis"): Likewise.
17422 ("alignaddrdi_vis"): Likewise.
17423 ("alignaddrlsi_vis"): Likewise.
17424 ("alignaddrldi_vis"): Likewise.
17425 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17426 ("fexpand_vis"): Likewise.
17427 ("fpmerge_vis"): Likewise.
17428 ("faligndata<VM64:mode>_vis"): Likewise.
17429 ("bshuffle<VM64:mode>_vis"): Likewise.
17430 ("cmask8<P:mode>_vis"): Likewise.
17431 ("cmask16<P:mode>_vis"): Likewise.
17432 ("cmask32<P:mode>_vis"): Likewise.
17433 ("fchksm16_vis"): Likewise.
17434 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17435 ("fmean16_vis"): Likewise.
17436 ("fp<plusminus_insn>64_vis"): Likewise.
17437 ("<plusminus_insn>v8qi3"): Likewise.
17438 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17439 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17440 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17441 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17442 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17443 ("*movqi_insn"): Likewise.
17444 ("*movhi_insn"): Likewise.
17445 ("*movsi_insn"): Likewise.
17446 ("movsi_pic_gotdata_op"): Likewise.
17447 ("*movdi_insn_sp32"): Likewise.
17448 ("*movdi_insn_sp64"): Likewise.
17449 ("movdi_pic_gotdata_op"): Likewise.
17450 ("*movsf_insn"): Likewise.
17451 ("*movdf_insn_sp32"): Likewise.
17452 ("*movdf_insn_sp64"): Likewise.
17453 ("*zero_extendhisi2_insn"): Likewise.
17454 ("*zero_extendqihi2_insn"): Likewise.
17455 ("*zero_extendqisi2_insn"): Likewise.
17456 ("*zero_extendqidi2_insn"): Likewise.
17457 ("*zero_extendhidi2_insn"): Likewise.
17458 ("*zero_extendsidi2_insn_sp64"): Likewise.
17459 ("ldfsr"): Likewise.
17460 ("prefetch_64"): Likewise.
17461 ("prefetch_32"): Likewise.
17462 ("tie_ld32"): Likewise.
17463 ("tie_ld64"): Likewise.
17464 ("*tldo_ldub_sp32"): Likewise.
17465 ("*tldo_ldub1_sp32"): Likewise.
17466 ("*tldo_ldub2_sp32"): Likewise.
17467 ("*tldo_ldub_sp64"): Likewise.
17468 ("*tldo_ldub1_sp64"): Likewise.
17469 ("*tldo_ldub2_sp64"): Likewise.
17470 ("*tldo_ldub3_sp64"): Likewise.
17471 ("*tldo_lduh_sp32"): Likewise.
17472 ("*tldo_lduh1_sp32"): Likewise.
17473 ("*tldo_lduh_sp64"): Likewise.
17474 ("*tldo_lduh1_sp64"): Likewise.
17475 ("*tldo_lduh2_sp64"): Likewise.
17476 ("*tldo_lduw_sp32"): Likewise.
17477 ("*tldo_lduw_sp64"): Likewise.
17478 ("*tldo_lduw1_sp64"): Likewise.
17479 ("*tldo_ldx_sp64"): Likewise.
17480 ("*mov<VM32:mode>_insn"): Likewise.
17481 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17482 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17484 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17486 * config/sparc/sparc.md ("type"): New insn type viscmp.
17487 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17489 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17490 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17491 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17492 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17494 ("n7_vis_logical_11cycle"): Likewise.
17495 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17496 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17497 * config/sparc/niagara.md ("niag_vis"): Likewise.
17498 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17499 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17501 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17503 * config/sparc/sparc.md: New instruction type `bmask'.
17504 (bmaskdi_vis): Use the `bmask' type.
17505 (bmasksi_vis): Likewise.
17506 * config/sparc/ultra3.md (us3_array): Likewise.
17507 * config/sparc/niagara7.md (n7_array): Likewise.
17508 * config/sparc/niagara4.md (n4_array): Likewise.
17509 * config/sparc/niagara2.md (niag2_vis): Likewise.
17510 (niag3_vis): Likewise.
17511 * config/sparc/niagara.md (niag_vis): Likewise.
17513 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17515 * ipa-comdats.c: Remove optimize check from gate.
17516 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17517 for functions not optimized.
17518 (ipa_fn_summary_read): Skip optimize check.
17519 (ipa_fn_summary_write): Likewise.
17520 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17522 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17524 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17525 (check_callers): Give up on caller which is not optimized.
17526 (inline_small_functions): Likewise.
17527 (ipa_inline): Do not give up when not optimizing.
17528 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17529 away unoptimizes cdtors.
17530 (whole_program_function_and_variable_visibility): Do
17531 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17532 * ipa.c (process_references): Do not check optimize.
17533 (symbol_table::remove_unreachable_nodes): Update optimize check.
17534 (set_writeonly_bit): Update optimize check.
17535 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17536 (pass_ipa_single_use::gate): Remove.
17538 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17540 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17541 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17542 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17543 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17544 permute_load, permute_store, adjust_extract, adjust_splat,
17545 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17546 replace_swap_with_copy, dump_swap_insn_table,
17547 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17548 recombine_lvx_pattern, recombine_stvx_pattern,
17549 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17550 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17551 to file rs6000-p8swap.c.
17552 * config/rs6000/rs6000-p8swap.c: New file.
17553 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17554 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17555 and rs6000*-*-* targets.
17557 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17559 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17561 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17563 * lto-wrapper.c (merge_and_complain): Do not merge
17564 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17565 fsigned_zeros, ftrapping_math, fwrapv.
17566 (append_compiler_options): Do not track these options.
17567 (append_linker_options): Likewie
17569 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17571 * cgraphunit.c (cgraph_node::finalize_function): When
17572 !flag_toplevel_reorde set no_reorder flag.
17573 (varpool_node::finalize_decl): Likewise.
17574 (symbol_table::compile): Drop no toplevel reorder path.
17576 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17578 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17579 edges; zero probability is not better than uninitialized.
17581 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17583 * asan.h (asan_sanitize_allocas_p): Declare.
17584 * asan.c (asan_sanitize_allocas_p): New function.
17585 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17586 (handle_builtin_alloca): Likewise.
17587 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17588 if !asan_sanitize_allocas_p.
17589 * params.def (asan-instrument-allocas): Add new option.
17590 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17591 * opts.c (common_handle_option): Disable allocas sanitization for
17594 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17596 * asan.c: Include gimple-fold.h.
17597 (get_last_alloca_addr): New function.
17598 (handle_builtin_stackrestore): Likewise.
17599 (handle_builtin_alloca): Likewise.
17600 (asan_emit_allocas_unpoison): Likewise.
17601 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17602 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17603 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17604 (instrument_builtin_call): Pass gimple iterator to
17605 get_mem_refs_of_builtin_call.
17606 (last_alloca_addr): New global.
17607 * asan.h (asan_emit_allocas_unpoison): Declare.
17608 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17609 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17610 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17611 if function calls alloca.
17612 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17613 * gimple-fold.h (replace_call_with_value): Declare.
17614 * internal-fn.c: Include asan.h.
17615 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17616 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17618 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17620 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17621 (C_SELFTEST_FLAGS): New.
17622 (CPP_SELFTEST_FLAGS): New.
17623 (SELFTEST_DEPS): New, from deps of s-selftest.
17624 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17625 (CPP_SELFTEST_DEPS): New.
17626 (selftest): Add dependency on s-selftest-c++.
17627 (s-selftest): Rename to...
17628 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17629 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17630 than SELFTEST_FLAGS.
17631 (selftest-gdb): Rename to...
17632 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17634 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17635 (selftest-valgrind): Rename to...
17636 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17638 (selftest-valgrind): Reintroduce as an alias for
17639 selftest-c-valgrind.
17640 (s-selftest-c++): New.
17641 (selftest-c++-gdb): New.
17642 (selftest-c++-valgrind): New.
17644 2017-07-06 Olivier Hainque <hainque@adacore.com>
17646 * gcc.c (process_command): When deciding if undefined variables
17647 should be ignored when processing specs, accept "gcc -v" as well.
17649 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17651 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17652 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17654 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17656 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17657 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17658 * config/arm/arm-tables.opt: Regenerate.
17659 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17661 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17663 2017-07-06 Carl Love <cel@us.ibm.com>
17665 * ChangeLog: Clean up from mid air collision
17667 2017-07-06 Carl Love <cel@us.ibm.com>
17669 * config/rs6000/rs6000-c.c: Add support for built-in functions
17670 vector signed int vec_subc (vector signed int, vector signed int);
17671 vector signed __int128 vec_subc (vector signed __int128,
17672 vector signed __int128);
17673 vector unsigned __int128 vec_subc (vector unsigned __int128,
17674 vector unsigned __int128);
17675 vector signed int vec_sube (vector signed int, vector signed int,
17676 vector signed int);
17677 vector unsigned int vec_sube (vector unsigned int,
17678 vector unsigned int,
17679 vector unsigned int);
17680 vector signed __int128 vec_sube (vector signed __int128,
17681 vector signed __int128,
17682 vector signed__int128);
17683 vector unsigned __int128 vec_sube (vector unsigned __int128,
17684 vector unsigned __int128,
17685 vector unsigned __int128);
17686 vector signed int vec_subec (vector signed int, vector signed int,
17687 vector signed int);
17688 vector unsigned int vec_subec (vector unsigned int,
17689 vector unsigned int,
17690 vector unsigned int);
17691 vector signed __int128 vec_subec (vector signed __int128,
17692 vector signed __int128,
17693 vector signed__int128);
17694 vector unsigned __int128 vec_subec (vector unsigned __int128,
17695 vector unsigned __int128,
17696 vector unsigned __int128);
17697 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17698 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17699 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17700 BU_ALTIVEC_OVERLOAD_X definitions.
17701 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17702 * doc/extend.texi: Update the built-in documentation file for the new
17703 built-in functions.
17705 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17708 * diagnostic-show-locus.c (layout::layout): Use start and finish
17709 spelling location for the start and finish of each range.
17710 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17711 Add unused aspect param.
17712 * input.c (expand_location_1): Add "aspect" param, and use it
17713 to access the correct part of the location.
17714 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17716 (expand_location_to_spelling_point): Likewise.
17717 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17718 param, and pass it to expand_location_1.
17720 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17722 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17723 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17724 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17725 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17726 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17727 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17728 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17729 _mm_maskz_getmant_ss): New intrinsics.
17730 (__builtin_ia32_getexpss128_mask): Changed to ...
17731 __builtin_ia32_getexpss128_round ... this.
17732 (__builtin_ia32_getexpsd128_mask): Changed to ...
17733 __builtin_ia32_getexpsd128_round ... this.
17734 * config/i386/i386-builtin-types.def
17735 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17736 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17737 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17738 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17739 __builtin_ia32_getmantss_mask_round): New builtins.
17740 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17741 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17742 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17743 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17744 * config/i386/sse.md
17745 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17746 avx512f_sgetexp<mode><mask_scalar_name>
17747 <round_saeonly_scalar_name> ... this.
17748 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17749 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17750 vgetexp<ssescalarmodesuffix>
17751 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17752 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17753 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17754 avx512f_vgetmant<mode><mask_scalar_name>
17755 <round_saeonly_scalar_name> ... this.
17756 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17757 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17758 vgetmant<ssescalarmodesuffix>
17759 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17760 %0<mask_scalar_operand4>, %1, %2
17761 <round_saeonly_scalar_mask_op4>, %3} ... this.
17762 * config/i386/subst.md (mask_scalar_operand4,
17763 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17764 round_saeonly_scalar_nimm_predicate): New subst attributes.
17766 2017-07-06 Julia Koval <julia.koval@intel.com>
17768 * config/i386/i386.c (ix86_erase_embedded_rounding):
17769 Remove code for old rounding pattern.
17771 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17773 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17775 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17777 * doc/sourcebuild.texi (Test Directives, Variants of
17778 dg-require-support): Add documentation for dg-require-stack-check.
17780 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17782 * config/i386/subst.md (mask_scalar, round_scalar,
17783 round_saeonly_scalar): New meta-templates.
17784 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17785 round_scalar_mask_operand3, round_scalar_mask_op3,
17786 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17787 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17788 round_saeonly_scalar_constraint,
17789 round_saeonly_scalar_prefix): New subst attribute.
17790 * config/i386/sse.md
17791 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17792 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17793 <round_scalar_name> ... this.
17794 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17795 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17796 <round_scalar_name> ... this.
17797 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17798 <sse>_vm<code><mode>3<mask_scalar_name>
17799 <round_saeonly_scalar_name> ... this.
17800 (v<plusminus_mnemonic><ssescalarmodesuffix>
17801 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17802 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17803 v<plusminus_mnemonic><ssescalarmodesuffix>
17804 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17805 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17806 (v<multdiv_mnemonic><ssescalarmodesuffix>
17807 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17808 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17809 v<multdiv_mnemonic><ssescalarmodesuffix>
17810 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17811 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17812 (v<maxmin_float><ssescalarmodesuffix>
17813 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17814 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17815 v<maxmin_float><ssescalarmodesuffix>
17816 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17817 %0<mask_scalar_operand3>, %1, %<iptr>2
17818 <round_saeonly_scalar_mask_op3>} ... this.
17820 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17822 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17823 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17825 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17826 Alan Hayward <alan.hayward@arm.com>
17827 David Sherwood <david.sherwood@arm.com>
17829 * combine.c (simplify_if_then_else): Remove "enum" before
17831 * compare-elim.c (can_eliminate_compare): Likewise.
17832 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17834 (aarch64_lookup_simd_builtin_type): Likewise.
17835 (aarch64_simd_builtin_type): Likewise.
17836 (aarch64_init_simd_builtin_types): Likewise.
17837 (aarch64_simd_expand_args): Likewise.
17838 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17840 (aarch64_reverse_mask): Likewise.
17841 (aarch64_simd_emit_reg_reg_move): Likewise.
17842 (aarch64_gen_adjusted_ldpstp): Likewise.
17843 (aarch64_ccmp_mode_to_code): Likewise.
17844 (aarch64_operands_ok_for_ldpstp): Likewise.
17845 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17846 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17848 (aarch64_min_divisions_for_recip_mul): Likewise.
17849 (aarch64_reassociation_width): Likewise.
17850 (aarch64_get_condition_code_1): Likewise.
17851 (aarch64_simd_emit_reg_reg_move): Likewise.
17852 (aarch64_simd_attr_length_rglist): Likewise.
17853 (aarch64_reverse_mask): Likewise.
17854 (aarch64_operands_ok_for_ldpstp): Likewise.
17855 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17856 (aarch64_gen_adjusted_ldpstp): Likewise.
17857 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17859 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17860 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17861 (arm_lookup_simd_builtin_type): Likewise.
17862 (arm_simd_builtin_type): Likewise.
17863 (arm_init_simd_builtin_types): Likewise.
17864 (arm_expand_builtin_args): Likewise.
17865 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17866 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17867 (ft32_setup_incoming_varargs): Likewise.
17868 (ft32_function_arg): Likewise.
17869 (ft32_function_arg_advance): Likewise.
17870 (ft32_pass_by_reference): Likewise.
17871 (ft32_arg_partial_bytes): Likewise.
17872 (ft32_valid_pointer_mode): Likewise.
17873 (ft32_addr_space_pointer_mode): Likewise.
17874 (ft32_addr_space_legitimate_address_p): Likewise.
17875 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17877 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17878 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17879 (iamcu_alignment): Likewise.
17880 (canonicalize_vector_int_perm): Likewise.
17881 (ix86_noce_conversion_profitable_p): Likewise.
17882 (ix86_mpx_bound_mode): Likewise.
17883 (ix86_operands_ok_for_move_multiple): Likewise.
17884 * config/microblaze/microblaze-protos.h
17885 (microblaze_expand_conditional_branch_reg): Likewise.
17886 * config/microblaze/microblaze.c
17887 (microblaze_expand_conditional_branch_reg): Likewise.
17888 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17890 (rs6000_reassociation_width): Likewise.
17891 (rs6000_invalid_binary_op): Likewise.
17892 (fusion_p9_p): Likewise.
17893 (emit_fusion_p9_load): Likewise.
17894 (emit_fusion_p9_store): Likewise.
17895 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17897 (riscv_hard_regno_mode_ok_p): Likewise.
17898 (riscv_address_insns): Likewise.
17899 (riscv_split_symbol): Likewise.
17900 (riscv_legitimize_move): Likewise.
17901 (riscv_function_value): Likewise.
17902 (riscv_hard_regno_nregs): Likewise.
17903 (riscv_expand_builtin): Likewise.
17904 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17905 (riscv_build_integer): Likewise.
17906 (riscv_split_integer): Likewise.
17907 (riscv_legitimate_constant_p): Likewise.
17908 (riscv_cannot_force_const_mem): Likewise.
17909 (riscv_regno_mode_ok_for_base_p): Likewise.
17910 (riscv_valid_base_register_p): Likewise.
17911 (riscv_valid_offset_p): Likewise.
17912 (riscv_valid_lo_sum_p): Likewise.
17913 (riscv_classify_address): Likewise.
17914 (riscv_legitimate_address_p): Likewise.
17915 (riscv_address_insns): Likewise.
17916 (riscv_load_store_insns): Likewise.
17917 (riscv_force_binary): Likewise.
17918 (riscv_split_symbol): Likewise.
17919 (riscv_force_address): Likewise.
17920 (riscv_legitimize_address): Likewise.
17921 (riscv_move_integer): Likewise.
17922 (riscv_legitimize_const_move): Likewise.
17923 (riscv_legitimize_move): Likewise.
17924 (riscv_address_cost): Likewise.
17925 (riscv_subword): Likewise.
17926 (riscv_output_move): Likewise.
17927 (riscv_canonicalize_int_order_test): Likewise.
17928 (riscv_emit_int_order_test): Likewise.
17929 (riscv_function_arg_boundary): Likewise.
17930 (riscv_pass_mode_in_fpr_p): Likewise.
17931 (riscv_pass_fpr_single): Likewise.
17932 (riscv_pass_fpr_pair): Likewise.
17933 (riscv_get_arg_info): Likewise.
17934 (riscv_function_arg): Likewise.
17935 (riscv_function_arg_advance): Likewise.
17936 (riscv_arg_partial_bytes): Likewise.
17937 (riscv_function_value): Likewise.
17938 (riscv_pass_by_reference): Likewise.
17939 (riscv_setup_incoming_varargs): Likewise.
17940 (riscv_print_operand): Likewise.
17941 (riscv_elf_select_rtx_section): Likewise.
17942 (riscv_save_restore_reg): Likewise.
17943 (riscv_for_each_saved_reg): Likewise.
17944 (riscv_register_move_cost): Likewise.
17945 (riscv_hard_regno_mode_ok_p): Likewise.
17946 (riscv_hard_regno_nregs): Likewise.
17947 (riscv_class_max_nregs): Likewise.
17948 (riscv_memory_move_cost): Likewise.
17949 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17950 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17951 (rl78_addr_space_address_mode): Likewise.
17952 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17954 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17955 (rs6000_reassociation_width): Likewise.
17956 (rs6000_invalid_binary_op): Likewise.
17957 (fusion_p9_p): Likewise.
17958 (emit_fusion_p9_load): Likewise.
17959 (emit_fusion_p9_store): Likewise.
17960 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17961 (ok_for_simple_move_operands): Likewise.
17962 (ok_for_simple_move_strict_operands): Likewise.
17963 (ok_for_simple_arith_logic_operands): Likewise.
17964 (visium_legitimize_reload_address): Likewise.
17965 (visium_select_cc_mode): Likewise.
17966 (output_cbranch): Likewise.
17967 (visium_split_double_move): Likewise.
17968 (visium_expand_copysign): Likewise.
17969 (visium_expand_int_cstore): Likewise.
17970 (visium_expand_fp_cstore): Likewise.
17971 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17972 (visium_function_arg): Likewise.
17973 (visium_function_arg_advance): Likewise.
17974 (visium_libcall_value): Likewise.
17975 (visium_setup_incoming_varargs): Likewise.
17976 (visium_legitimate_constant_p): Likewise.
17977 (visium_legitimate_address_p): Likewise.
17978 (visium_legitimize_address): Likewise.
17979 (visium_secondary_reload): Likewise.
17980 (visium_register_move_cost): Likewise.
17981 (visium_memory_move_cost): Likewise.
17982 (prepare_move_operands): Likewise.
17983 (ok_for_simple_move_operands): Likewise.
17984 (ok_for_simple_move_strict_operands): Likewise.
17985 (ok_for_simple_arith_logic_operands): Likewise.
17986 (visium_function_value_1): Likewise.
17987 (rtx_ok_for_offset_p): Likewise.
17988 (visium_legitimize_reload_address): Likewise.
17989 (visium_split_double_move): Likewise.
17990 (visium_expand_copysign): Likewise.
17991 (visium_expand_int_cstore): Likewise.
17992 (visium_expand_fp_cstore): Likewise.
17993 (visium_split_cstore): Likewise.
17994 (visium_select_cc_mode): Likewise.
17995 (visium_split_cbranch): Likewise.
17996 (output_cbranch): Likewise.
17997 (visium_print_operand_address): Likewise.
17998 * expmed.c (flip_storage_order): Likewise.
17999 * expmed.h (emit_cstore): Likewise.
18000 (flip_storage_order): Likewise.
18001 * genrecog.c (validate_pattern): Likewise.
18002 * hsa-gen.c (gen_hsa_addr): Likewise.
18003 * internal-fn.c (expand_arith_overflow): Likewise.
18004 * ira-color.c (allocno_copy_cost_saving): Likewise.
18005 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18006 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18007 (process_invariant_for_inheritance): Likewise.
18008 * lra-eliminations.c (move_plus_up): Likewise.
18009 * omp-low.c (lower_oacc_reductions): Likewise.
18010 * simplify-rtx.c (simplify_subreg): Likewise.
18011 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18012 (TARGET_CHKP_BOUND_MODE): Likewise..
18013 * targhooks.c (default_chkp_bound_mode): Likewise.
18014 (default_setup_incoming_vararg_bounds): Likewise.
18015 * targhooks.h (default_chkp_bound_mode): Likewise.
18016 (default_setup_incoming_vararg_bounds): Likewise.
18017 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18018 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18019 (have_whole_vector_shift): Likewise.
18020 * tree-vect-stmts.c (vectorizable_load): Likewise.
18021 * doc/tm.texi: Regenerate.
18023 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18025 Graceful degrade if Binutils PR21472 is not available.
18028 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18029 .rodata in flash test fails.
18030 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18031 * confgure: Regenerate.
18032 * config.in: Regenerate.
18033 * config/avr/avr.c (avr_asm_named_section)
18034 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18035 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18036 (avr_asm_init_sections): Same.
18038 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18040 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18041 (fma<VH:mode>4_intrinsic): Likewise.
18042 (*fmsub<VCVTF:mode>4): Likewise.
18043 (*fmsub<VH:mode>4_intrinsic): Likewise.
18045 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18048 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18049 Don't depend on "optimize > 0".
18050 (out_movhi_r_mr, out_movqi_mr_r): Same.
18051 (out_movhi_mr_r, out_movqi_r_mr): Same.
18052 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18053 io_address_operand on "optimize > 0".
18055 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18057 * tree-loop-distribution.c: Add general explanantion on the pass.
18058 (generate_loops_for_partition): Mark distributed loop.
18059 (pg_add_dependence_edges): New parameter. Handle alias data
18060 dependence specially and record it in the parameter if asked.
18061 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18062 (init_partition_graph_vertices, add_partition_graph_edge): New.
18063 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18064 (free_partition_graph_vdata, build_partition_graph): New.
18065 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18066 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18067 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18068 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18069 (version_for_distribution_p, finalize_partitions): New.
18070 (distribute_loop): Handle alias data dependence specially. Factor
18071 out loop fusion code as functions and call these functions.
18073 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18075 * tree-loop-distribution.c (classify_partition): New parameter and
18076 better handle reduction statement.
18077 (rdg_build_partitions): Revise comment.
18078 (distribute_loop): Compute statements in all partitions and pass it
18079 to classify_partition.
18081 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18083 * tree-loop-distribution.c (enum partition_type): New.
18084 (struct partition): New field type.
18085 (partition_merge_into): Add parameter. Update partition type.
18086 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18087 (build_rdg_partition_for_vertex): Compute partition type.
18088 (rdg_build_partitions): Dump partition type.
18089 (distribute_loop): Update calls to partition_merge_into.
18091 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18093 * tree-loop-distribution.c (struct ddr_hasher): New.
18094 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18096 (classify_partition): Call get_data_dependence.
18097 (pg_add_dependence_edges): Ditto.
18098 (distribute_loop): Release data dependence hash table.
18100 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18102 * tree-loop-distribution.c (ref_base_address): Delete.
18103 (similar_memory_accesses): Rename ...
18104 (share_memory_accesses): ... to this. Check if partitions access
18105 the same memory reference.
18106 (distribute_loop): Call share_memory_accesses.
18108 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18110 * tree-loop-distribution.c (struct partition): New field recording
18111 its data reference.
18112 (partition_alloc, partition_free): Init and release data refs.
18113 (partition_merge_into): Merge data refs.
18114 (build_rdg_partition_for_vertex): Collect data refs for partition.
18115 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18117 (distribute_loop): Remve data refs from vertice data of partition
18120 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18122 * tree-loop-distribution.c (params.h): Include header file.
18123 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18124 (datarefs_vec): New global var.
18125 (create_rdg_vertices): Use datarefs_vec directly.
18126 (free_rdg): Don't free data references.
18127 (build_rdg): Update use. Don't free data references.
18128 (distribute_loop): Compute global variable for data references.
18129 Bail out if there are too many data references.
18131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18133 * tree-loop-distribution.c (loop_nest): New global var.
18134 (build_rdg): Use loop directly, rather than loop nest.
18135 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18137 (distribute_loop): Compute global variable loop nest. Update use.
18139 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18141 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18142 (partition_merge_into): New parameter. Dump reason for fusion.
18143 (distribute_loop): Update use of partition_merge_into.
18145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18147 * tree-loop-distribution.c (bb_top_order_index): New.
18148 (bb_top_order_index_size, bb_top_order_cmp): New.
18149 (stmts_from_loop): Use topological order.
18150 (pass_loop_distribution::execute): Compute and release topological
18151 order for basic blocks.
18153 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18155 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18158 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18160 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18161 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18162 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18163 * internal-fn.def (LOOP_DIST_ALIAS): New.
18164 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18165 (fold_loop_internal_call): ... this.
18166 (vect_loop_dist_alias_call): New function.
18167 (set_uid_loop_bbs): Call fold_loop_internal_call.
18168 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18171 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18174 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18175 Require dead FLAGS_REG at the beginning of a peephole.
18177 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18180 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18181 arguments in the call to __builtin_ia32_sbb_u32.
18182 (_subborrow_u64): Swap _X and _Y arguments in the call to
18183 __builtin_ia32_sbb_u64.
18185 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18188 * tree-vrp.c (compare_assert_loc): Turn into a function template
18189 with stable template parameter. Only test if a->e is NULL,
18190 !a->e == !b->e has been verified already. Use e == NULL or
18191 e != NULL instead of e or ! e tests. If stable is true, don't use
18192 iterative_hash_expr, on the other side allow a or b or both NULL
18193 and sort the NULLs last.
18194 (process_assert_insertions): Sort using compare_assert_loc<false>
18195 instead of compare_assert_loc, later sort using
18196 compare_assert_loc<true> before calling process_assert_insertions_for
18197 in a loop. Use break instead of continue once seen NULL pointer.
18199 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18201 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18202 Cortex-R7 and Cortex-R8 processors.
18204 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18206 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18207 uninitialized while src is not.
18209 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18211 * common/config/arm/arm-common.c: Adjust include path for
18213 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18214 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18215 (arm-cpu-data.h): Likewise.
18216 (arm-cpu-cdata.h): Likewise.
18217 * config/arm/arm-cpu.h: Delete.
18218 * config/arm/arm-cpu-cdata.h: Delete.
18219 * config/arm/arm-cpu-data.h: Delete.
18221 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18223 * config/arm/arm-cpus.in (cortex-a55): New.
18224 (cortex-a75): Likewise.
18225 (cortex-a75.cortex-a55): Likewise.
18226 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18228 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18229 * config/arm/arm-cpu-cdata.h: Regenerate.
18230 * config/arm/arm-cpu-data.h: Regenerate.
18231 * config/arm/arm-cpu.h: Regenerate.
18232 * config/arm/arm-tables.opt: Regenerate.
18233 * config/arm/arm-tune.md: Regenerate.
18235 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18237 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18239 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18241 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18244 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18246 PR tree-optimization/81292
18247 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18248 full_string_p, also call adjust_related_strinfos if the adjustment
18249 is simple, otherwise invalidate related strinfos.
18251 2017-07-04 Martin Liska <mliska@suse.cz>
18254 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18255 newly created variable as DECL_IGNORED_P.
18257 2017-07-04 Martin Liska <mliska@suse.cz>
18260 * ipa-inline.c (inline_small_functions):
18261 Use xstrdup_for_dump.
18263 2017-07-04 Tom de Vries <tom@codesourcery.com>
18265 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18267 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18270 * config/darwin.c (darwin_function_switched_text_sections):
18273 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18275 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18277 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18279 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18281 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18283 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18284 min_profitable_iters, and th as inclusive lower bounds.
18285 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18286 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18287 for min_profitable_iters and min_profitable_estimate.
18288 (vect_transform_loop): Treat th as an inclusive lower bound.
18289 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18291 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18294 * config/darwin.c (darwin_function_switched_text_sections):
18295 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18296 in two pieces, and suppress the use of buf.
18298 2017-07-03 Nathan Sidwell <nathan@acm.org>
18300 * hash-table.h (hash_table_mod1): Fix indentation.
18302 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18304 PR middle-end/81290
18305 * predict.c (force_edge_cold): Be more careful about propagation
18307 * profile-count.h (profile_probability::guessed,
18308 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18310 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18312 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18314 * doc/invoke.texi (rcpc architecture extension): Document it.
18316 2017-07-03 Richard Biener <rguenther@suse.de>
18318 PR tree-optimization/60510
18319 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18320 the scalar reduction PHI and use it.
18321 (vectorizable_reduction): Properly guard the single_defuse_cycle
18322 path for non-SLP reduction chains where we cannot use it.
18323 Rework reduc_def/index and vector type deduction. Rework
18324 vector operand gathering during reduction op code-gen.
18325 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18326 chains dissolve the chain and leave it to non-SLP reduction
18329 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18331 * tree-data-ref.h (dr_alignment): Declare.
18332 * tree-data-ref.c (dr_alignment): New function.
18333 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18334 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18336 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18338 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18340 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18341 and base_misalignment fields.
18342 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18343 * tree-data-ref.c: Include builtins.h.
18344 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18345 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18346 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18347 * tree-vect-data-refs.c: Include tree-cfg.h.
18348 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18349 fields instead of calculating an alignment here.
18350 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18351 innermost_loop_behavior fields.
18353 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18355 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18357 (DR_STEP_ALIGNMENT): New macro.
18358 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18359 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18360 (create_data_ref): Print it.
18361 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18362 to tell whether the step preserves vector (mis)alignment.
18363 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18364 Move the check for an integer step and generalise to all INTEGER_CST.
18365 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18366 Print the outer step alignment.
18368 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18370 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18371 with offset_alignment.
18372 (DR_ALIGNED_TO): Delete.
18373 (DR_OFFSET_ALIGNMENT): New macro.
18374 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18375 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18376 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18377 (create_data_ref): Likewise.
18378 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18379 (vect_analyze_data_refs): Likewise.
18380 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18381 creating dummy innermost behavior.
18383 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18385 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18386 with a "innermost_loop_behavior *" and refeence tree.
18387 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18388 (create_data_ref): Update call accordingly.
18389 * tree-predcom.c (find_looparound_phi): Likewise.
18391 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18393 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18394 fields with dr_wrt_vec_loop.
18395 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18396 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18397 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18398 (vect_dr_behavior): New function.
18399 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18400 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18401 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18402 track whether the step preserves the misalignment.
18403 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18404 Use vect_dr_behavior.
18405 (vect_setup_realignment): Update call accordingly.
18406 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18407 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18408 call to vect_create_addr_base_for_vector_ref.
18409 (vect_create_cond_for_align_checks): Likewise.
18410 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18411 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18412 (vect_recog_mask_conversion_pattern): Likewise.
18413 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18414 (new_stmt_vec_info): Remove redundant zeroing.
18416 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18418 * common/config/arm/arm-common.c (arm_be8_option): New function.
18419 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18420 (ISA_ARMv6): Add isa_bit_be8.
18421 * config/arm/arm.h (arm_be8_option): Add prototype.
18422 (BE8_SPEC_FUNCTION): New define.
18423 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18424 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18425 (mlittle-endian): Similarly.
18426 (mbe8, mbe32): New options.
18427 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18428 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18430 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18432 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18434 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18436 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18437 (cleanup_tree_cfg_bb): Use it.
18438 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18440 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18442 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18445 * loop-doloop.c (add_test): Update profile.
18447 2017-07-03 Martin Liska <mliska@suse.cz>
18450 * sanopt.c (rewrite_usage_of_param): New function.
18451 (sanitize_rewrite_addressable_params): Likewise.
18452 (pass_sanopt::execute): Call rewrite_usage_of_param.
18454 2017-07-03 Richard Biener <rguenther@suse.de>
18456 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18457 back to using VIEW_CONVERT_EXPR.
18459 2017-07-03 Martin Liska <mliska@suse.cz>
18462 * doc/extend.texi: Document when a resolver function is
18463 generated for target_clones.
18465 2017-07-03 Martin Liska <mliska@suse.cz>
18467 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18468 and shadow memory of auto variables which are subject of
18469 use-after-scope sanitization.
18470 (asan_expand_mark_ifn): Add do set only when is_poison.
18472 2016-07-03 Richard Biener <rguenther@suse.de>
18474 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18476 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18477 (vectorizable_reduction): Perform reduction PHI creation when
18478 visiting a reduction PHI and adjust and simplify code generation
18479 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18480 (vect_transform_loop): Visit reduction PHIs.
18481 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18482 defs into the SLP tree.
18483 (vect_build_slp_tree): Reduction defs terminate the recursion.
18484 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18486 (vect_get_vec_defs_for_stmt_copy): Export.
18487 (vect_get_vec_defs): Likewise.
18488 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18490 (vect_get_vec_defs_for_stmt_copy): Declare.
18491 (vect_get_vec_defs): Likewise.
18493 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18495 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18496 parameter with a "loop" parameter and use it instead of the
18497 loop containing DR_STMT. Don't check simple_iv when doing
18498 BB analysis. Describe the two analysis modes in the comment.
18500 2017-07-03 Tom de Vries <tom@codesourcery.com>
18502 PR tree-optimization/69468
18503 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18504 (find_same_succ_bb): Handle ignore_edge_flags.
18506 2017-07-03 Tom de Vries <tom@codesourcery.com>
18508 PR tree-optimization/81192
18509 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18511 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18513 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18515 2017-07-03 Tom de Vries <tom@codesourcery.com>
18517 PR tree-optimization/81192
18518 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18519 BB_SAME_SUCC (bb) == NULL.
18521 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18523 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18526 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18528 * dumpfile.c: Include profile-count.h
18529 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18531 (insert_cond_bb): Update profile.
18532 * tree-cfg.h (insert_cond_bb): Update prototype.
18533 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18534 * tree-dump.c: Do not include tree-cfg.
18536 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18538 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18540 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18542 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18545 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18547 * tree-complex.c (expand_complex_div_wide): update profile.
18549 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18550 Alan Hayward <alan.hayward@arm.com>
18551 David Sherwood <david.sherwood@arm.com>
18553 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18554 (CORETYPES_H): New define.
18555 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18556 (insn-modes-inline.h, s-modes-inline-h): New rules.
18557 (generated_files): Add insn-modes-inline.h.
18558 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18559 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18560 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18561 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18562 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18563 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18564 (build/gencodes.o, build/genconditions.o): Likewise.
18565 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18566 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18567 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18568 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18569 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18570 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18571 * coretypes.h: Include everything up to real.h for generators.
18572 Include insn-modes.h first. Include wide-int-print.h after
18573 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18574 * machmode.h: Don't include insn-modes.h here.
18575 * function-tests.c: Remove includes of signop.h, machmode.h,
18576 double-int.h and wide-int.h.
18578 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18580 * optc-save-gen.awk: Likewise.
18581 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18582 * godump.c: Remove include of wide-int-print.h.
18583 * pretty-print.h: Likewise.
18584 * wide-int-print.cc: Likewise.
18585 * wide-int.cc: Likewise.
18586 * hash-map-tests.c: Remove include of signop.h.
18587 * hash-set-tests.c: Likewise.
18588 * rtl-tests.c: Likewise.
18589 * mkconfig.sh: Remove include of machmode.h.
18590 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18592 (emit_insn_modes_inline_h): ...this new function. Emit the code
18593 into an insn-modes-inline.h header file, adding appropriate
18594 include guards and end comments.
18595 (emit_insn_modes_c_header): Remove include of machmode.h.
18596 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18598 (main): Handle -i flag and call emit_insn_modes_inline_h when
18601 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18603 * tree-ssa-strlen.c (strinfo): Rename the length field to
18604 nonzero_chars. Add a full_string_p field.
18605 (compare_nonzero_chars, zero_length_string_p): New functions.
18606 (get_addr_stridx): Add an offset_out parameter.
18607 Use compare_nonzero_chars.
18608 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18609 (new_strinfo): Update after above changes to strinfo.
18610 (set_endptr_and_length): Set full_string_p.
18611 (get_string_length): Update after above changes to strinfo.
18612 (unshare_strinfo): Update call to new_strinfo.
18613 (maybe_invalidate): Likewise.
18614 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18615 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18616 as a uhwi instead of an shwi. Update after above changes to
18617 strinfo and new_strinfo.
18618 (zero_length_string): Assert that chainsi contains full strings.
18619 Use zero_length_string_p. Update call to new_strinfo.
18620 (adjust_related_strinfos): Update after above changes to strinfo.
18621 Copy full_string_p from origsi.
18622 (adjust_last_stmt): Use zero_length_string_p.
18623 (handle_builtin_strlen): Update after above changes to strinfo and
18624 new_strinfo. Install the lhs as the string length if the previous
18625 entry didn't describe a full string.
18626 (handle_builtin_strchr): Update after above changes to strinfo
18628 (handle_builtin_strcpy): Likewise.
18629 (handle_builtin_strcat): Likewise.
18630 (handle_builtin_malloc): Likewise.
18631 (handle_pointer_plus): Likewise.
18632 (handle_builtin_memcpy): Likewise. Track nonzero characters
18633 that aren't necessarily followed by a nul terminator.
18634 (handle_char_store): Likewise.
18636 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18638 PR tree-optimization/80769
18639 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18640 for malloc and calloc. Document the new invariant that all related
18641 strinfos have delayed lengths or none do.
18642 (verify_related_strinfos): Move earlier in file.
18643 (set_endptr_and_length): New function, split out from...
18644 (get_string_length): ...here. Also set the lengths of related
18646 (zero_length_string): Assert that chainsi has known (rather than
18648 (adjust_related_strinfos): Likewise.
18650 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18652 PR tree-optimization/81136
18653 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18654 assert that two references with the same misalignment have the same
18655 compile-time misalignment if those compile-time misalignments
18658 2017-07-01 Andi Kleen <ak@linux.intel.com>
18660 * print-tree.c (print_node): Print all attributes.
18662 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18664 * cfg.c (scale_bbs_frequencies): New function.
18665 * cfg.h (scale_bbs_frequencies): Declare it.
18666 * cfgloopanal.c (single_likely_exit): Cleanup.
18667 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18669 (scale_loop_profile): Likewise.
18670 (loop_version): Likewise.
18671 (create_empty_loop_on_edge): Update.
18672 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18673 scale_loop_frequencies, scale_loop_profile, loopify,
18674 loop_version): Update prototypes.
18675 * modulo-sched.c (sms_schedule): Update.
18676 * predict.c (unlikely_executed_edge_p): Also check probability.
18677 (probably_never_executed_edge_p): Fix typo.
18678 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18679 * tree-parloops.c (gen_parallel_loop): Update.
18680 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18681 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18682 * tree-ssa-loop-split.c (split_loop): Update.
18683 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18684 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18685 (vect_loop_versioning): Update.
18686 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18690 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18692 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18694 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18695 to keep profile consistent.
18697 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18699 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18700 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18701 * profile-count.h (max_safe_multiplier): Make unsigned.
18702 (profile_count::guessed_zero): New.
18704 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18706 * bb-reorder.c (fix_up_crossing_landing_pad,
18707 fix_crossing_conditional_branches): Use make_single_succ_edge
18708 to keep profile consistent.
18710 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18712 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18715 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18718 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18719 the right scopes, make sure cond_jump isn't preserved between multiple
18720 iterations. Search for fallthru edge whenever there are 3+ edges and
18721 use find_fallthru_edge for it.
18723 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18725 Patch by Alexander Monakov <amonakov@ispras.ru>
18726 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18727 probabilities consistently.
18729 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18731 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18732 emit_cmp_and_jump_insns.
18734 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18737 * tree-inline.c (expand_call_inline): Combine profile statuses.
18739 2017-06-30 Andrew Pinski <apinski@cavium.com>
18741 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18742 fold_stmt returned true.
18744 2017-06-30 Nathan Sidwell <nathan@acm.org>
18746 * ggc.h (empty_string): Delete.
18747 * cfgexpand.c (expand_asm_stmt): Use plain "".
18748 * optabs.c (expand_asm_memory_barrier): Likewise.
18749 * stringpool.c (empty_string): Delete.
18750 (digit_vector, digit_string): Delete.
18751 (ggc_alloc_string): Use plain "", don't optimize single digit
18752 strings. Use ggc_alloc_atomic.
18754 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18756 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18757 comparison set and one other set, use the cost of the non-comparison
18760 2017-06-30 Nathan Sidwell <nathan@acm.org>
18762 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18765 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18767 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18768 loops. Remove now unneeded calls to gimple_switch_set_label() that
18769 just set removed labels to NULL_TREE.
18771 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18773 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18774 (set_range_info): ...here. Only call set_range_info_raw if domain
18776 (set_nonzero_bits): Call set_range_info_raw.
18777 * tree-ssanames.h (set_range_info_raw): New.
18779 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18782 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18783 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18784 of nonimmediate_operand and <store_mask_constraint> instead of m
18785 for the input operand. For V8FI iterator, always split if input
18786 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18787 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18788 <store_mask_predicate> instead of register_operand and
18789 <store_mask_constraint> instead of v for the input operand. Make
18790 sure both operands aren't MEMs for if not <mask_applied>.
18792 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18794 * lto-wrapper.c (copy_file) Close both file descriptors before
18797 2017-06-30 Martin Liska <mliska@suse.cz>
18800 * multiple_target.c (create_dispatcher_calls): Make ifunc
18801 also for function that don't have calls or are not referenced.
18803 2017-06-30 Richard Biener <rguenther@suse.de>
18805 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18806 analyze the first scalar stmt. Move vector type computation
18807 for the BB case here from ...
18808 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18809 live operation processing in the SLP case properly.
18811 2017-06-30 Richard Biener <rguenther@suse.de>
18813 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18815 2017-06-30 Martin Liska <mliska@suse.cz>
18818 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18819 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18821 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18823 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18826 2017-06-30 Richard Biener <rguenther@suse.de>
18828 PR tree-optimization/81249
18829 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18830 condition reduction result to original scalar type.
18832 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18834 * profile-count.h (enum profile_quality): Fix typos and whitespace
18837 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18839 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18840 type for branch probabilities.
18842 2017-06-29 Julian Brown <julian@codesourcery.com>
18843 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18845 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18846 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18847 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18848 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18850 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18852 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18853 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18854 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18855 CC usage from generic code to here.
18856 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18857 CC usage into the target macros.
18859 2017-06-29 Maya Rashish <coypu@sdf.org>
18861 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18864 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18866 * arm/arm-builtins.c: Include profile-count.h
18867 * except.c (sjlj_emit_function_enter): Use
18868 profile_probability::unlikely.
18870 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18872 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18873 and tocrel_offset be pointer args rather than implicitly using
18875 (legitimate_constant_pool_address_p, rs6000_emit_move,
18876 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18877 tocrel_offset and use in toc_relative_expr_p call.
18878 (print_operand, print_operand_address): Use static tocrel_base_oac
18879 and tocrel_offset_oac.
18880 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18883 2017-06-29 Maya Rashish <coypu@sdf.org>
18885 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18887 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18889 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18890 objects, take into account only the alignment of 'op0' and 'mode1' if
18893 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18895 * ccmp.c (ccmp_tree_comparison_p): New function.
18896 (ccmp_candidate_p): Update to use above function.
18897 (get_compare_parts): New function.
18898 (expand_ccmp_next): Update to use new functions.
18899 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18901 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18902 take mode as argument.
18903 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18904 * expr.c (expand_expr_real_1): Pass mode as argument.
18906 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18908 * combine.c (combine_instructions): Print insns to dump_file, together
18911 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18913 * asan.c (asan_emit_stack_protection): Update.
18914 (create_cond_insert_point): Update.
18915 * auto-profile.c (afdo_propagate_circuit): Update.
18916 * basic-block.h (struct edge_def): Turn probability to
18917 profile_probability.
18918 (EDGE_FREQUENCY): Update.
18919 * bb-reorder.c (find_traces_1_round): Update.
18920 (better_edge_p): Update.
18921 (sanitize_hot_paths): Update.
18922 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18923 (make_single_succ_edge): Update.
18924 (check_bb_profile): Update.
18925 (dump_edge_info): Update.
18926 (update_bb_profile_for_threading): Update.
18927 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18928 probabilitycount to 0.
18929 * cfgbuild.c (compute_outgoing_frequencies): Update.
18930 * cfgcleanup.c (try_forward_edges): Update.
18931 (outgoing_edges_match): Update.
18932 (try_crossjump_to_edge): Update.
18933 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18934 (expand_gimple_tailcall): Update.
18935 (construct_init_block): Use make_single_succ_edge.
18936 (construct_exit_block): Use make_single_succ_edge.
18937 * cfghooks.c (verify_flow_info): Update.
18938 (redirect_edge_succ_nodup): Update.
18939 (split_edge): Update.
18940 (account_profile_record): Update.
18941 * cfgloopanal.c (single_likely_exit): Update.
18942 * cfgloopmanip.c (scale_loop_profile): Update.
18943 (set_zero_probability): Remove.
18944 (duplicate_loop_to_header_edge): Update.
18945 * cfgloopmanip.h (loop_version): Update prototype.
18946 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18947 (force_nonfallthru_and_redirect): Update.
18948 (update_br_prob_note): Update.
18949 (rtl_verify_edges): Update.
18950 (purge_dead_edges): Update.
18951 (rtl_lv_add_condition_to_bb): Update.
18952 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18953 * cgraphunit.c (init_lowered_empty_function): Update.
18954 (cgraph_node::expand_thunk): Update.
18955 * cilk-common.c: Include profile-count.h
18956 * dojump.c (inv): Remove.
18957 (jumpifnot): Update.
18958 (jumpifnot_1): Update.
18959 (do_jump_1): Update.
18961 (do_jump_by_parts_greater_rtx): Update.
18962 (do_compare_rtx_and_jump): Update.
18963 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18964 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18965 * dwarf2cfi.c: Include profile-count.h
18966 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18967 (sjlj_emit_dispatch_table): Likewise.
18968 * explow.c: Include profile-count.h
18969 * expmed.c (emit_store_flag_force): Update.
18970 (do_cmp_and_jump): Update.
18971 * expr.c (compare_by_pieces_d::generate): Update.
18972 (compare_by_pieces_d::finish_mode): Update.
18973 (emit_block_move_via_loop): Update.
18974 (store_expr_with_bounds): Update.
18975 (store_constructor): Update.
18976 (expand_expr_real_2): Update.
18977 (expand_expr_real_1): Update.
18978 * expr.h (try_casesi, try_tablejump): Update prototypes.
18979 * gimple-pretty-print.c (dump_probability): Update.
18980 (dump_profile): New.
18981 (dump_gimple_label): Update.
18982 (dump_gimple_bb_header): Update.
18983 * graph.c (draw_cfg_node_succ_edges): Update.
18984 * hsa-gen.c (convert_switch_statements): Update.
18985 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18986 (find_if_case_1): Update.
18987 (find_if_case_2): Update.
18988 * internal-fn.c (expand_arith_overflow_result_store): Update.
18989 (expand_addsub_overflow): Update.
18990 (expand_neg_overflow): Update.
18991 (expand_mul_overflow): Update.
18992 (expand_vector_ubsan_overflow): Update.
18993 * ipa-cp.c (good_cloning_opportunity_p): Update.
18994 * ipa-split.c (split_function): Use make_single_succ_edge.
18995 * ipa-utils.c (ipa_merge_profiles): Update.
18996 * loop-doloop.c (add_test): Update.
18997 (doloop_modify): Update.
18998 * loop-unroll.c (compare_and_jump_seq): Update.
18999 (unroll_loop_runtime_iterations): Update.
19000 * lra-constraints.c (lra_inheritance): Update.
19001 * lto-streamer-in.c (input_cfg): Update.
19002 * lto-streamer-out.c (output_cfg): Update.
19003 * mcf.c (adjust_cfg_counts): Update.
19004 * modulo-sched.c (sms_schedule): Update.
19005 * omp-expand.c (expand_omp_for_init_counts): Update.
19006 (extract_omp_for_update_vars): Update.
19007 (expand_omp_ordered_sink): Update.
19008 (expand_omp_for_ordered_loops): Update.
19009 (expand_omp_for_generic): Update.
19010 (expand_omp_for_static_nochunk): Update.
19011 (expand_omp_for_static_chunk): Update.
19012 (expand_cilk_for): Update.
19013 (expand_omp_simd): Update.
19014 (expand_omp_taskloop_for_outer): Update.
19015 (expand_omp_taskloop_for_inner): Update.
19016 * omp-simd-clone.c (simd_clone_adjust): Update.
19017 * optabs.c (expand_doubleword_shift): Update.
19018 (expand_abs): Update.
19019 (emit_cmp_and_jump_insn_1): Update.
19020 (expand_compare_and_swap_loop): Update.
19021 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19022 * predict.c (predictable_edge_p): Update.
19023 (edge_probability_reliable_p): Update.
19024 (set_even_probabilities): Update.
19025 (combine_predictions_for_insn): Update.
19026 (combine_predictions_for_bb): Update.
19027 (propagate_freq): Update.
19028 (estimate_bb_frequencies): Update.
19029 (force_edge_cold): Update.
19030 * profile-count.c (profile_count::dump): Add missing space into dump.
19031 (profile_count::debug): Add newline.
19032 (profile_count::differs_from_p): Explicitly convert to unsigned.
19033 (profile_count::stream_in): Update.
19034 (profile_probability::dump): New member function.
19035 (profile_probability::debug): New member function.
19036 (profile_probability::differs_from_p): New member function.
19037 (profile_probability::differs_lot_from_p): New member function.
19038 (profile_probability::stream_in): New member function.
19039 (profile_probability::stream_out): New member function.
19040 * profile-count.h (profile_count_quality): Rename to ...
19041 (profile_quality): ... this one.
19042 (profile_probability): New.
19043 (profile_count): Update.
19044 * profile.c (compute_branch_probabilities): Update.
19045 * recog.c (peep2_attempt): Update.
19046 * sched-ebb.c (schedule_ebbs): Update.
19047 * sched-rgn.c (find_single_block_region): Update.
19048 (compute_dom_prob_ps): Update.
19049 (schedule_region): Update.
19050 * sel-sched-ir.c (compute_succs_info): Update.
19051 * stmt.c (struct case_node): Update.
19052 (do_jump_if_equal): Update.
19053 (get_outgoing_edge_probs): Update.
19054 (conditional_probability): Update.
19055 (emit_case_dispatch_table): Update.
19056 (expand_case): Update.
19057 (expand_sjlj_dispatch_table): Update.
19058 (emit_case_nodes): Update.
19059 * targhooks.c: Update.
19060 * tracer.c (better_p): Update.
19061 (find_best_successor): Update.
19062 * trans-mem.c (expand_transaction): Update.
19063 * tree-call-cdce.c: Update.
19064 * tree-cfg.c (gimple_split_edge): Upate.
19065 (move_sese_region_to_fn): Upate.
19066 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19067 * tree-eh.c (lower_resx): Upate.
19068 (cleanup_empty_eh_move_lp): Upate.
19069 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19070 * tree-inline.c (copy_edges_for_bb): Update.
19071 (copy_cfg_body): Update.
19072 * tree-parloops.c (gen_parallel_loop): Update.
19073 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19074 (gimple_gen_time_profiler): Update.
19075 * tree-ssa-dce.c (remove_dead_stmt): Update.
19076 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19077 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19078 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19079 (unloop_loops): Update.
19080 (try_peel_loop): Update.
19081 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19082 * tree-ssa-loop-split.c (connect_loops): Update.
19083 (split_loop): Update.
19084 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19085 (hoist_guard): Update.
19086 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19087 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19088 (value_replacement): Update.
19089 * tree-ssa-reassoc.c (branch_fixup): Update.
19090 * tree-ssa-tail-merge.c (replace_block_by): Update.
19091 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19092 (create_edge_and_update_destination_phis): Update.
19093 (compute_path_counts): Update.
19094 (recompute_probabilities): Update.
19095 (update_joiner_offpath_counts): Update.
19096 (freqs_to_counts_path): Update.
19097 (duplicate_thread_path): Update.
19098 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19099 (struct switch_conv_info): Update.
19100 (gen_inbound_check): Update.
19101 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19102 (vect_do_peeling): Update.
19103 (vect_loop_versioning): Update.
19104 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19105 (optimize_mask_stores): Update.
19106 * ubsan.c (ubsan_expand_null_ifn): Update.
19107 * value-prof.c (gimple_divmod_fixed_value): Update.
19108 (gimple_divmod_fixed_value_transform): Update.
19109 (gimple_mod_pow2): Update.
19110 (gimple_mod_pow2_value_transform): Update.
19111 (gimple_mod_subtract): Update.
19112 (gimple_mod_subtract_transform): Update.
19113 (gimple_ic): Update.
19114 (gimple_stringop_fixed_value): Update.
19115 (gimple_stringops_transform): Update.
19116 * value-prof.h: Update.
19118 2017-06-29 Carl Love <cel@us.ibm.com>
19120 * config/rs6000/rs6000-c.c: Add support for built-in functions
19121 vector signed int vec_signed (vector float);
19122 vector signed long long vec_signed (vector double);
19123 vector signed int vec_signed2 (vector double, vector double);
19124 vector signed int vec_signede (vector double);
19125 vector signed int vec_signedo (vector double);
19126 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19127 instruction generator.
19128 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19129 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19130 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19132 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19133 vunsignede_v2df): Add define_expands.
19134 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19135 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19136 VEC_UNSIGNEDO): Add definitions.
19137 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19138 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19139 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19140 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19141 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19142 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19143 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19144 vec_unsignede, vec_unsignedo): Add builtin defines.
19145 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19147 * doc/extend.texi: Update the built-in documentation file for the
19148 new built-in functions.
19150 2017-06-29 Richard Biener <rguenther@suse.de>
19152 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19153 reduction chains to LOOP_VINFO_REDUCTIONS.
19154 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19155 SLP reductions after processing reduction chains.
19157 2017-06-29 Nathan Sidwell <nathan@acm.org>
19159 * builtins.c (fold_builtin_FUNCTION): Use
19160 lang_hooks.decl_printable_name.
19162 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19164 PR middle-end/81194
19165 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19166 with only one label.
19167 * stmt.c (expand_case): Assert NCASES is greater than one.
19169 2017-06-29 Richard Biener <rguenther@suse.de>
19171 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19173 (group_case_labels): Likewise.
19174 (find_taken_edge): Push sanity checking on val to workers...
19175 (find_taken_edge_cond_expr): ... here
19176 (find_taken_edge_switch_expr): ... and here, handle cases
19177 with just a default label.
19178 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19179 (group_case_labels): Likewise.
19180 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19181 group_case_labels does anything cleanup the CFG again.
19183 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19185 PR tree-optimization/81196
19186 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19187 exit condition comparing two IVs.
19189 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19191 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19192 profile to the dummy entry at the end of the list of architectures.
19193 * config/arm/arm-cpu-cdata.h: Regenerated.
19195 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19196 Michael Collison <michael.collison@arm.com>
19199 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19201 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19202 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19203 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19204 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19205 of shift when the shift amount is masked with constant equal to
19206 the size of the mode.
19207 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19210 2017-06-29 Martin Liska <mliska@suse.cz>
19212 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19214 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19217 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19220 (edge_growth_cache_entry::edge_growth_cache_entry): New
19222 (reset_edge_growth_cache): Update to use constructor.
19224 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19226 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19227 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19228 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19230 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19232 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19233 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19235 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19237 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19238 (*-linux-uclibc*): Add t-uclibc tmake_file.
19239 * config/t-musl: New.
19240 * config/t-uclibc: New.
19242 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19244 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19246 (gen_comm_data): Emit architectural setting of arch_prof.
19247 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19249 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19250 (armv8-m.base, armv8-m.main): Likewise.
19251 * arm-protos.h (arm_build_target): Add profile field.
19252 (arch_option): Likewise.
19253 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19255 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19256 arm_active_target.profile.
19258 2017-06-28 Richard Biener <rguenther@suse.de>
19260 PR middle-end/81227
19261 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19262 TYPE_OVERFLOW_WRAPS.
19263 * match.pd (negate_expr_p): Likewise.
19264 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19265 fold_build2, not fold_binary.
19267 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19269 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19270 Convert memory address to Pmode.
19271 (aarch64_print_operand): Assert MEM operands are always Pmode.
19273 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19276 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19277 Remove redundant if.
19278 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19279 * config/arm/aarch-common-protos.h
19280 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19281 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19283 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19286 * multiple_target.c (create_dispatcher_calls): Set the default
19287 clone to be static, not public.
19289 2017-06-28 Richard Biener <rguenther@suse.de>
19291 * tree-vect-loop.c (vectorizable_reduction): Move special
19292 cond reduction IV var creation ...
19293 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19294 parameter. Use STMT_VINFO_VECTYPE.
19295 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19298 2017-06-28 Martin Liska <mliska@suse.cz>
19301 * ipa-visibility.c (non_local_p): Handle visibility.
19303 2017-06-28 Martin Liska <mliska@suse.cz>
19306 * common.opt: Add IntegerRange to various options.
19307 * opt-functions.awk (integer_range_info): New function.
19308 * optc-gen.awk: Add integer_range_info to cl_options struct.
19309 * opts-common.c (decode_cmdline_option): Handle
19310 CL_ERR_INT_RANGE_ARG.
19311 (cmdline_handle_error): Likewise.
19312 * opts.c (print_filtered_help): Show valid interval in
19313 when --help is provided.
19314 * opts.h (struct cl_option): Add range_min and range_max fields.
19315 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19317 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19319 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19320 (x * C EQ/NE y * C): New transformation.
19322 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19324 * genmultilib (combination_space): Accept '+' in option names.
19326 2017-06-28 Martin Liska <mliska@suse.cz>
19329 * asan.c (instrument_derefs): Bail out inner references
19330 that are hard register variables.
19332 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19335 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19336 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19338 2017-06-28 Richard Biener <rguenther@suse.de>
19340 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19341 (vect_get_slp_defs): Adjust.
19342 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19344 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19346 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19347 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19348 (vectorizable_reduction): Adjust.
19349 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19351 (vect_get_slp_defs): Likewise.
19352 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19353 (vectorizable_bswap): Adjust.
19354 (vectorizable_call): Likewise.
19355 (vectorizable_conversion): Likewise.
19356 (vectorizable_assignment): Likewise.
19357 (vectorizable_shift): Likewise.
19358 (vectorizable_operation): Likewise.
19359 (vectorizable_store): Likewise.
19360 (vectorizable_condition): Likewise.
19361 (vectorizable_comparison): Likewise.
19363 2017-06-28 Michael Collison <michael.collison@arm.com>
19366 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19368 (arm_gen_movmemqi): Removed unused variable 'i'.
19369 Convert 'for' loop into 'while' loop.
19370 (arm_expand_prologue): Remove last unnecessary set of insn.
19371 (thumb_pop): Remove unused variable 'pushed_words'.
19372 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19374 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19376 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19377 * config/s390/s390-protos.h: Add prototype of
19378 s390_rel_address_ok_p.
19379 * config/s390/s390.c (s390_got_symbol): New function.
19380 (s390_rel_address_ok_p): New function.
19381 (legitimize_pic_address): Use s390_rel_address_ok_p.
19382 (s390_load_got): Use s390_got_symbol.
19383 (s390_option_override): Issue error if
19384 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19385 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19387 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19389 2017-06-27 Andrew Pinski <apinski@cavium.com>
19391 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19392 (X * copysign (1.0, X)): New pattern.
19393 (X * copysign (1.0, -X)): New pattern.
19394 (copysign (-1.0, CST)): New pattern.
19396 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19398 * genmultilib (combination_space): Remove variable.
19399 Validate reuse rules against regular expression for any sequence
19400 of multilib options in any order.
19402 2017-06-27 Michael Collison <michael.collison@arm.com>
19404 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19405 call aarch64_split_simd_combine.
19406 * (aarch64_combine_internal<mode>): Delete pattern.
19407 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19408 Allow register and subreg operands.
19410 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19412 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19413 specific need, just fallback on defaults.
19414 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19416 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19417 Olivier Hainque <hainque@adacore.com>
19419 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19421 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19422 targets. Pick a default if no particular attempt applied.
19423 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19426 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19428 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19429 (x86_64-wrs-vxworks7): Likewise.
19431 2017-06-27 Marek Polacek <polacek@redhat.com>
19434 * ubsan.c (instrument_null): Check get_base_address's result for null.
19436 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19438 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19440 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19442 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19443 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19444 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19445 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19446 New function types.
19447 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19448 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19449 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19450 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19451 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19452 BUILT_IN_FEUPDATEENV): New builtins.
19453 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19454 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19455 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19456 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19458 (builtin_structptr_types): Adjust size.
19459 * tree.c (builtin_structptr_types): Add four entries.
19461 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19462 Olivier Hainque <hainque@adacore.com>
19464 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19465 (TLS_SYM): New local macro, forcing reference to __tls__ on
19466 link command lines for VxWorks 7 RTPs, triggering initialization
19468 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19469 OS features TLS support, true for RTPs on VxWorks 7.
19470 * config/vxworks.c (vxworks_override_options): Setup emutls
19473 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19475 * predict.c (test_prediction_value_range): Use -1U instead of -1
19476 to avoid narrowing conversion warning.
19477 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19478 to avoid narrowing conversion warning.
19479 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19481 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19483 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19485 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19486 64bit configurations.
19487 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19488 (SIZE_TYPE): Likewise.
19489 * config/vxworks.c (vxworks_emutls_var_fields): Use
19490 long_unsigned_type_node instead of unsigned_type_node as the offset
19491 field type, which is "pointer" mode in emutls.c.
19493 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19496 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19498 PR middle-end/81207
19499 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19500 gimple_vuse copying separately from gimple_vdef copying.
19502 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19504 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19507 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19508 Olivier Hainque <hainque@adacore.com>
19510 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19511 for all vxworks7 targets.
19512 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19513 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19514 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19515 variations for VX6/VX7 and 32/64bits later on in ...
19516 (VXWORKS_LIB_SPEC): Leverage new macros.
19517 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19518 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19520 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19522 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19523 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19525 2017-06-26 Carl Love <cel@us.ibm.com>
19527 * config/rs6000/rs6000-c.c: Add support for built-in functions
19528 vector bool char vec_reve (vector bool char);
19529 vector signed char vec_reve (vector signed char);
19530 vector unsigned char vec_reve (vector unsigned char);
19531 vector bool int vec_reve (vector bool int);
19532 vector signed int vec_reve (vector signed int);
19533 vector unsigned int vec_reve (vector unsigned int);
19534 vector bool long long vec_reve (vector bool long long);
19535 vector signed long long vec_reve (vector signed long long);
19536 vector unsigned long long vec_reve (vector unsigned long long);
19537 vector bool short vec_reve (vector bool short);
19538 vector signed short vec_reve (vector signed short);
19539 vector double vec_reve (vector double);
19540 vector float vec_reve (vector float);
19541 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19542 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19543 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19544 (altivec_vreve): New pattern.
19545 * config/rs6000/altivec.h (vec_reve): New define.
19546 * doc/extend.texi (vec_rev): Update the built-in documentation file
19547 for the new built-in functions.
19549 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19551 PR tree-optimization/71815
19552 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19554 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19556 (slsr_process_phi): Likewise.
19557 (replace_uncond_cands_and_profitable_phis): Don't replace a
19558 multiply candidate with a stride of 1 (copy or cast).
19559 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19561 (lowest_cost_path): Likewise.
19562 (total_savings): Likewise.
19564 2017-06-26 Richard Biener <rguenther@suse.de>
19567 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19568 Use def_builtin_pure for all gather builtins.
19570 2017-06-26 Richard Biener <rguenther@suse.de>
19572 PR tree-optimization/81203
19573 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19574 non-dominating BBs.
19576 2017-06-26 Marek Polacek <polacek@redhat.com>
19579 * doc/invoke.texi: Document -Wmultistatement-macros.
19581 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19583 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19584 arm_neon_ok_no_float_abi effective target.
19586 2017-06-26 Richard Biener <rguenther@suse.de>
19588 PR tree-optimization/80928
19589 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19590 (copy_bbs): Set BB_DUPLICATED flag early.
19591 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19593 (execute_on_shrinking_pred): Likewise.
19594 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19595 BB_DUPLICATED blocks.
19596 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19597 iterate over all PHIs considering removal of *gsi.
19599 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19601 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19604 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19606 * config/rs6000/rs6000-string.c: (expand_block_clear,
19607 do_load_for_compare, select_block_compare_mode,
19608 compute_current_alignment, expand_block_compare,
19609 expand_strncmp_align_check, expand_strn_compare,
19610 expand_block_move, rs6000_output_load_multiple)
19611 Move functions related to string/block move/compare
19612 to a separate file.
19613 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19614 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19615 for this function which is now used in two files.
19616 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19617 * config.gcc: Add rs6000-string.o to extra_objs for
19618 targets powerpc*-*-* and rs6000*-*-*.
19620 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19623 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19624 32-bit, since indexed is not valid for DImode.
19625 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19626 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19627 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19628 (define_peephole2 for Altivec d-form store): Likewise.
19631 * multiple_target.c (create_dispatcher_calls): Only create the
19632 dispatcher call if the function is the default clone of a
19633 versioned function.
19635 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19637 PR middle-end/80902
19638 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19639 a call, force the call to not be a tail call.
19641 2017-06-23 Jeff Law <law@redhat.com>
19643 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19646 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19648 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19649 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19650 handling for early expansion of vector shifts (sl,sr,sra,rl).
19651 (builtin_function_type): Add vector shift right instructions
19652 to the unsigned argument list.
19654 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19656 rtl-optimizatoin/79286
19657 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19658 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19659 trap. PIC register plus a const unspec without offset can never trap.
19661 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19663 * tree.h (builtin_structptr_type): New type.
19664 (builtin_structptr_types): Declare new array.
19665 * tree.c (builtin_structptr_types): New array.
19666 (free_lang_data, build_common_tree_nodes): Use it.
19668 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19671 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19674 2017-06-22 Matt Turner <mattst88@gmail.com>
19676 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19677 Lake models to skylake case. Assume skylake for unknown
19678 models with clflushopt.
19680 2017-06-22 Jeff Law <law@redhat.com>
19682 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19683 frame sizes that do not satisfy aarch64_uimm12_shift.
19685 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19687 * profile-count.h (apply_probability,
19688 apply_scale, probability_in): Fix checks for zero.
19690 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19692 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19693 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19695 2016-06-22 Richard Biener <rguenther@suse.de>
19697 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19698 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19699 REDUC_MAX_EXPR support.
19700 (vectorizable_reduction): Likewise.
19701 (vect_create_epilog_for_reduction): Likewise.
19703 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19705 * match.pd (A / (1 << B) -> A >> B): New.
19706 * generic-match-head.c: Include optabs-tree.h.
19707 * gimple-match-head.c: Likewise.
19708 * optabs-tree.h (target_supports_op_p): New.
19709 * optabs-tree.c (target_supports_op_p): New.
19711 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19713 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19714 $gcc_cv_ld --help output.
19715 (gcc_cv_ld_demangle): Likewise.
19716 (gcc_cv_ld_eh_frame_hdr): Likewise.
19717 (gcc_cv_ld_pie): Likewise.
19718 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19719 (gcc_cv_ld_buildid): Likewise.
19720 (gcc_cv_ld_sysroot): Likewise.
19721 (ld_bndplt_support): Likewise.
19722 (ld_pushpopstate_support): Likewise.
19723 * configure: Regenerate.
19724 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19726 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19729 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19730 operands indexes to avoid gap between operands and match_dups.
19732 2017-06-21 Andrew Pinski <apinski@cavium.com>
19734 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19735 Increment Arith_shift and Arith_shift_reg by 1.
19736 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19738 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19739 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19740 (aarch64_strip_extend): Add new argument and test for it.
19741 (aarch64_cheap_mult_shift_p): New function.
19742 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19743 add a cost if it is true.
19744 Update calls to aarch64_strip_extend.
19745 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19747 2017-06-21 Andrew Pinski <apinski@cavium.com>
19749 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19751 (thunderxt88): Likewise.
19752 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19753 (thunderx_prefetch_tune): New variable.
19754 (thunderx2t99_prefetch_tune): Update for the correct values.
19755 (thunderxt88_tunings): New variable.
19756 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19757 generic_prefetch_tune.
19758 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19760 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19762 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19763 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19764 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19765 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19766 (aarch64_atomic_cas<mode>, GPI): Likewise.
19768 2017-06-21 Martin Liska <mliska@suse.cz>
19770 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19771 statements on cold and hot labels.
19772 * predict.c (tree_estimate_probability_bb): Remove the
19773 prediction from this place.
19775 2017-06-21 Martin Liska <mliska@suse.cz>
19777 PR tree-optimization/79489
19778 * gimplify.c (maybe_add_early_return_predict_stmt): New
19780 (gimplify_return_expr): Call the function.
19781 * predict.c (tree_estimate_probability_bb): Remove handling
19783 * predict.def: Update comment about early return predictor.
19784 * gimple-predict.h (is_gimple_predict): New function.
19785 * predict.def: Change default value of early return to 66.
19786 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19788 * passes.def: Put pass_strip_predict_hints to the beginning of
19791 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19793 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19794 FUNCTION_DECL declarations.
19795 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19797 (dwaf2out_decl): Likewise.
19798 * godump.c (go_early_global_decl): Skip call to the real debug hook
19799 for FUNCTION_DECL declarations.
19800 * passes.c (rest_of_decl_compilation): Skip call to the
19801 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19802 -fdump-go-spec is passed.
19804 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19806 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19807 Reorder for compactness.
19808 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19809 (def_builtin_pure, def_builtin_pure2): New functions.
19810 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19812 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19814 * match.pd (nop_convert): New predicate.
19815 ((A +- CST1) +- CST2): Allow some NOP conversions.
19817 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19820 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19821 with ctors/dtors if GOVD_SHARED is set.
19823 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19825 * config/aarch64/aarch64.md (movti_aarch64):
19826 Emit mov rather than orr.
19827 (movtf_aarch64): Likewise.
19828 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19829 Emit mov rather than orr.
19831 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19833 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19834 Swap alternatives, make integer dup more expensive.
19836 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19838 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19839 Return true for non-tls symbols.
19841 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19843 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19844 (cortex-a75): Likewise.
19845 (cortex-a75.cortex-a55): Likewise.
19846 * config/aarch64/aarch64-tune.md: Regenerate.
19847 * doc/invoke.texi (-mtune): Document new values for -mtune.
19849 2017-06-21 Tom de Vries <tom@codesourcery.com>
19851 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19852 stack_size feature.
19853 (Effective-Target Keywords, Other attributes): Suggest using
19854 dg-add-options stack_size feature to get stack limit in stack_size
19855 effective target documentation.
19857 2017-06-21 Julian Brown <julian@codesourcery.com>
19858 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19860 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19861 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19862 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19864 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19865 attribute type list for neon_multiply.
19866 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19867 attribute type list for neon_multiply.
19868 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19869 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19870 attribute type list for neon_multiply.
19871 * config/arm/types.md (crypto_pmull): Add.
19872 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19873 attribute type list.
19875 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19877 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19880 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19882 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19883 to make sure not to dereference a NULL cost_classes_ptr pointer.
19885 2017-06-20 Carl Love <cel@us.ibm.com>
19887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19888 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19889 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19890 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19891 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19892 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19893 VMULOSW): New enum "unspec" values.
19894 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19895 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19896 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19897 altivec_vmulosw): New patterns.
19898 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19899 VMULOSW): Add definitions.
19901 2017-06-20 Julia Koval <julia.koval@intel.com>
19903 * config/i386/i386.c: Fix rounding expand for new pattern.
19904 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19906 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19908 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19909 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19911 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19913 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19916 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19918 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19919 architecture, then by implementer ID.
19920 * config/aarch64/aarch64-tune.md: Regenerate.
19922 2017-06-20 Richard Biener <rguenther@suse.de>
19924 PR middle-end/81097
19925 * fold-const.c (split_tree): Fold to type before negating.
19927 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19929 * diagnostic-show-locus.c
19930 (selftest::test_fixit_deletion_affecting_newline): New function.
19931 (selftest::diagnostic_show_locus_c_tests): Call it.
19933 2017-06-20 Andreas Schwab <schwab@suse.de>
19936 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19939 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19941 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19942 __ARM_FEATURE_COPROC according to support.
19944 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19946 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19947 Rewritten to avoid overflow for > 32-bit pointers.
19950 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19951 by removing enum keyword.
19952 (ubsan_type_descriptor): Likewise. Formatting fix.
19955 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19956 splitter): Require TARGET_SSE2 in the condition.
19958 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19961 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19962 for doing vector set of SFmode on ISA 3.0.
19963 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19964 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19966 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19967 SFmode value into a V4SF variable that was extracted from another
19968 V4SF variable without converting the element to double precision
19969 and back to single precision vector format.
19970 (vsx_insert_extract_v4sf_p9_2): Likewise.
19972 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19974 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19975 in UWHI to avoid undefined overflow.
19978 * ubsan.h (enum ubsan_encode_value_phase): New.
19979 (ubsan_encode_value): Change second argument to
19980 enum ubsan_encode_value_phase with default value of
19981 UBSAN_ENCODE_VALUE_GENERIC.
19982 * ubsan.c (ubsan_encode_value): Change second argument to
19983 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19984 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19985 create_tmp_var_raw instead of create_tmp_var and use a
19987 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19988 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19989 ubsan_encode_value callers.
19992 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19993 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19994 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19996 2017-06-19 Richard Biener <rguenther@suse.de>
19998 PR middle-end/81118
19999 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20000 estimates if we changed anything.
20002 2017-06-19 Richard Biener <rguenther@suse.de>
20004 PR tree-optimization/80887
20005 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20006 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20007 simplified lookups, then reset mprts_hook.
20008 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20010 (try_to_simplify): Likewise.
20012 2017-06-19 Martin Liska <mliska@suse.cz>
20015 * gimplify.c (gimplify_switch_expr):
20016 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20018 2017-06-19 Martin Liska <mliska@suse.cz>
20020 * doc/install.texi: Document that PGO runs in 4 stages.
20022 2017-06-19 Martin Liska <mliska@suse.cz>
20025 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20026 to dispatcher function name.
20027 * multiple_target.c (replace_function_decl): New function.
20028 (create_dispatcher_calls): Redirect both edges and references.
20030 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20032 * profile-count.c (profile_count::dump): Dump quality.
20033 (profile_count::differs_from_p): Update for unsigned val.
20034 * profile-count.h (profile_count_quality): New enum.
20035 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20037 2017-06-19 Richard Biener <rguenther@suse.de>
20039 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20040 struct function as arg.
20041 (estimate_numbers_of_iterations): Export overload with loop arg.
20042 (free_numbers_of_iterations_estimates_loop): Use an overload of
20043 free_numbers_of_iterations_estimates instead.
20044 * tree-cfg.c (remove_bb): Adjust.
20045 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20046 * tree-parloops.c (gen_parallel_loop): Likewise.
20047 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20049 (tree_unroll_loops_completely): Likewise.
20050 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20051 Use an overload instead and export.
20052 (estimated_loop_iterations): Adjust.
20053 (max_loop_iterations): Likewise.
20054 (likely_max_loop_iterations): Likewise.
20055 (estimate_numbers_of_iterations): Take struct function as arg
20057 (loop_exits_before_overflow): Adjust.
20058 (free_numbers_of_iterations_estimates_loop): Use an overload.
20059 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20060 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20062 2017-06-19 Richard Biener <rguenther@suse.de>
20065 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20066 RANGE_EXPR conservatively.
20068 2017-06-16 Carl Love <cel@us.ibm.com>
20070 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20071 definitions for vec_float, vec_float2, vec_floato,
20072 vec_floate built-ins.
20073 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20074 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20076 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20077 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20078 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20079 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20080 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20081 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20082 vec_floato): Add builtin defines.
20083 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20084 Update the built-in documentation file for the new built-in
20087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20089 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20090 (mthumb): Mark as the negative of -marm.
20092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20094 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20096 (ARM Options, -mtune): Document that this accepts the same
20097 extension options as -mcpu.
20098 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20102 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20103 permitted extensions.
20105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20107 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20108 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20109 (armv8-m.main): Add option +nodsp.
20110 * config/arm/arm-cpu-cdata.h: Regenerated.
20112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20114 * config/arm/t-fuchsia: New file.
20115 * config.gcc (arm*-*-fuchsia*): Use it.
20117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20119 * config/arm/t-symbian: Rewrite for new option infrastructure.
20121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20123 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20124 (MULTILIB_REQUIRED): Likewise.
20126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20128 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20129 (MULTILIB_RESUE): Likewise.
20130 (MULTILIB_MATCHES): Likewise.
20131 (MULTLIB_REQUIRED): Likewise.
20133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20135 * config/arm/t-rtems: Rewrite for new option framework.
20137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20139 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20140 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20141 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20142 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20143 * config/arm/t-multilib: ... here.
20144 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20145 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20146 armv7-a and armv8*-a when A-profile libraries have not been built.
20147 * config/arm/t-rmprofile: Rewrite.
20149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20151 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20152 with a backslash. Remove the backslash after substituting unescaped
20154 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20158 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20159 the driver to pass through the default setting of -mfloat-abi.
20160 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20162 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20163 (all_feat_combs): New rule.
20164 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20166 * config/arm/t-aprofile: Rewrite.
20168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20170 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20171 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20172 fpu is not specified by the user/command-line.
20173 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20174 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20175 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20176 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20177 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20178 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20182 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20183 * config/arm/t-arm-elf: Rewritten.
20185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20187 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20188 have some floating-point instructions.
20189 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20190 (TARGET_MAYBE_HARD_FLOAT): New macro.
20191 * config/arm/arm-builtins.c (arm_init_builtins): Use
20192 TARGET_MAYBE_HARD_FLOAT.
20193 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20197 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20198 (configargs.h): Include it.
20199 (arm_print_hint_for_fpu_option): New function.
20200 (arm_parse_fpu_option): New function.
20201 (candidate_extension): New class.
20202 (arm_canon_for_multilib): New function.
20203 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20204 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20205 (ARCH_CANONICAL_SPECS): New macro.
20206 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20208 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20210 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20211 are set after handling multilib fragments. Set target_cpu_default2
20214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20216 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20218 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20219 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20221 (check_arch): Likewise.
20222 * config/arm/arm.c (arm_configure_build_target): Handle
20223 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20224 options in the default.
20226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20228 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20229 when an option is an alias of another.
20230 * config/arm/parsecpu.awk (optalias): New parser token.
20231 (gen_comm_data): Mark non-alias options as such. Emit entries
20232 for extension aliases.
20233 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20234 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20235 (armv6kz, armv6zk, armv6t2): Likewise.
20236 (armv7): Make vfpv3-d16 an alias.
20237 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20239 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20240 Sort in canonical order.
20241 (armv8-a): Sort in canonical order.
20242 (armv8.1-a, armv8.2-a): Likewise.
20243 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20245 (cortex-a9): Sort in canonical order.
20246 * config/arm/arm.c (selftests.h): Include it.
20247 (arm_test_cpu_arch_data): New function.
20248 (arm_run_self_tests): New function.
20249 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20250 (targetm): Move declaration to the end of the file.
20251 * arm-cpu-cdata.h: Regenerated.
20253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20255 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20256 call to target_mode_check describing the type of option passed.
20257 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20258 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20259 arm_parse_cpu_option_name to match parameters against list of
20261 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20262 arm_arch_core_flags data structure.
20263 * config/arm/arm-cpu_cdata.h: Regenerated.
20265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20267 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20269 (arm_print_hint_for_cpu_option): Likewise.
20270 (arm_print_hint_for_arch_option): Likewise.
20271 (arm_parse_cpu_option_name): Likewise.
20272 (arm_parse_arch_option_name): Likewise.
20273 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20274 of entries in the all_fpus list.
20275 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20276 (arm_parse_cpu_option_name): Declare.
20277 (arm_parse_arch_option_name): Declare.
20278 (arm_parse_option_features): Declare.
20279 (arm_intialize_isa): Declare.
20280 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20282 (gen_comm_data): ... here. Make definitions non-static.
20283 * config/arm/arm-cpu-data.h: Regenerated.
20284 * config/arm/arm-cpu-cdata.h: Regenerated.
20286 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20288 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20289 (cpu_arch_extension): New structure.
20290 (cpu_arch_option, arch_option, cpu_option): New structures.
20291 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20292 architecture types.
20293 (gen_data): Generate new format data tables.
20294 * config/arm/arm.c (cpu_tune): New structure.
20295 (cpu_option, processors): Delete.
20296 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20297 (arm_print_hint_for_cpu_option): ... this and ...
20298 (arm_print_hint_for_arch_option): ... this.
20299 (arm_parse_arch_cpu_name): Delete. Replace with ...
20300 (arm_parse_cpu_option_name): ... this and ...
20301 (arm_parse_arch_option_name): ... this.
20302 (arm_unrecognized_feature): Change type of target parameter to
20304 (arm_parse_arch_cpu_features): Delete. Replace with ...
20305 (arm_parse_option_features): ... this.
20306 (arm_configure_build_target): Rework to use new configuration data
20308 (arm_print_tune_info): Rework for new configuration data tables.
20309 * config/arm/arm-cpu-data.h: Regenerated.
20310 * config/arm/arm-cpu.h: Regenerated.
20312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20314 * Makefile.in (OBJS): Move sbitmap.o from here ...
20315 (OBJS-libcommon): ... to here.
20317 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20319 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20320 (ISA_ALL_CRYPTO): New macro.
20321 (ISA_ALL_SIMD): New macro
20322 (ISA_ALL_FP): New macro.
20323 * config/arm/arm.c (fpu_bitlist): Update initializer.
20324 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20326 (arm9e): Add fpu. Add option for nofp
20327 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20328 (arm926ej-s, arm1026ej-s): Likewise.
20329 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20330 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20331 neon-fp16, neon-vfpv4, nofp and nosimd.
20332 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20333 (cortex-a8): Add fpu. Add option for nofp.
20334 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20335 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20336 (cortex-r4f): Add fpu.
20337 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20338 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20340 (cortex-r8): Likewise.
20341 (cortex-m4): Add fpu. Add option for nofp.
20342 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20343 (cortex-a17.cortex-a7): Likewise.
20344 (cortex-a32): Add fpu. Add options for crypto and nofp.
20345 (cortex-a35, cortex-a53): Likewise.
20346 (cortex-a57): Add fpu. Add option for crypto.
20347 (cortex-a72, cortex-a73): Likewise.
20348 (exynos-m1): Likewise.
20349 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20350 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20351 (cortex-m33): Add fpu. Add option for nofp.
20352 * config/arm/arm-cpu-cdata.h: Regenerated
20353 * config/arm/arm-cpu-data.h: Regenerated.
20355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20357 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20358 (armv5te, armv5tej): Likewise.
20359 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20360 (armv7): Add options fp and vfpv3-d16.
20361 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20362 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20364 (armv7ve): Likewise.
20365 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20366 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20367 (armv8-a): Add nocrypto option.
20368 (armv8.1-a, armv8.2-a): Likewise.
20369 (armv8-m.main): add options fp, fp.dp and nofp.
20371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20373 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20375 (armv8-a+crc): Delete.
20376 (armv8.1-a): Add options simd, crypto and nofp.
20377 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20378 (armv8.2-a+fp16): Delete.
20379 (armv8-m.main): Add option dsp.
20380 (armv8-m.main+dsp): Delete.
20381 (cortex-a8): Add fpu. Add nofp option.
20382 (cortex-a9): Add fpu. Add nofp and nosimd options.
20383 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20384 link to main cpu and architecture data structures.
20385 (gen_comm_data): Only put isa attributes from the main architecture
20387 (option): New statement for architecture and CPU entries.
20388 * arm.c (struct cpu_option): New structure.
20389 (struct processors): Add entry for options.
20390 (arm_unrecognized_feature): New function.
20391 (arm_parse_arch_cpu_name): Ignore any characters after the first
20393 (arm_parse_arch_cpu_feature): New function.
20394 (arm_configure_build_target): Separate out any CPU and architecture
20395 features and parse separately. Don't error out if -mfpu=auto is
20396 used with only an architecture string.
20397 (arm_print_asm_arch_directives): New function.
20398 (arm_file_start): Call it.
20399 * config/arm/arm-cpu-cdata.h: Regenerated.
20400 * config/arm/arm-cpu-data.h: Likewise.
20401 * config/arm/arm-tables.opt: Likewise.
20403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20405 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20406 assembler when it is not -mfpu=auto.
20408 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20410 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20411 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20412 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20413 (ASM_CPU_SPEC): Rewrite.
20414 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20415 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20416 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20417 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20418 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20419 copied string is NUL-terminated. Also strip any characters prefixed
20421 (arm_rewrite_selected_arch): New function.
20422 (arm_rewrite_march): New function.
20424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20426 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20427 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20428 (march, mcpu, mtune): Convert to string-based options.
20429 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20430 (arm_parse_arch_cpu_name): New function.
20431 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20432 identify selected architecture or CPU.
20433 (arm_option_save): New function.
20434 (TARGET_OPTION_SAVE): Redefine.
20435 (arm_option_restore): Restore string options.
20436 (arm_option_print): Print string options.
20438 2017-06-16 Martin Sebor <msebor@redhat.com>
20440 PR tree-optimization/80933
20441 PR tree-optimization/80934
20442 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20443 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20444 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20445 (gimple_fold_builtin): Call them.
20447 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20449 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20450 as unlikely; update profile.
20452 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20454 * predict.c (force_edge_cold): Handle declaring edges impossible
20457 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20459 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20461 (try_unroll_loop_completely): Fix reporting.
20463 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20465 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20467 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20470 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20471 if given a non-constant argument for an intrinsic which requires a
20474 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20476 * profile.c (compare_freqs): New function.
20477 (branch_prob): Sort edge list.
20478 (find_spanning_tree): Assume that the list is priority sorted.
20480 2017-06-16 Richard Biener <rguenther@suse.de>
20482 PR tree-optimization/81090
20483 * passes.def (pass_record_bounds): Remove.
20484 * tree-pass.h (make_pass_record_bounds): Likewise.
20485 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20486 make_pass_record_bounds): Likewise.
20487 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20488 not free niter estimates at the beginning but at the end.
20489 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20491 2017-06-16 Richard Biener <rguenther@suse.de>
20493 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20494 initializer to workaround ICE in host GCC 4.8.
20496 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20498 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20500 (clone_inlined_nodes): Update.
20502 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20504 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20505 prefetch settings, and enable prefetching by default at -O3.
20507 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20509 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20510 Set flag_prefetch_loop_arrays according to tuning data.
20512 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20514 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20515 New tune structure.
20516 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20517 [Unrelated to main purpose of the patch] Place the pointer field last
20518 to enable type checking errors when tune structure are wrongly merged.
20519 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20520 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20521 (thunderx2t99_prefetch_tune): New tune constants.
20522 (tune_params *_tunings): Update all tunings (no functional change).
20523 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20524 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20525 from tunings structures.
20527 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20530 * ubsan.c (instrument_null): Add T argument, use it instead
20531 of computing it based on IS_LHS.
20532 (instrument_object_size): Likewise.
20533 (pass_ubsan::execute): Adjust instrument_null and
20534 instrument_object_size callers to pass gimple_get_lhs or
20535 gimple_assign_rhs1 result to it. Use instrument_null instead of
20536 calling get_base_address and instrument_mem_ref. Handle
20537 aggregate call arguments for object-size sanitization.
20539 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20541 PR tree-optimization/81089
20542 * tree-vrp.c (is_masked_range_test): Validate operands of
20545 2017-06-15 Martin Sebor <msebor@redhat.com>
20548 * dumpfile.c (dump_register): Avoid calling memset to initialize
20549 a class with a default ctor.
20550 * gcc.c (struct compiler): Remove const qualification.
20551 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20552 * hash-table.h: Ditto.
20553 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20555 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20556 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20558 * params.h (struct param_info): Make struct members non-const.
20559 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20560 with default initialization.
20561 * vec.h (vec_copy_construct, vec_default_construct): New helper
20563 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20564 with vec_copy_construct.
20565 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20566 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20567 * doc/invoke.texi (-Wclass-memaccess): Document.
20569 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20571 * emit-rtl.h (is_leaf): Update comment about local
20572 register allocator.
20574 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20577 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20578 for a variable to have a section before checking if the section has a
20580 Set section to.persistent if persistent attribute is set.
20581 Warn if .persistent attribute is used on an automatic variable.
20583 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20585 PR rtl-optimization/80474
20586 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20588 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20590 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20593 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20595 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20596 (casesi): Emit bounds checking as RTL.
20597 (casesi_internal_mips16_<mode>): Remove bounds checking.
20599 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20601 * config/xtensa/xtensa.c (xtensa_option_override): Append
20602 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20603 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20604 xtensa_doloop_hooks): Define unconditionally.
20605 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20607 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20608 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20609 for it in xtensa_option_override.
20610 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20611 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20613 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20615 * doc/cppopts.texi: Document '-' special value to -MF.
20617 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20619 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20620 (cortex_a53_fconst): Likewise.
20621 (cortex_a53_fpmul): Likewise.
20622 (cortex_a53_f_load_64): Likewise.
20623 (cortex_a53_f_load_many): Likewise.
20624 (cortex_a53_advsimd_alu): Likewise.
20625 (cortex_a53_advsimd_alu_q): Likewise.
20626 (cortex_a53_advsimd_mul): Likewise.
20627 (cortex_a53_advsimd_mul_q): Likewise.
20628 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20629 Add missing fmul, r2f_cvt and fconst cases.
20631 2017-06-14 Richard Biener <rguenther@suse.de>
20633 PR middle-end/81088
20634 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20636 (fold_binary_loc): When associating do not treat pre-existing
20637 TREE_OVERFLOW on literal constants as a reason to allow
20638 TREE_OVERFLOW on associated literal constants.
20640 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20642 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20643 (MASK_FEATURES): New macro.
20644 * config/sparc/sparc.c (sparc_option_override): Remove the special
20645 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20647 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20649 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20650 a division of 0 if non-call exceptions are enabled.
20652 2017-06-14 Andrew Pinski <apinski@cavium.com>
20653 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20656 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20657 Improve vector initialization code gen for only variable case.
20659 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20661 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20663 2017-06-14 Richard Biener <rguenther@suse.de>
20665 PR tree-optimization/81083
20666 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20669 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20671 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20672 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20673 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20674 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20675 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20676 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20680 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20681 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20683 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20685 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20687 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20689 * config/rs6000/t-rtems: Don't handle SPE.
20691 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20693 * config/rs6000/t-linux: Don't handle SPE.
20695 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20697 * config/rs6000/eabispe.h: Delete file.
20699 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20701 * config/rs6000/t-spe: Delete file.
20703 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20705 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20706 (rs6000_legitimate_offset_address_p): Return false for anything in
20707 V2SImode or V2SFmode.
20709 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20711 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20712 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20713 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20715 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20716 (rs6000_legitimate_offset_address_p): Ditto.
20717 (rs6000_emit_move): Ditto.
20718 (rs6000_init_builtins): Remove V4HI_type_node.
20720 2017-06-13 Martin Liska <mliska@suse.cz>
20723 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20724 (gate_asan): Likewise.
20725 * asan.h (asan_no_sanitize_address_p): Remove the function.
20726 (sanitize_flags_p): New function.
20727 * builtins.def: Fix coding style.
20728 * common.opt: Use renamed enum value.
20729 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20730 * doc/extend.texi: Document no_sanitize attribute.
20731 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20732 to SANITIZE_UNDEFINED_NONDEFAULT.
20733 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20734 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20735 Use sanitize_flags_p.
20736 * gimplify.c (gimplify_function_tree): Likewise.
20737 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20738 * opts.c (parse_no_sanitize_attribute): New function.
20739 (common_handle_option): Use renamed enum value.
20740 * opts.h (parse_no_sanitize_attribute): Declare.
20741 * tree.c (sanitize_flags_p): New function.
20742 * tree.h: Declared here.
20743 * tsan.c: Use sanitize_flags_p.
20744 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20745 (instrument_mem_ref): Likewise.
20746 (instrument_bool_enum_load): Likewise.
20747 (do_ubsan_in_current_function): Remove the function.
20748 (pass_ubsan::execute): Use sanitize_flags_p.
20749 * ubsan.h: Remove do_ubsan_in_current_function
20750 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20751 (dump_function_to_file): Use it here.
20753 2017-06-13 Martin Jambor <mjambor@suse.cz>
20755 PR tree-optimization/80803
20756 PR tree-optimization/81063
20757 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20758 (propagate_subaccesses_across_link): Enqueue subtree whenever
20759 necessary instead of relying on the caller.
20761 2017-06-13 Martin Jambor <mjambor@suse.cz>
20763 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20764 that have a first_link.
20765 (sort_and_splice_var_accesses): Do not check first_link before
20767 (subtree_mark_written_and_enqueue): Likewise.
20768 (propagate_all_subaccesses): Likewise and do not stop at first
20769 parent with a first_link.
20771 2017-06-13 Martin Jambor <mjambor@suse.cz>
20773 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20776 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20778 * match.pd: New pattern.
20780 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20782 * tree-vrp.c (is_masked_range_test): New function.
20783 (register_edge_assert_for): Determine ranges for
20786 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20788 PR tree-optimization/67328
20789 * fold-const.c (maskable_range_p): New function.
20790 (build_range_check): Generate bittests if possible.
20792 2017-06-13 Martin Liska <mliska@suse.cz>
20794 * gimple-pretty-print.c (dump_probability): Add new argument.
20795 (dump_edge_probability): Dump both probability and count.
20796 (dump_gimple_label): Likewise.
20797 (dump_gimple_bb_header): Likewise.
20799 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20802 * config/avr/avr-devices.c: Fix indentation.
20803 * config/avr/gen-avr-mmcu-specs.c: Dito.
20805 2017-06-13 Richard Biener <rguenther@suse.de>
20807 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20808 instead get vector type from stmt_info.
20809 (vectorizable_reduction): Adjust. Remove dead code.
20811 2017-06-13 Richard Biener <rguenther@suse.de>
20813 PR middle-end/81065
20814 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20815 case of C * (x * C2 + C3).
20816 (fold_addr_of_array_ref_difference): Properly fold index difference.
20818 2017-06-12 David S. Miller <davem@davemloft.net>
20821 * config/sparc/sparc.md (return expander): Emit frame blockage if
20822 function uses alloca.
20824 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20826 * combine.c (make_field_assignment): Check len rather than the mode
20827 precision when calling force_to_mode.
20829 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20831 Support multilibs and devices that see flash in RAM address range.
20834 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20835 (avr_mcu_t) <flash_pm_offset>: New field.
20836 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20837 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20838 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20839 (AVR_TINY_PM_OFFSET): Remove macro.
20840 * config/avr/avr.opt (-mshort-calls): New option.
20841 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20842 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20843 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20844 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20845 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20846 instead of avr_arch->have_jmp_call.
20847 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20848 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20849 avr_arch->flash_pm_offset to define.
20850 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20851 new field flash_pm_offset. Add entry for avrxmega3.
20852 (avr_texinfo): Add entry for avrxmega3.
20853 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20854 attiny212, attiny214,
20855 attiny412, attiny414, attiny416, attiny417,
20856 attiny814, attiny816, attiny817,
20857 attiny1614, attiny1616, attiny1617,
20858 attiny3214, attiny3216, attiny3217.
20859 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20860 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20861 (avr_print_operand_address) [AVR_TINY]: Same.
20862 (avr_asm_init_sections) <readonly_data_section>: Only patch
20863 callback if avr_arch->flash_pm_offset = 0.
20864 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20865 for rodata if avr_arch->flash_pm_offset != 0.
20866 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20867 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20868 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20869 (m_options): Append opt_rcall.
20870 (m_dirnames): Append dir_rcall.
20871 * config/avr/t-multilib: Regenerate.
20873 * configure.ac [target=avr]: Check whether avrxmega3 default
20874 linker description file works as needed.
20875 * configure: Regenerate.
20876 * doc/avr-mmcu.texi: Regenerate.
20877 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20878 <__AVR_ARCH__>: Document avrxmega3 and 103.
20879 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20880 <__AVR_SHORT_CALLS__>: Document it.
20881 <__AVR_PM_BASE_ADDRESS__>: Document it.
20882 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20883 (AVR Variable Attributes) <progmem>: Document this is
20884 not needed for avrxmega3.
20885 (AVR Named Address Spaces) <__flash>: Dito.
20887 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20889 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20891 2017-06-12 Doug Rupp <rupp@adacore.com>
20893 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20894 Append vxworks-stdint.h to the tm_file list.
20895 * config/vxworks-stdint.h: New file.
20897 2017-06-12 Martin Liska <mliska@suse.cz>
20899 PR tree-optimization/81041
20900 * tree-profile.c (gimple_gen_ic_func_profiler):
20901 Create an extra BB in profile-generate
20902 (gimple_gen_time_profiler): Likewise.
20904 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20906 PR tree-optimization/81003
20907 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20908 (update_range_test): Use it instead of force_gimple_operand_gsi.
20910 2017-06-12 Richard Biener <rguenther@suse.de>
20912 PR tree-optimization/81053
20913 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20914 with backedge value not defined in loop. Simplify def stmt
20917 2017-06-11 Tom de Vries <tom@codesourcery.com>
20920 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20922 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20923 nvptx_cannot_force_const_mem.
20925 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20927 * opts.c (finish_options): Move test for flag_split_stack after
20928 it has been initialized.
20930 2017-06-11 Jason Merrill <jason@redhat.com>
20932 * tree.h (id_equal): New.
20933 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20934 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20935 instead of strcmp of IDENTIFIER_POINTER.
20937 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20939 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20940 (mark_all_inlined_calls_cdtor): Fix formating.
20941 (inline_transform): Rescale profile before inlining.
20943 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20945 * cgraph.h (cgraph_edge::clone): Update prototype.
20946 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20947 (cgraph_node::create_clone): Update.
20948 (cgraph_node::create_version_clone): Update.
20949 * tree-inline.c (copy_bb): Update.
20950 (expand_call_inline): Update.
20952 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20955 factored out from ...
20956 (rs6000_emit_prologue): ... here.
20958 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20960 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20961 factored out from ...
20962 (rs6000_emit_prologue): ... here.
20964 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20966 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20968 (handle_missing_profiles): Fix computation of tp_first_run.
20969 (counts_to_freqs): Do not touch freqs when count is 0.
20971 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20973 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20976 2017-06-10 Tom de Vries <tom@codesourcery.com>
20978 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20979 attributes): Document signal effective target.
20981 2017-06-10 Tom de Vries <tom@codesourcery.com>
20983 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20984 Document effective target stack_size.
20986 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20988 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20989 to the edit_context if they can be auto-applied.
20991 2017-06-9 Ian Lance Taylor <iant@golang.org>
20993 * opts.c (finish_options): If -fsplit-stack, disable implicit
20994 -forder-blocks-and-partition.
20995 * doc/invoke.texi (Optimize Options): Document that when using
20996 -fsplit-stack -forder-blocks-and-partition is not implicitly
20999 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21001 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21002 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21003 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21004 * builtins.def (abort, trap, unreachable): Declare cold.
21005 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21006 * tree-core.h (ECF_COLD): New.
21007 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21008 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21010 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21012 * predict.c (unlikely_executed_stmt_p): Cleanup.
21014 2017-06-09 Richard Biener <rguenther@suse.de>
21016 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21017 model if the ref is always written to.
21019 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21021 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21023 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21025 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21028 2017-06-09 Tom de Vries <tom@codesourcery.com>
21031 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21032 "target cannot support label values" when encountering LABEL_REF.
21034 2017-06-09 Martin Liska <mliska@suse.cz>
21036 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21037 (gimple_gen_ic_func_profiler): Emit direct comparison
21038 of __gcov_indirect_call_callee with NULL.
21039 (gimple_gen_time_profiler): Change probability from
21040 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21042 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21044 * profile.c (edge_gcov_counts): Turn to pointer.
21045 (compute_branch_probabilities, compute_branch_probabilities): Update.
21046 (branch_prob): Do not clear edge_gcov_count.
21047 * profile.h (edge_gcov_counts): Turn to pointer.
21048 (edge_gcov_count): Update.
21050 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21052 * gimple.h (gimple_check_failed): Mark cold.
21054 2017-06-09 Richard Biener <rguenther@suse.de>
21056 PR tree-optimization/66623
21057 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21058 refactor check_reduction into two parts, properly computing
21059 whether we have to check reduction validity for outer loop
21062 2017-06-09 Richard Biener <rguenther@suse.de>
21064 PR tree-optimization/79483
21065 * graphite-scop-detection.c (order): New global.
21066 (get_order): Compute bb to order mapping that satisfies code
21067 generation constraints.
21068 (cmp_pbbs): New helper.
21069 (build_scops): Start domwalk at entry block, sort generated
21072 2017-06-09 Richard Biener <rguenther@suse.de>
21074 PR middle-end/81007
21075 * ipa-polymorphic-call.c
21076 (ipa_polymorphic_call_context::restrict_to_inner_class):
21077 Skip FIELD_DECLs with error_mark_node type.
21078 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21081 2017-06-09 Martin Liska <mliska@suse.cz>
21083 * predict.c (struct branch_predictor): New struct.
21084 (test_prediction_value_range): New test.
21085 (predict_c_tests): New function.
21086 * selftest-run-tests.c (selftest::run_tests): Run the function.
21087 * selftest.h: Declare new tests.
21089 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21092 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21093 gen_add3_insn did not fail.
21094 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21095 r0, construct that number in a temporary reg and add that reg to r0.
21096 If asked to put the result in r0 as well, fail.
21098 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21100 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21101 for early expansion of vec_eqv.
21103 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21105 PR middle-end/81005
21106 * ubsan.c (instrument_null): Avoid pointless code temporary.
21107 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21109 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21113 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21116 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21119 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21121 * predict.c (unlikely_executed_edge_p): Move ahead.
21122 (probably_never_executed_edge_p): Use it.
21124 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21126 PR middle-end/79988
21127 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21128 gimple_call_builtin_p call.
21130 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21132 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21133 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21134 rtl_check_failed_type2, rtl_check_failed_code1,
21135 rtl_check_failed_code2, rtl_check_failed_code_mode,
21136 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21137 rtvec_check_failed_bounds, rtl_check_failed_flag,
21138 _fatal_insn_not_found, _fatal_insn): Likewise.
21139 * tree.h (tree_contains_struct_check_failed,
21140 tree_check_failed, tree_not_check_failed,
21141 tree_class_check_failed, tree_range_check_failed,
21142 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21143 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21144 tree_operand_check_failed, omp_clause_check_failed,
21145 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21148 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21150 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21151 flag_branch_probabilities.
21152 * ipa-inline.c (edge_badness): Likewise.
21153 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21154 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21155 * predict.c (maybe_hot_frequency_p): Likewise.
21156 (probably_never_executed): Likewise.
21157 * sched-ebb.c (schedule_ebbs): Likewise.
21158 * sched-rgn.c (find_single_block_region): Likewise.
21159 * tracer.c (tail_duplicate): Likewise.
21161 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21163 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21164 longer requires x_flag_profile_use.
21166 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21168 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21169 instead of flag_reorder_blocks_and_partition.
21170 * dbxout.c (dbxout_function_end): Likewise.
21171 * dwarf2out.c (gen_subprogram_die): Likewise.
21172 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21173 * hw-doloop.c (reorg_loops): Likewise.
21174 * varasm.c (assemble_start_function,
21175 assemble_end_function): Likewise.
21176 (decide_function_section): Do not check for
21177 flag_reorder_blocks_and_partition.
21179 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21181 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21183 (chkp_get_hard_register_fake_addr_expr): Ditto.
21184 (chkp_build_addr_expr): Add check for hard reg case.
21185 (chkp_parse_array_and_component_ref): Ditto.
21186 (chkp_find_bounds_1): Ditto.
21187 (chkp_process_stmt): Don't generate bounds store for
21190 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21192 * predict.c (maybe_hot_bb_p): Do not check profile status.
21193 (maybe_hot_edge_p): Likewise.
21194 (probably_never_executed): Check for zero counts even if profile
21196 (unlikely_executed_edge_p): New function.
21197 (unlikely_executed_stmt_p): New function.
21198 (unlikely_executed_bb_p): New function.
21199 (set_even_probabilities): Use unlikely predicates.
21200 (combine_predictions_for_bb): Likewise.
21201 (predict_paths_for_bb): Likewise.
21202 (predict_paths_leading_to_edge): Likewise.
21203 (determine_unlikely_bbs): New function.
21204 (estimate_bb_frequencies): Use it.
21205 (compute_function_frequency): Use zero counts even if profile is
21207 * profile-count.h: Fix typo.
21209 2017-08-08 Julia Koval <julia.koval@intel.com>
21211 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21212 _mm512_mask_cvtsepi16_storeu_epi8,
21213 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21214 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21215 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21216 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21217 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21218 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21219 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21220 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21221 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21222 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21223 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21224 __builtin_ia32_pmovuswb256mem_mask,
21225 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21226 __builtin_ia32_pmovwb512mem_mask): New builtins.
21228 2017-08-08 Julia Koval <julia.koval@intel.com>
21230 PR target/73350,80862
21231 * config/i386/subst.md (round): Fix round pattern.
21232 * config/i386/i386.c (ix86_erase_embedded_rounding):
21233 Fix erasing rounding for the fixed pattern.
21235 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21237 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21239 2017-06-08 Martin Liska <mliska@suse.cz>
21241 PR gcov-profile/80911
21242 * gcov.c (block_info::block_info): New constructor.
21244 2017-06-07 Carl Love <cel@us.ibm.com>
21246 * config/rs6000/rs6000-c: The return type of the following
21247 built-in functions was implemented as int not long long. Fix sign
21248 of return value for the unsigned version of vec_mulo and vec_mule.
21249 vector unsigned long long vec_bperm (vector unsigned long long,
21250 vector unsigned char)
21251 vector signed long long vec_mule (vector signed int,
21253 vector unsigned long long vec_mule (vector unsigned int,
21254 vector unsigned int)
21255 vector signed long long vec_mulo (vector signed int,
21257 vector unsigned long long vec_mulo (vector unsigned int,
21258 vector unsigned int)
21259 * doc/extend.texi: Fix the documentation for the built-in
21262 2017-06-07 Carl Love <cel@us.ibm.com>
21265 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21268 2017-06-07 Carl Love <cel@us.ibm.com>
21270 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21271 support, Generate doublehv for signed int/float for BE case only.
21273 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21275 * doc/invoke.texi (mcx16): Rewrite.
21277 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21279 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21280 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21281 *mov<mode>_softfloat, and an anonymous splitter): Use
21282 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21284 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21286 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21288 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21289 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21290 (rs6000_debug_reg_global): Adjust.
21291 (rs6000_init_hard_regno_mode_ok): Adjust.
21292 (rs6000_dbx_register_number): Adjust.
21293 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21294 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21295 Remove SPE_ACC and SPEFSCR.
21296 (REG_ALLOC_ORDER): Ditto.
21297 (FRAME_POINTER_REGNUM): Change to 111.
21298 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21299 (REG_CLASS_NAMES): Ditto.
21300 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21301 (REGISTER_NAMES): Ditto.
21302 (ADDITIONAL_REG_NAMES): Ditto.
21303 (rs6000_reg_names): Ditto.
21304 * config/rs6000/rs6000.md: Renumber some register number
21307 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21309 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21311 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21312 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21314 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21315 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21316 Delete the SPE high registers.
21317 (REG_ALLOC_ORDER): Ditto.
21318 (enum reg_class): Remove SPE_HIGH_REGS.
21319 (REG_CLASS_NAMES): Ditto.
21320 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21321 (REGISTER_NAMES): Ditto.
21322 (rs6000_reg_names): Ditto.
21323 * doc/tm.texi.in: Remove SPE as example.
21324 * doc/tm.texi: Regenerate.
21326 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21328 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21329 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21330 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21331 * config/rs6000/rs6000.md (type): Remove "brinc".
21333 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21335 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21336 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21337 * config/rs6000/linuxspe.h: Delete file.
21338 * config/rs6000/rs6000.md: Don't include spe.md.
21339 * config/rs6000/spe.h: Delete file.
21340 * config/rs6000/spe.md: Delete file.
21341 * config/rs6000/t-rs6000: Remove spe.md.
21343 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21345 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21346 (reg_or_none500mem_operand): Delete.
21347 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21348 instead of reg_or_none500mem_operand.
21350 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21352 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21353 handling of SPE flags.
21354 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21356 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21358 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21360 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21361 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21362 paired_divv2sf3): Similar.
21363 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21364 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21365 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21367 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21368 Rename the paired_* instruction patterns.
21369 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21371 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21372 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21373 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21374 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21375 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21376 PAIRED_VECTOR_MODE.
21377 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21378 (spe_func_has_64bit_regs_p): Delete.
21379 (spe_expand_predicate_builtin): Delete.
21380 (spe_expand_evsel_builtin): Delete.
21381 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21382 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21383 (invalid_e500_subreg): Delete.
21384 (rs6000_legitimize_address): Always force_reg op2 as well, for
21385 paired single memory accesses.
21386 (rs6000_member_type_forces_blk): Delete.
21387 (rs6000_spe_function_arg): Delete.
21388 (rs6000_expand_unop_builtin): Delete SPE handling.
21389 (rs6000_expand_binop_builtin): Ditto.
21390 (spe_expand_stv_builtin): Delete.
21391 (bdesc_2arg_spe): Delete.
21392 (spe_expand_builtin): Delete.
21393 (spe_expand_predicate_builtin): Delete.
21394 (spe_expand_evsel_builtin): Delete.
21395 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21396 (spe_init_builtins): Delete.
21397 (spe_func_has_64bit_regs_p): Delete.
21398 (savres_routine_name): Delete "info" parameter. Adjust callers.
21399 (rs6000_emit_stack_reset): Ditto.
21400 (rs6000_dwarf_register_span): Delete.
21401 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21402 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21403 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21405 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21407 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21408 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21409 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21410 mulv2sf3, divv2sf3): Delete expanders.
21412 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21414 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21418 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21419 * config/rs6000/rs6000.c: Ditto.
21421 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21423 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21424 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21425 comparison_operator.
21427 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21429 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21430 * config/rs6000/rs6000.opt: Ditto.
21431 * config/rs6000/t-rtems: Ditto.
21433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21435 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21436 TARGET_E500_SINGLE by 0, simplify.
21437 * config/rs6000/rs6000.c: Ditto.
21438 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21439 (spe_build_register_parallel): Delete.
21440 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21441 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21442 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21443 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21444 (E500_CONVERT): Delete.
21445 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21449 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21450 * config/rs6000/dfp.md: Ditto.
21451 (negdd2, *negdd2_fpr): Merge.
21452 (absdd2, *absdd2_fpr): Merge.
21453 (negtd2, *negtd2_fpr): Merge.
21454 (abstd2, *abstd2_fpr): Merge.
21455 * config/rs6000/e500.h: Delete file.
21456 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21457 TARGET_FPRS by 1 and simplify.
21458 * config/rs6000/rs6000-c.c: Ditto.
21459 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21460 TARGET_DF_SPE by 0.
21461 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21463 * config/rs6000/rs6000.md: Ditto.
21464 (floatdidf2, *floatdidf2_fpr): Merge.
21465 (move_from_CR_gt_bit): Delete.
21466 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21467 (E500_CR_IOR_COMPARE): Delete.
21468 (All patterns that require !TARGET_FPRS): Delete.
21469 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21471 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21473 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21475 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21477 * graphds.c (add_edge): Intitialize edge's attached data.
21478 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21479 pointer parameter. Call pointed function on each edge during
21480 graph traversing. Skip traversing the edge when the function
21482 (graphds_dfs, graphds_scc): Ditto.
21483 (for_each_edge): New parameter. Pass the new parameter to callback
21485 * graphds.h (skip_edge_callback): New function pointer type.
21486 (graphds_dfs, graphds_scc): New function pointer parameter.
21487 (graphds_edge_callback, for_each_edge): New parameter.
21489 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21491 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21492 out code checking if runtime alias check is possible to below ...
21493 Call the new function.
21494 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21495 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21497 2017-06-07 Marek Polacek <polacek@redhat.com>
21500 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21501 TYPE_OVERFLOW_WRAPS check.
21503 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21505 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21506 if versioning is required.
21507 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21508 peeling with the check for versioning.
21510 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21512 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21513 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21514 Set true to new parameter if new ssa variable is defined.
21515 (vect_gen_vector_loop_niters): Refactor. Set range information
21516 for the new vector loop bound variable.
21517 (vect_do_peeling): Ditto.
21519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21521 * tree-affine.c (ssa.h): Include header file.
21522 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21523 has wrapping overflow behavior.
21525 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21527 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21531 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21532 (tree_to_aff_combination): ... here.
21534 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21536 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21537 reg_pressure model function.
21538 (ivopts_global_cost_for_size): Delete.
21539 (determine_set_costs, iv_ca_recount_cost): Call new model function
21540 ivopts_estimate_reg_pressure.
21542 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21544 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21545 expensive than udiv. Remove floating point cases from mod.
21547 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21549 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21550 Increase idiv cost.
21552 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21554 * config/aarch64/aarch64.md
21555 (copysignsf3): Fix mask generation.
21557 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21559 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21561 (class dump_manager): Add register_dumps method.
21562 * dumpfile.c: Include langhooks.h.
21563 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21564 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21565 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21566 (dump_manager::dump_register): Start with 512 entries instead of 32.
21567 (dump_manager::register_dumps): New method.
21568 * toplev.c (general_init): Instead of invoking register_dumps
21569 langhook, invoke register_dumps method on the dump manager.
21570 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21573 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21575 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21576 Say that # requires an associated define_split to exist, and that
21577 the define_split must be suitable for use after register allocation.
21579 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21581 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21582 (compute_outgoing_frequencies): Also initialize zero counts.
21583 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21584 around loops; preserve more of profile when nothing changes.
21586 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21588 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21590 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21591 * config/arm/arm-cpu-cdata.h: Regenerate.
21592 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21593 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21594 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21595 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21596 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21598 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21600 * config/arm/t-rmprofile: Likewise.
21601 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21603 2017-06-06 David S. Miller <davem@davemloft.net>
21606 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21607 blockage if function uses alloca.
21609 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21611 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21612 New "uid" fields to hold pretty-print IDs of group and ref.
21613 Memory references are now identified as <group_id>:<ref_id>
21614 instead of using [random] addresses.
21615 (dump_mem_details): Simplify, no functional change.
21616 (dump_mem_ref): Simplify and make output more concise.
21617 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21618 (find_or_create_group): Initialize group uid.
21619 (record_ref): Initialize ref uid. Improve debug output.
21620 (prune_group_by_reuse, should_issue_prefetch_p,)
21621 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21622 (mark_nontemporal_store, determine_loop_nest_reuse):
21623 Improve debug output.
21625 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21627 * dbgcnt.def (prefetch): New debug counter.
21628 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21629 (schedule_prefetches): Stop issueing prefetches if debug counter
21632 2017-06-06 Tom de Vries <tom@codesourcery.com>
21634 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21635 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21637 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21639 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21640 Use aarch64_reg_or_zero predicate for operand 4.
21641 (aarch64_compare_and_swap<mode> define_insn_and_split):
21642 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21643 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21645 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21647 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21648 (arm_compute_save_core_reg_mask): This.
21649 (thumb1_compute_save_reg_mask): Rename into ...
21650 (thumb1_compute_save_core_reg_mask): This.
21651 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21652 (arm_compute_frame_layout): Likewise.
21654 2017-06-06 Richard Biener <rguenther@suse.de>
21656 PR tree-optimization/80974
21657 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21658 keep or clear leaders SSA info.
21660 2017-06-06 Tom de Vries <tom@codesourcery.com>
21662 * config/nvptx/nvptx.c (split_mode_p): New function.
21663 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21665 2017-06-06 Tom de Vries <tom@codesourcery.com>
21667 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21672 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21675 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21677 * shrink-wrap.c (handle_simple_exit): Update profile.
21678 (try_shrink_wrapping): Upate profile.
21680 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21682 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21683 (tree_guess_outgoing_edge_probabilities): New.
21684 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21685 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21687 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21689 * ipa-split.c (split_function): Initialize return bb profile.
21691 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21693 * profile.c (compute_branch_probabilities): Also initialize
21694 EXIT_BLOCK profile.
21696 2017-06-06 Richard Biener <rguenther@suse.de>
21698 PR tree-optimization/80928
21699 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21700 (vect_analyze_loop_operations): Properly guard analysis for
21702 (vect_transform_loop): Likewise.
21703 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21704 (vect_model_induction_cost): Do not cost for pure SLP.
21705 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21706 of induction in inner loop vectorization.
21707 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21708 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21709 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21711 (vect_analyze_slp_cost_1): Cost induction.
21712 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21713 (vect_get_slp_vect_defs): Likewise.
21714 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21715 (vect_transform_stmt): Handle SLP reductions.
21716 * tree-vectorizer.h (vectorizable_induction): Adjust.
21718 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21720 * config/rs6000/rs6000.c (make_resolver_func): Update
21721 init_lowered_empty_function call.
21723 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21725 * doc/invoke.texi: Document the -fprofile-abs-path option.
21726 * common.opt (fprofile-abs-path): New option.
21727 * gcov-io.h (gcov_write_filename): Declare.
21728 * gcov-io.c (gcov_write_filename): New function.
21729 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21730 * profile.c (output_location): Likewise.
21732 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21734 * shring-wrap.c: Revert accidental commit.
21736 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21738 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21740 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21742 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21744 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21745 profile in callgraph edge.
21746 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21747 (apply_scale): Likewise.
21748 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21749 Also scale profile when inlining function with zero profile.
21750 (initialize_cfun): Update exit block profile even when it is zero.
21751 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21752 when profile is read.
21754 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21756 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21757 (CLONE_*): New constants to define the processors we can generate
21758 code for with the target_clone attribute.
21759 (rs6000_clone_map): New array to identify which clone processors
21760 the current program is running on.
21761 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21762 target_clone attribute.
21763 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21764 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21765 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21766 (cpu_expand_builtin): Add support for target_clone attribute.
21767 (rs6000_valid_attribute_p): Allow "default" attribute.
21768 (get_decl_name): New debug function to simplify printing the
21769 current function name in debugging statements.
21770 (rs6000_clone_priority): New functions to support the target_clone
21771 attribute, and be able to generate code to switch between ISA 2.05
21772 through ISA 3.0 (power6 through power9).
21773 (rs6000_compare_version_priority): Likewise.
21774 (rs6000_get_function_versions_dispatcher): Likewise.
21775 (make_resolver_func): Likewise.
21776 (add_condition_to_bb): Likewise.
21777 (dispatch_function_versions): Likewise.
21778 (rs6000_generate_version_dispatcher_body): Likewise.
21779 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21780 (fusion_gpr_load_p): Fix a spacing issue.
21781 * doc/extend.texi (Common Function Attributes): Document that the
21782 PowerPC supports the target_clone attribute.
21784 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21786 * config/arm/arm.h: explain F symbol found in description of ARM
21787 register allocation in its legend.
21789 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21791 * config/mips/frame-header-opt.c: Include profile-count.h.
21792 * config/riscv/riscv.c: Include profile-count.h
21794 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21796 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21798 (sm_set_flag_if_changed): Add bbs field.
21799 (execute_sm_if_changed_flag_set): Pass BBS.
21800 (execute_sm): Update.
21802 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21804 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21807 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21809 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21810 (peephole2): New peephole2 to emit the above.
21811 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21813 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21815 * config/aarch64/aarch64.c (define_peephole2 above
21816 *sub_<shift>_<mode>): New peephole.
21818 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21820 * config/i386/i386.c (make_resolver_func): Update.
21821 * Makefile.in: Add profile-count.h and profile-count.o
21822 * auto-profile.c (afdo_indirect_call): Update to new API.
21823 (afdo_set_bb_count): Update.
21824 (afdo_propagate_edge): Update.
21825 (afdo_propagate_circuit): Update.
21826 (afdo_calculate_branch_prob): Update.
21827 (afdo_annotate_cfg): Update.
21828 * basic-block.h: Include profile-count.h
21829 (struct edge_def): Turn count to profile_count.
21830 (struct basic_block_def): Likewie.
21831 (REG_BR_PROB_BASE): Move to profile-count.h
21832 (RDIV): Move to profile-count.h
21833 * bb-reorder.c (max_entry_count): Turn to profile_count.
21834 (find_traces): Update.
21835 (rotate_loop):Update.
21836 (connect_traces):Update.
21837 (sanitize_hot_paths):Update.
21838 * bt-load.c (migrate_btr_defs): Update.
21839 * cfg.c (RDIV): Remove.
21840 (init_flow): Use alloc_block.
21841 (alloc_block): Uninitialize count.
21842 (unchecked_make_edge): Uninitialize count.
21843 (check_bb_profile): Update.
21844 (dump_edge_info): Update.
21845 (dump_bb_info): Update.
21846 (update_bb_profile_for_threading): Update.
21847 (scale_bbs_frequencies_int): Update.
21848 (scale_bbs_frequencies_gcov_type): Update.
21849 (scale_bbs_frequencies_profile_count): New.
21850 * cfg.h (update_bb_profile_for_threading): Update.
21851 (scale_bbs_frequencies_profile_count): Declare.
21852 * cfgbuild.c (compute_outgoing_frequencies): Update.
21853 (find_many_sub_basic_blocks): Update.
21854 * cfgcleanup.c (try_forward_edges): Update.
21855 (try_crossjump_to_edge): Update.
21856 * cfgexpand.c (expand_gimple_tailcall): Update.
21857 (construct_exit_block): Update.
21858 * cfghooks.c (verify_flow_info): Update.
21859 (dump_bb_for_graph): Update.
21860 (split_edge): Update.
21861 (make_forwarder_block): Update.
21862 (duplicate_block): Update.
21863 (account_profile_record): Update.
21864 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21865 (get_estimated_loop_iterations): Update.
21866 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21867 (single_likely_exit): Update.
21868 * cfgloopmanip.c (scale_loop_profile): Update.
21870 (set_zero_probability): Update.
21871 (lv_adjust_loop_entry_edge): Update.
21872 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21873 (purge_dead_edges): Update.
21874 (rtl_account_profile_record): Update.
21875 * cgraph.c (cgraph_node::create): Uninitialize count.
21876 (symbol_table::create_edge): Uninitialize count.
21877 (cgraph_update_edges_for_call_stmt_node): Update.
21878 (cgraph_edge::dump_edge_flags): Update.
21879 (cgraph_node::dump): Update.
21880 (cgraph_edge::maybe_hot_p): Update.
21881 * cgraph.h: Include profile-count.h
21882 (create_clone), create_edge, create_indirect_edge): Update.
21883 (cgraph_node): Turn count to profile_count.
21884 (cgraph_edge0: Likewise.
21885 (make_speculative, clone): Update.
21886 (create_edge): Update.
21887 (init_lowered_empty_function): Update.
21888 * cgraphclones.c (cgraph_edge::clone): Update.
21889 (duplicate_thunk_for_node): Update.
21890 (cgraph_node::create_clone): Update.
21891 * cgraphunit.c (cgraph_node::analyze): Update.
21892 (cgraph_node::expand_thunk): Update.
21893 * final.c (dump_basic_block_info): Update.
21894 * gimple-streamer-in.c (input_bb): Update.
21895 * gimple-streamer-out.c (output_bb): Update.
21896 * graphite.c (print_global_statistics): Update.
21897 (print_graphite_scop_statistics): Update.
21898 * hsa-brig.c: Include basic-block.h.
21899 * hsa-dump.c: Include basic-block.h.
21900 * hsa-gen.c (T sum_slice): Update.
21901 (convert_switch_statements):Update.
21902 * hsa-regalloc.c: Include basic-block.h.
21903 * ipa-chkp.c (chkp_produce_thunks): Update.
21904 * ipa-cp.c (struct caller_statistics): Update.
21905 (init_caller_stats): Update.
21906 (gather_caller_stats): Update.
21907 (ipcp_cloning_candidate_p): Update.
21908 (good_cloning_opportunity_p): Update.
21909 (get_info_about_necessary_edges): Update.
21910 (dump_profile_updates): Update.
21911 (update_profiling_info): Update.
21912 (update_specialized_profile): Update.
21913 (perhaps_add_new_callers): Update.
21914 (decide_about_value): Update.
21915 (ipa_cp_c_finalize): Update.
21916 * ipa-devirt.c (struct odr_type_warn_count): Update.
21917 (struct decl_warn_count): Update.
21918 (struct final_warning_record): Update.
21919 (possible_polymorphic_call_targets): Update.
21920 (ipa_devirt): Update.
21921 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21922 * ipa-icf.c (sem_function::merge): Update.
21923 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21924 * ipa-inline.c (compute_uninlined_call_time): Update.
21925 (compute_inlined_call_time): Update.
21926 (want_inline_small_function_p): Update.
21927 (want_inline_self_recursive_call_p): Update.
21928 (edge_badness): Update.
21929 (lookup_recursive_calls): Update.
21930 (recursive_inlining): Update.
21931 (inline_small_functions): Update.
21932 (dump_overall_stats): Update.
21933 (dump_inline_stats): Update.
21934 * ipa-profile.c (ipa_profile_generate_summary): Update.
21935 (ipa_propagate_frequency): Update.
21936 (ipa_profile): Update.
21937 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21938 * ipa-utils.c (ipa_merge_profiles): Update.
21939 * loop-doloop.c (doloop_modify): Update.
21940 * loop-unroll.c (report_unroll): Update.
21941 (unroll_loop_runtime_iterations): Update.
21942 * lto-cgraph.c (lto_output_edge): Update.
21943 (lto_output_node): Update.
21944 (input_node): Update.
21945 (input_edge): Update.
21946 (merge_profile_summaries): Update.
21947 * lto-streamer-in.c (input_cfg): Update.
21948 * lto-streamer-out.c (output_cfg): Update.
21949 * mcf.c (create_fixup_graph): Update.
21950 (adjust_cfg_counts): Update.
21951 (sum_edge_counts): Update.
21952 * modulo-sched.c (sms_schedule): Update.
21953 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21954 * predict.c (maybe_hot_count_p): Update.
21955 (probably_never_executed): Update.
21956 (dump_prediction): Update.
21957 (combine_predictions_for_bb): Update.
21958 (propagate_freq): Update.
21959 (handle_missing_profiles): Update.
21960 (counts_to_freqs): Update.
21961 (rebuild_frequencies): Update.
21962 (force_edge_cold): Update.
21963 * predict.h: Include profile-count.h
21964 (maybe_hot_count_p, counts_to_freqs): UPdate.
21965 * print-rtl-function.c: Do not include cfg.h
21966 * print-rtl.c: Include basic-block.h
21967 * profile-count.c: New file.
21968 * profile-count.h: New file.
21969 * profile.c (is_edge_inconsistent): Update.
21970 (correct_negative_edge_counts): Update.
21971 (is_inconsistent): Update.
21972 (set_bb_counts): Update.
21973 (read_profile_edge_counts): Update.
21974 (compute_frequency_overlap): Update.
21975 (compute_branch_probabilities): Update; Initialize and deinitialize
21977 (branch_prob): Update.
21978 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21979 (edge_gcov_count): New.
21980 (bb_gcov_count): New.
21981 * shrink-wrap.c (try_shrink_wrapping): Update.
21982 * tracer.c (better_p): Update.
21983 * trans-mem.c (expand_transaction): Update.
21984 (ipa_tm_insert_irr_call): Update.
21985 (ipa_tm_insert_gettmclone_call): Update.
21986 * tree-call-cdce.c: Update.
21987 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21988 (gimple_duplicate_sese_tail): Update.
21989 (gimple_account_profile_record): Update.
21990 (execute_fixup_cfg): Update.
21991 * tree-inline.c (copy_bb): Update.
21992 (copy_edges_for_bb): Update.
21993 (initialize_cfun): Update.
21994 (freqs_to_counts): Update.
21995 (copy_cfg_body): Update.
21996 (expand_call_inline): Update.
21997 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21998 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21999 (try_unroll_loop_completely): Update.
22000 (try_peel_loop): Update.
22001 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22002 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22003 * tree-ssa-loop-split.c (connect_loops): Update.
22004 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22005 * tree-ssa-reassoc.c (branch_fixup): Update.
22006 * tree-ssa-tail-merge.c (replace_block_by): Update.
22007 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22008 (compute_path_counts): Update.
22009 (update_profile): Update.
22010 (recompute_probabilities): Update.
22011 (update_joiner_offpath_counts): Update.
22012 (estimated_freqs_path): Update.
22013 (freqs_to_counts_path): Update.
22014 (clear_counts_path): Update.
22015 (ssa_fix_duplicate_block_edges): Update.
22016 (duplicate_thread_path): Update.
22017 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22018 (struct switch_conv_info): Update.
22019 * tree-tailcall.c (decrease_profile): Update.
22020 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22021 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22022 * value-prof.c (check_counter): Update.
22023 (gimple_divmod_fixed_value): Update.
22024 (gimple_mod_pow2): Update.
22025 (gimple_mod_subtract): Update.
22026 (gimple_ic_transform): Update.
22027 (gimple_stringop_fixed_value): Update.
22028 * value-prof.h (gimple_ic): Update.
22030 2017-06-02 Carl Love <cel@us.ibm.com>
22032 * config/rs6000/rs6000-c: Add support for built-in functions
22033 vector double vec_doublee (vector signed int);
22034 vector double vec_doublee (vector unsigned int);
22035 vector double vec_doublee (vector float);
22036 vector double vec_doubleh (vector signed int);
22037 vector double vec_doubleh (vector unsigned int);
22038 vector double vec_doubleh (vector float);
22039 vector double vec_doublel (vector signed int);
22040 vector double vec_doublel (vector unsigned int);
22041 vector double vec_doublel (vector float);
22042 vector double vec_doubleo (vector signed int);
22043 vector double vec_doubleo (vector unsigned int);
22044 vector double vec_doubleo (vector float);.
22045 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22046 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22048 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22049 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22050 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22052 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22053 vec_doublel, vec_doubleh.
22054 * doc/extend.texi: Update the built-in documentation file for the
22055 new built-in functions.
22057 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22060 * ipa-inline-analysis.c (free_growth_caches): Set
22061 edge_removal_hook_holder to NULL after removing it.
22063 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22065 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22066 comparision with zero.
22068 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22069 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22070 for early expansion of vec_min and vec_max builtins.
22071 (builtin_function_type): Add min/max unsigned variants to those
22072 identified as having unsigned arguments.
22074 2017-06-02 Olivier Hainque <hainque@adacore.com>
22076 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22078 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22080 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22081 Use VALL_F16 iterator rather than VALL.
22083 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22085 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22086 Emit CBNZ inside loop when doing a strong exchange and comparing
22087 against zero. Generate the CC flags after the loop.
22089 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22091 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22092 (dl_section_ref): New.
22093 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22094 On AIX, append an expression to subtract the size of the
22095 section length to dl_section_ref.
22097 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22099 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22100 for early expansion of vector absolute builtins.
22102 2017-06-02 Richard Biener <rguenther@suse.de>
22104 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22105 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22107 2017-06-02 Richard Biener <rguenther@suse.de>
22109 PR tree-optimization/80948
22110 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22111 stmt order as well.
22113 2017-06-02 Richard Biener <rguenther@suse.de>
22115 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22117 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22118 for inductions as relevant.
22120 2017-06-02 Richard Biener <rguenther@suse.de>
22122 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22123 (vectorizable_induction): ... this. Remove dead code.
22125 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22127 * builtins. (expand_builtin_alloca): Remove second parameter and
22128 infer its value from the first parameter instead.
22129 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22131 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22133 PR rtl-optimization/80903
22134 * loop-doloop.c (add_test): Unshare sequence.
22136 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22138 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22140 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22142 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22144 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22145 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22146 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22147 (xlogue_layout::get_stub_rtx): Make static.
22148 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22149 (xlogue_layout::compute_stub_managed_regs): Rename to...
22150 (xlogue_layout::count_stub_managed_regs): ...this.
22151 (xlogue_layout::is_stub_managed_reg): New function.
22152 (xlogue_layout::m_stub_names): Rename to...
22153 (xlogue_layout::s_stub_names): ...this, make static.
22154 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22155 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22156 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22157 xlogue_layout::s_stub_names): Instantiate statics.
22158 (stub_managed_regs): Remove.
22159 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22160 (disable_call_ms2sysv_xlogues): Rename to...
22161 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22162 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22164 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22165 change after reload_completed.
22166 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22168 (ix86_expand_prologue): Likewise.
22169 (ix86_expand_epilogue): Likewise.
22170 (ix86_expand_split_stack_prologue): Likewise.
22171 (ix86_compute_frame_layout): Remove frame parameter ...
22172 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22173 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22175 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22176 (ix86_frame): Move from here ...
22177 * config/i386/i386.h (ix86_frame): ... to here.
22178 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22179 complete ix86_frame data structure instead. Remove some_ld_name.
22181 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22183 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22184 symbols that hold a DECL_VALUE_EXPR.
22186 2017-06-01 Martin Jambor <mjambor@suse.cz>
22188 PR tree-optimization/80898
22189 * tree-sra.c (process_subtree_disqualification): Removed.
22190 (disqualify_candidate): Do not acll
22191 process_subtree_disqualification.
22192 (subtree_mark_written_and_enqueue): New function.
22193 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22194 RHS has been disqualified and re-queue LHS if necessary. Apart
22195 from that, ignore disqualified RHS.
22197 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22199 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22200 address fetch for z10 or later.
22202 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22204 * config/arc/arc.md (tst_movb): Add guard when splitting.
22206 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22208 * config/arc/arc.c (arc_can_eliminate): Test against
22209 arc_frame_pointer_needed.
22211 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22213 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22214 to prevent store reordering.
22215 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22216 (type): Add block type.
22217 (stack_tie): Define special instruction to be used in
22220 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22222 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22223 constraint. It is not valid for the pattern.
22224 (noncommutative_binary_comparison): Likewise.
22226 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22228 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22231 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22233 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22234 be used by the reg-alloc.
22236 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22238 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22239 reg-alloc when having mul64 or mul32x16 instructions.
22240 (mulsidi3): Likewise.
22241 (umulsidi3): Likewise.
22242 (mulsi32x16): New pattern.
22243 (mulsi64): Likewise.
22244 (mulsidi64): Likewise.
22245 (umulsidi64): Likewise.
22246 (MUL32x16_REG): Define.
22247 (mul64_600): Use MUL32x16_REG.
22248 (mac64_600): Likewise.
22249 (umul64_600): Likewise.
22250 (umac64_600): Likewise.
22252 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22254 * config/arc/arc.md (mulsi3_700): Make it commutative.
22256 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22258 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22260 (*sign_extendsidi2_insn): Likewise for movstosw.
22262 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22264 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22265 the type of the input discriminant value. Convert the
22266 discriminant value of signedness vary.
22268 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22270 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22271 Add to -Wall section.
22273 2017-06-01 Richard Biener <rguenther@suse.de>
22275 PR middle-end/66313
22276 * fold-const.c (fold_plusminus_mult_expr): If the factored
22277 factor may be zero use a wrapping type for the inner operation.
22278 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22279 and handle moved defs.
22280 (process_assignment): Properly guard the unary op case. Return a
22281 tri-state indicating that moving the stmt before the call may allow
22282 to continue. Pass through to_move.
22283 (find_tail_calls): Handle moving unrelated defs before
22286 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22289 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22290 splitter result in the canonical way.
22292 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22294 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22295 also for 32bit target. Update insn attributes.
22296 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22298 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22300 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22301 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22302 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22303 (_mm_maskz_min_ss): New intrinsics.
22305 2017-05-31 Martin Liska <mliska@suse.cz>
22307 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22308 Change comment style to one we normally use.
22309 (vectorizable_reduction): Likewise.
22310 (vectorizable_induction): Likewise.
22311 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22312 (vectorizable_call): Likewise.
22313 (vectorizable_simd_clone_call): Likewise.
22314 (vectorizable_conversion): Likewise.
22315 (vectorizable_assignment): Likewise.
22316 (vectorizable_shift): Likewise.
22317 (vectorizable_operation): Likewise.
22318 (vectorizable_store): Likewise.
22319 (vectorizable_load): Likewise.
22320 * tree-vectorizer.h: Likewise.
22322 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22324 * passes.c (emergency_dump_function): New.
22325 * tree-pass.h (emergency_dump_function): Declare.
22326 * plugin.c (plugins_internal_error_function): Remove.
22327 * plugin.h (plugins_internal_error_function): Remove declaration.
22328 * toplev.c (internal_error_function): New static function. Use it...
22329 (general_init): ...here.
22331 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22333 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22334 (arc_rtx_costs): Add costs for new patterns.
22335 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22336 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22338 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22340 * tree-ssa-strlen.c (get_next_strinfo): New function.
22341 (get_stridx_plus_constant): Use it.
22342 (zero_length_string): Likewise.
22343 (adjust_related_strinfos): Likewise.
22344 (adjust_last_stmt): Likewise.
22346 2017-05-31 Richard Biener <rguenther@suse.de>
22349 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22350 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22352 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22354 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22355 loop_vinfo argument and use of dependence distance vectors.
22356 Check instead whether the two references differ only in their
22357 initial value and assume that they have the same alignment if the
22358 difference is a multiple of the vector alignment.
22359 (vect_analyze_data_refs_alignment): Update call accordingly.
22361 2017-05-31 Martin Liska <mliska@suse.cz>
22364 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22366 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22368 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22369 (create_intersect_range_checks): Move from ...
22370 * tree-data-ref.c (create_intersect_range_checks_index)
22371 (create_intersect_range_checks): ... to here.
22372 (create_runtime_alias_checks): New function factored from ...
22373 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22374 here. Call above function.
22375 * tree-data-ref.h (create_runtime_alias_checks): New function.
22377 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22379 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22380 segment length for dr_b and compute it in wide_int.
22382 2017-05-31 Richard Biener <rguenther@suse.de>
22384 PR tree-optimization/80906
22385 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22386 and pass through iv_map.
22387 (copy_bb_and_scalar_dependences): Adjust.
22388 (translate_pending_phi_nodes): Likewise.
22389 (copy_loop_close_phi_args): Handle code-generating IVs instead
22392 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22394 * diagnostic-color.c (color_dict): Add "type-diff".
22395 (parse_gcc_colors): Update comment.
22396 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22397 -fdiagnostics-show-template-tree and -fno-elide-type.
22398 (GCC_COLORS): Add type-diff to example.
22400 (-fdiagnostics-show-template-tree): New.
22401 (-fno-elide-type): New.
22402 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22403 the pp_format_decoder callback. Call any m_format_postprocessor's
22405 (pretty_printer::pretty_printer): Initialize
22406 m_format_postprocessor.
22407 (pretty_printer::~pretty_printer): Delete any
22408 m_format_postprocessor.
22409 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22410 (class format_postprocessor): New class.
22411 (struct pretty_printer::format_decoder): Document the new parameters.
22412 (struct pretty_printer::m_format_postprocessor): New field.
22413 * tree-diagnostic.c (default_tree_printer): Update for new
22414 bool and const char ** params.
22415 * tree-diagnostic.h (default_tree_printer): Likewise.
22417 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22419 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22420 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22421 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22422 rs6000_gen_cell_microcode code.
22423 (rs6000_final_prescan_insn): Delete.
22424 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22425 "warn-cell-microcode" entries.
22426 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22427 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22428 throughout. Change cc_reg_not_micro_cr0_operand to
22429 cc_reg_not_cr0_operand throughout.
22430 (*extendhi<mode>2_noload): Delete.
22431 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22432 (mwarn-cell-microcode): Delete.
22433 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22434 -mgen-cell-microcode and -mwarn-cell-microcode.
22436 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22439 * config/i386/constraints.md (Yd): New constraint.
22441 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22442 and (?Yd, r) alternatives. Update insn attributes.
22443 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22444 and (?*Yd, r) alternatives. Update insn attributes.
22445 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22447 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22449 * gimplify.c (gimplify_modify_expr): Don't create a
22450 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22453 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22455 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22457 2017-05-30 Richard Biener <rguenther@suse.de>
22459 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22460 and reduc_def fields.
22461 (STMT_VINFO_REDUC_TYPE): New define.
22462 (STMT_VINFO_REDUC_DEF): Likewise.
22463 (vect_force_simple_reduction): Adjust prototype.
22464 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22465 (vect_is_simple_reduction): Remove check_reduction argument.
22466 (vect_force_simple_reduction): Adjust and set
22467 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22468 (vectorizable_reduction): Do not re-do reduction analysis
22469 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22470 * tree-parloops.c (gather_scalar_reductions): Adjust.
22472 2017-05-30 Richard Biener <rguenther@suse.de>
22474 PR middle-end/80901
22475 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22478 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22480 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22481 Introduce unknown_misalignment parameter and remove vf.
22482 (vect_peeling_hash_get_lowest_cost):
22483 Pass unknown_misalignment parameter.
22484 (vect_enhance_data_refs_alignment):
22485 Fix unsupportable data ref treatment.
22487 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22489 * tree-vect-data-refs.c (vect_get_data_access_cost):
22490 Workaround for SLP handling.
22491 (vect_enhance_data_refs_alignment):
22492 Compute costs for doing no peeling at all, compare to the best
22493 peeling costs so far and avoid peeling if cheaper.
22495 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22497 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22498 Return peeling info and set costs to zero for unlimited cost
22500 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22501 with unknown misalignment. Compute and costs for unknown
22502 misalignment, compare them to the costs for known misalignment
22503 and choose the cheapest for peeling.
22505 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22507 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22508 (vect_get_peeling_costs_all_drs): Create function.
22509 (vect_peeling_hash_get_lowest_cost):
22510 Use vect_get_peeling_costs_all_drs.
22511 (vect_peeling_supportable): Create function.
22512 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22514 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22516 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22517 DR_HAS_NEGATIVE_STEP.
22518 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22519 (vect_enhance_data_refs_alignment): Use.
22520 (vect_duplicate_ssa_name_ptr_info): Use.
22521 * tree-vectorizer.h (dr_misalignment): Use.
22522 (known_alignment_for_access_p): Use.
22524 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22527 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22529 (has_section_name): New function.
22531 2017-05-30 Martin Liska <mliska@suse.cz>
22534 * auto-profile.c (get_function_decl_from_block): Fix
22537 2017-05-30 Richard Biener <rguenther@suse.de>
22539 PR middle-end/80876
22540 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22542 2017-05-30 Martin Liska <mliska@suse.cz>
22544 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22545 * dumpfile.h (struct dump_file_info): Remove ctors.
22547 2017-05-30 Martin Liska <mliska@suse.cz>
22549 * predict.def: Fix GNU coding style.
22551 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22553 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22554 Mark 'to' argument with ATTRIBUTE_UNUSED.
22556 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22558 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22559 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22560 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22563 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22565 * doc/install.texi (Options specification): Restore entry of
22566 --enable-sjlj-exceptions.
22568 2017-05-27 Michael Eager <eager@eagercon.com>
22571 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22573 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22575 * config/microblaze/microblaze.h
22576 (FIXED_REGISTERS): Update in macro.
22577 (CALL_USED_REGISTERS): Update in macro.
22579 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22581 * doc/install.texi: Add links to macOS binary distributions.
22583 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22587 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22589 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22591 2017-05-26 Martin Liska <mliska@suse.cz>
22593 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22595 2017-05-26 Martin Liska <mliska@suse.cz>
22597 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22598 always leading ';; '.
22599 (dump_bb_info): Likewise.
22600 (brief_dump_cfg): Likewise.
22601 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22602 * dumpfile.c: Remove usage of TDF_VERBOSE.
22603 * dumpfile.h (enum dump_kind): Likewise.
22604 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22605 * print-tree.c (debug_verbose): Remove.
22606 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22607 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22608 * tree-diagnostic.c (default_tree_printer): Replace
22609 TDF_DIAGNOSTIC with TDF_SLIM.
22611 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22613 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22614 in parameter loop, rather than loop_vinfo.
22615 (create_intersect_range_checks): Ditto.
22616 (vect_create_cond_for_alias_checks): Update call to above functions.
22618 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22620 PR tree-optimization/80815
22621 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22622 for merging runtime alias checks. Handle negative DR_STEPs.
22624 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22626 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22628 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22629 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22630 out code pruning runtime alias checks.
22631 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22632 factored out from above.
22633 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22635 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22637 (prune_runtime_alias_test_list): New decalaration.
22639 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22641 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22642 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22643 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22644 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22645 (operator==, comp_dr_with_seg_len_pair): Ditto.
22646 (vect_prune_runtime_alias_test_list): Ditto.
22648 2017-05-26 Martin Liska <mliska@suse.cz>
22651 * params.def: Bound partial-inlining-entry-probability param.
22653 2017-05-26 Marek Polacek <polacek@redhat.com>
22656 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22659 2017-05-26 Richard Biener <rguenther@suse.de>
22661 PR tree-optimization/80842
22662 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22665 2017-05-26 Richard Biener <rguenther@suse.de>
22667 PR tree-optimization/80844
22668 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22670 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22672 * doc/md.texi (Machine Constraints): Update x86 family
22673 machine constraints section to match 'config/i386/constraints.md'.
22675 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22677 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22679 2017-05-25 Nathan Sidwell <nathan@acm.org>
22681 * doc/invoke.texi (--enable-languages): Update documentation.
22683 2017-05-25 Martin Liska <mliska@suse.cz>
22685 * dumpfile.c: Add TDF_FOLDING.
22686 * dumpfile.h (enum dump_kind): Likewise.
22687 * genmatch.c (dt_simplify::gen_1): Use it.
22689 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22691 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22693 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22695 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22696 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22698 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22700 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22701 * match.pd (X == C): Rewrite it here.
22702 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22703 with_certain_nonzero_bits2): New predicates.
22704 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22706 2017-05-24 Nathan Sidwell <nathan@acm.org>
22708 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22711 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22714 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22716 * config/powerpcspe: New port. Files are copied from the rs6000
22717 port, with "rs6000" in filenames replaced by "powerpcspe".
22719 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22721 PR rtl-optimization/80754
22722 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22724 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22726 * config/sparc/sparc.md (length): Return the correct value for -mflat
22727 sibcalls to match output_sibcall.
22729 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22733 * config/rs6000/rs6000.c (struct machine_function): Add new field
22735 (rs6000_get_separate_components): Init that field, use it.
22736 (rs6000_components_for_bb): Use the field.
22738 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22740 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22742 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22744 PR middle-end/80823
22745 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22747 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22750 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22751 address against address_operand predicate.
22752 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22754 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22756 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22757 parameters passed indirectly.
22759 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22761 * config/i386/i386.md (*movdi_internal): Remove SSE4
22762 alternative 18 (?r, *v). Update insn attributes.
22763 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22764 Update insn attributes.
22765 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22766 Update insn attributes.
22767 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22768 alternative 1 (r, v). Remove isa attribute.
22769 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22770 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22771 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22773 2017-05-23 Tom de Vries <tom@codesourcery.com>
22775 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22778 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22780 * cgraphunit.c (symbol_table::process_new_functions): Update.
22781 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22782 (inline_generate_summary): Rename to ...
22783 (ipa_fn_summary_generate): ... this one.
22784 (inline_read_summary): Rename to ...
22785 (ipa_fn_summary_read): ... this one.
22786 (inline_write_summary): Rename to ...
22787 (ipa_fn_summary_write): ... this one.
22788 (inline_free_summary): Rename to ...
22789 (ipa_free_fn_summary): ... this one.
22790 (pass_data_local_fn_summary, pass_local_fn_summary,
22791 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22792 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22793 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22794 make_pass_ipa_fn_summary): New.
22795 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22796 inline_write_summary, inline_free_summary): Remove.
22797 (ipa_free_fn_summary) : New.
22798 * ipa-inline.c (ipa_inline): Update.
22799 (pass_ipa_inline): Do not generate summaries.
22800 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22802 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22803 and add pass_ipa_fn_summary.
22804 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22806 (make_pass_inline_parameters): Remove.
22808 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22810 * omp-low.c (struct omp_context): Remove "default_kind" member.
22813 * omp-offload.c (execute_oacc_device_lower): Remove the
22814 parallelism dimensions function attributes for unparallelized
22815 OpenACC kernels constructs.
22817 2017-05-23 Martin Liska <mliska@suse.cz>
22819 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22821 (cgraph_edge::make_speculative): Likewise.
22822 (cgraph_edge::resolve_speculation): Likewise.
22823 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22824 (cgraph_node::dump): Likewise.
22825 * cgraph.h: Likewise.
22826 * cgraphunit.c (analyze_functions): Likewise.
22827 (symbol_table::compile): Likewise.
22828 * ipa-cp.c (print_all_lattices): Likewise.
22829 (determine_versionability): Likewise.
22830 (initialize_node_lattices): Likewise.
22831 (ipcp_verify_propagated_values): Likewise.
22832 (estimate_local_effects): Likewise.
22833 (update_profiling_info): Likewise.
22834 (create_specialized_node): Likewise.
22835 (perhaps_add_new_callers): Likewise.
22836 (decide_about_value): Likewise.
22837 (decide_whether_version_node): Likewise.
22838 (identify_dead_nodes): Likewise.
22839 (ipcp_store_bits_results): Likewise.
22840 * ipa-devirt.c (dump_targets): Likewise.
22841 (ipa_devirt): Likewise.
22842 * ipa-icf.c (sem_item::dump): Likewise.
22843 (sem_function::equals): Likewise.
22844 (sem_variable::equals): Likewise.
22845 (sem_item_optimizer::read_section): Likewise.
22846 (sem_item_optimizer::execute): Likewise.
22847 (congruence_class::dump): Likewise.
22848 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22849 (dump_inline_summary): Likewise.
22850 (estimate_node_size_and_time): Likewise.
22851 (inline_analyze_function): Likewise.
22852 * ipa-inline-transform.c (inline_call): Likewise.
22853 * ipa-inline.c (report_inline_failed_reason): Likewise.
22854 (want_early_inline_function_p): Likewise.
22855 (edge_badness): Likewise.
22856 (update_edge_key): Likewise.
22857 (inline_small_functions): Likewise.
22858 * ipa-profile.c (ipa_profile): Likewise.
22859 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22860 (ipa_make_edge_direct_to_target): Likewise.
22861 (remove_described_reference): Likewise.
22862 (ipa_impossible_devirt_target): Likewise.
22863 (propagate_controlled_uses): Likewise.
22864 (ipa_print_node_params): Likewise.
22865 (ipcp_transform_function): Likewise.
22866 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22867 (propagate_pure_const): Likewise.
22868 * ipa-reference.c (generate_summary): Likewise.
22869 (read_write_all_from_decl): Likewise.
22870 (propagate): Likewise.
22871 (ipa_reference_read_optimization_summary): Likewise.
22872 * ipa-utils.c (ipa_merge_profiles): Likewise.
22873 * ipa.c (walk_polymorphic_call_targets): Likewise.
22874 (symbol_table::remove_unreachable_nodes): Likewise.
22875 (ipa_single_use): Likewise.
22876 * passes.c (execute_todo): Likewise.
22877 * predict.c (drop_profile): Likewise.
22878 * symtab.c (symtab_node::get_dump_name): New function.
22879 (symtab_node::dump_name): Likewise.
22880 (symtab_node::dump_asm_name): Likewise.
22881 (symtab_node::dump_references): Likewise.
22882 (symtab_node::dump_referring): Likewise.
22883 (symtab_node::dump_base): Likewise.
22884 (symtab_node::debug_symtab): Likewise.
22885 * tree-sra.c (convert_callers_for_node): Likewise.
22886 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22887 * value-prof.c (init_node_map): Likewise.
22889 2017-05-23 Martin Liska <mliska@suse.cz>
22891 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22892 and symtab_node::debug_symtab to symbol_table::debug.
22893 * cgraphunit.c (analyze_functions): Use the renamed function.
22894 (symbol_table::compile): Likewise.
22895 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22896 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22897 * passes.c (execute_todo): Likewise.
22898 * symtab.c (symbol_table::dump): New function.
22899 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22901 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22903 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22904 that nonconst implies exec.
22906 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22908 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22909 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22910 (inline_edge_summary_vec): Turn into ...
22911 (ipa_call_summaries): ... this one.
22912 (redirect_to_unreachable, edge_set_predicate,
22913 evaluate_properties_for_edge, inline_summary_alloc,
22914 reset_ipa_call_summary, reset_inline_summary,
22915 inline_summary_t::duplicate): Update.
22916 (inline_edge_duplication_hook): Turn to ...
22917 (ipa_call_summary_t::duplicate): ... this one.
22918 (inline_edge_removal_hook): Turn to ...
22919 (ipa_call_summary_t::remove): ... this one.
22920 (dump_inline_edge_summary): Turn to ...
22921 (dump_ipa_call_summary): ... this one.
22922 (estimate_function_body_sizes): Update.
22923 (inline_update_callee_summaries): Update.
22924 (remap_edge_change_prob): Update.
22925 (remap_edge_summaries): Update.
22926 (inline_merge_summary): Update.
22927 (do_estimate_edge_time): Update.
22928 (inline_generate_summary): Update.
22929 (inline_read_section): Update.
22930 (inline_read_summary): Update.
22931 (inline_free_summary): Update.
22932 * ipa-inline.c (can_inline_edge_p): Update.
22933 (compute_inlined_call_time): Update.
22934 (want_inline_small_function_p): Update.
22935 (edge_badness): Update.
22936 (early_inliner): Update.
22937 * ipa-inline.h (inline_edge_summary): Turn to ...
22938 (ipa_call_summary): ... this one.
22939 (ipa_call_summary_t): New class.
22940 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22941 (ipa_call_summaries): New.
22942 (inline_edge_summary): Remove.
22943 (estimate_edge_growth): Update.
22944 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22945 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22946 * ipa-split.c (execute_split_functions): Update.
22947 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22949 2017-05-23 Tom de Vries <tom@codesourcery.com>
22951 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22952 attributes): Document rdrand effective target.
22954 2017-05-23 Tom de Vries <tom@codesourcery.com>
22956 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22957 attributes): Sort alphabetically.
22959 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22961 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22963 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22966 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22967 V2DF/V2DI splat into two separate patterns, one that handles
22968 registers, and the other that only handles memory. Drop support
22969 for splatting from a GPR on ISA 2.07 and then splitting the
22970 splat into direct move and splat.
22971 (vsx_splat_<mode>_reg): Likewise.
22972 (vsx_splat_<mode>_mem): Likewise.
22974 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22976 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22978 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22980 PR middle-end/80809
22981 * omp-low.c (finish_taskreg_remap): New function.
22982 (finish_taskreg_scan): If unit size of ctx->record_type
22983 is non-constant, unshare the size expression and replace
22984 decls in it with possible outer var refs.
22986 PR middle-end/80809
22987 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22988 GOVD_SHARED rather than GOVD_PRIVATE with it.
22989 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22990 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22992 PR middle-end/80853
22993 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22994 as last argument to build_outer_var_ref for pointer bases of array
22995 section reductions.
22997 2017-05-19 Martin Sebor <msebor@redhat.com>
22999 * print-tree.c (print_node): Print DECL_READ_P flag.
23001 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23003 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23004 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23005 * cgraph.c: Likewise.
23006 * cgraphunit.c: Likewise.
23007 * gengtype.c: Likewise.
23008 * ipa-cp.c: Likewise.
23009 * ipa-devirt.c: Likewise.
23010 * ipa-icf.c: Likewise.
23011 * ipa-predicate.c: Likewise.
23012 * ipa-profile.c: Likewise.
23013 * ipa-prop.c: Likewise.
23014 * ipa-split.c: Likewise.
23016 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23017 edge_predicate_pool, dump_inline_hints,
23018 inline_summary::account_size_time, redirect_to_unreachable,
23019 edge_set_predicate, set_hint_predicate,
23020 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23021 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23022 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23023 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23024 ipa_call_summary_t::remove, initialize_growth_caches,
23025 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23026 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23027 mark_modified, unmodified_parm_1, unmodified_parm,
23028 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23029 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23030 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23031 will_be_nonconstant_predicate, record_modified_bb_info,
23032 get_minimal_bb, record_modified, param_change_prob,
23033 phi_result_unknown_predicate, predicate_for_phi_result,
23034 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23035 estimate_function_body_sizes, compute_inline_parameters,
23036 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23037 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23038 inline_update_callee_summaries, remap_edge_change_prob,
23039 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23040 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23041 inline_analyze_function, inline_summary_t::insert,
23042 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23043 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23044 inline_free_summary): Move to ipa-fnsummary.h
23045 (predicate_t): Remove.
23046 * ipa-fnsummary.c: New file.
23047 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23048 (enum inline_hints_vals, inline_hints, agg_position_info,
23049 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23050 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23051 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23052 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23053 inline_read_summary, inline_write_summary, inline_free_summary,
23054 inline_analyze_function, initialize_inline_failed,
23055 inline_merge_summary, inline_update_overall_summary,
23056 compute_inline_parameters): Move to ipa-fnsummary.h
23057 * ipa-fnsummary.h: New file.
23058 * ipa-inline-transform.h: Include ipa-inline.h.
23059 * ipa-inline.c: LIkewise.
23061 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23063 * ipa-inline.c (edge_badness): Use inlined_time instead of
23064 inline_summaries->get.
23066 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23068 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23070 2017-05-22 Nathan Sidwell <nathan@acm.org>
23072 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23073 (fdump-lang): Document 'raw' option.
23074 * dumpfile.h (TDI_tu): Delete.
23075 * dumpfile.c (dump_files): Remove translation-unit.
23076 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23078 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23080 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23081 command option from $(AWK) call.
23082 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23083 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23084 [FORMAT]: Remove handling of variable.
23085 * config/avr/t-multilib: Regenerate.
23087 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23089 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23091 (dump_inline_summary): Do not print self_time.
23092 (estimate_function_body_sizes): Do not set self_time.
23093 (compute_inline_parameters): Likewise.
23094 (inline_read_section, inline_write_summary): Do not stream self_time.
23095 * ipa-inline.h (inline_summary): Drop self_time.
23097 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23099 * ipa-inline-analysis.c (account_size_time): Rename to ...
23100 (inline_summary::account_size_time): ... this one.
23101 (reset_ipa_call_summary): Turn to ...
23102 (ipa_call_summary::reset): ... this one.
23103 (reset_inline_summary): Turn to ...
23104 (inline_summary::reset): ... this one.
23105 (inline_summary_t::remove): Update.
23106 (inline_summary_t::duplicate): Update.
23107 (ipa_call_summary_t::remove): Update.
23108 (dump_inline_summary): Update.
23109 (estimate_function_body_sizes): Update.
23110 (compute_inline_parameters): Update.
23111 (estimate_node_size_and_time): Update.
23112 (inline_merge_summary): Update.
23113 (inline_update_overall_summary): Update.
23114 (inline_read_section): Update.
23115 (inline_write_summary): Update.
23116 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23117 add account_size_time and reset member functions.
23118 (ipa_call_summary): Add reset function.
23119 * ipa-predicate.h (predicate::operator &): Constify.
23121 2017-05-22 Richard Biener <rguenther@suse.de>
23123 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23125 2017-05-19 Jason Merrill <jason@redhat.com>
23127 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23129 2017-05-19 Marek Polacek <polacek@redhat.com>
23132 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23133 TYPE_OVERFLOW_WRAPS checks.
23135 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23137 * tree-core.h (enum omp_clause_default_kind): Add
23138 "OMP_CLAUSE_DEFAULT_PRESENT".
23139 * tree-pretty-print.c (dump_omp_clause): Handle it.
23140 * gimplify.c (enum gimplify_omp_var_data): Add
23141 "GOVD_MAP_FORCE_PRESENT".
23142 (gimplify_adjust_omp_clauses_1): Map it to
23143 "GOMP_MAP_FORCE_PRESENT".
23144 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23146 * gimplify.c (oacc_default_clause): Clarify.
23148 2017-05-19 Nathan Sidwell <nathan@acm.org>
23150 LANG_HOOK_REGISTER_DUMPS
23151 * toplev.c (general_init): Call register dump lang hook.
23152 * doc/invoke.texi: Document -fdump-lang option family.
23153 * dumpfile.c (dump_files): Remove class dump here.
23154 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23155 * dumpfile.h (tree_dump_index): Remove TDI_class.
23156 * langhooks-def.h (lhd_register_dumps): Declare.
23157 (LANG_HOOKS_REGISTER_DUMPS): Define.
23158 (LANG_HOOKS_INITIALIZER): Add it.
23159 * langhooks.c (lhd_register_dumps): Define.
23160 * langhooks.h (struct lang_hooks): Add register_dumps.
23162 2017-05-19 Nathan Sidwell <nathan@acm.org>
23164 * context.h (context::set_passes): New.
23165 * context.c (context::context): Do not create pass manager.
23166 * toplev.c (general_init): Create pass manager here.
23168 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23170 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23171 use this splitter if two add or or instructions would also work for
23172 the constant we want to generate.
23174 2017-05-19 Richard Biener <rguenther@suse.de>
23177 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23178 predicate evaluation.
23180 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23182 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23184 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23187 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23189 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23190 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23191 (inline_edge_summary_vec): Turn into ...
23192 (ipa_call_summaries): ... this one.
23193 (redirect_to_unreachable, edge_set_predicate,
23194 evaluate_properties_for_edge, inline_summary_alloc,
23195 reset_ipa_call_summary, reset_inline_summary,
23196 inline_summary_t::duplicate): Update.
23197 (inline_edge_duplication_hook): Turn to ...
23198 (ipa_call_summary_t::duplicate): ... this one.
23199 (inline_edge_removal_hook): Turn to ...
23200 (ipa_call_summary_t::remove): ... this one.
23201 (dump_inline_edge_summary): Turn to ...
23202 (dump_ipa_call_summary): ... this one.
23203 (estimate_function_body_sizes): Update.
23204 (inline_update_callee_summaries): Update.
23205 (remap_edge_change_prob): Update.
23206 (remap_edge_summaries): Update.
23207 (inline_merge_summary): Update.
23208 (do_estimate_edge_time): Update.
23209 (inline_generate_summary): Update.
23210 (inline_read_section): Update.
23211 (inline_read_summary): Update.
23212 (inline_free_summary): Update.
23213 * ipa-inline.c (can_inline_edge_p): Update.
23214 (compute_inlined_call_time): Update.
23215 (want_inline_small_function_p): Update.
23216 (edge_badness): Update.
23217 (early_inliner): Update.
23218 * ipa-inline.h (inline_edge_summary): Turn to ...
23219 (ipa_call_summary): ... this one.
23220 (ipa_call_summary_t): New class.
23221 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23222 (ipa_call_summaries): New.
23223 (inline_edge_summary): Remove.
23224 (estimate_edge_growth): Update.
23225 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23226 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23227 * ipa-split.c (execute_split_functions): Update.
23228 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23230 2017-05-19 Richard Biener <rguenther@suse.de>
23232 PR middle-end/80764
23233 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23235 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23237 * config/rs6000/rs6000.c (struct machine_function): Add field
23238 fpr_is_wrapped_separately.
23239 (rs6000_get_separate_components): Use 64 components. Handle the
23240 new FPR components.
23241 (rs6000_components_for_bb): Handle the FPR components.
23242 (rs6000_emit_prologue_components): Handle the FPR components.
23243 (rs6000_emit_epilogue_components): Handle the FPR components.
23244 (rs6000_set_handled_components): Handle the FPR components.
23245 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23246 that are already separately shrink-wrapped.
23247 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23248 that are already separately shrink-wrapped.
23250 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23253 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23256 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23257 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23258 cases where the register allocator uses a move and an offsettable
23259 memory operation to/from a FPR register on ISA 2.06/2.07.
23260 (define_peephole2 for Altivec d-form store): Likewise.
23262 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23265 * config/i386/mmx.md (*mov<mode>_internal): Enable
23266 alternatives 11, 12, 13 and 14 also for 32bit targets.
23267 Remove alternatives 15, 16, 17 and 18.
23268 * config/i386/sse.md (vec_concatv2di): Change
23269 alternative (!x, *y) to (x, ?!*Yn).
23271 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23273 * dumpfile.h (enum dump_kind): Remove stray comma.
23275 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23277 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23278 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23279 predicate::num_conditions
23280 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23281 (CHANGED): turn into predicate::changed.
23282 (agg_position_info): Move to ipa-predicate.h
23283 (add_condition, predicate::add_clause, predicate::operator &=,
23284 predicate::or_with, predicate::evaluate, predicate::probability,
23285 dump_condition, dump_clause, predicate::dump,
23286 predicate::remap_after_duplication, predicate::remap_after_inlining,
23287 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23288 (evaluate_conditions_for_known_args): Update.
23289 (set_cond_stmt_execution_predicate): Update.
23290 * ipa-inline.h: Include ipa-predicate.h
23291 (condition, inline_param_summary, conditions, agg_position_info,
23292 predicate): Move to ipa-predicate.h
23293 * ipa-predicate.c: New file.
23294 * ipa-predicate.h: New file.
23296 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23298 * final.c (leaf_function_p): Check we are not in a sequence.
23300 2017-05-18 Martin Liska <mliska@suse.cz>
23302 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23303 * dumpfile.c (dump_register): Use new enum dump_kind.
23304 (get_dump_file_name): Likewise.
23305 (dump_enable_all): Likewise.
23306 (dump_switch_p_1): Likewise.
23307 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23308 * dumpfile.h (enum dump_kind): New enum type.
23309 (struct dump_file_info): Create constructor and
23310 format fields and comments.
23311 * passes.c (pass_manager::register_one_dump_file):
23313 * statistics.c (statistics_early_init): Likewise.
23314 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23315 TDF_TREE with TDF_SLIM.
23316 (gather_memory_references_ref): Likewise.
23318 2017-05-18 Martin Liska <mliska@suse.cz>
23320 * vec.h (struct vnull): Use it.
23322 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23324 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23325 (true_predicate, false_predicate, true_predicate_p,
23326 false_predicate_p): Remove.
23327 (single_cond_predicate, not_inlined_predicate): Turn to member function
23329 (add_condition): Update.
23330 (add_clause): Turn to...
23331 (predicate::add_clause): ... this one; update; allow passing NULL
23333 (and_predicates): Turn to ...
23334 (predicate::operator &=): ... this one.
23335 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23336 (or_predicates): Turn to ...
23337 (predicate::or_with): ... this one.
23338 (evaluate_predicate): Turn to ...
23339 (predicate::evaluate): ... this one.
23340 (predicate_probability): Turn to ...
23341 (predicate::probability): ... this one.
23342 (dump_condition): Update.
23343 (dump_predicate): Turn to ...
23344 (predicate::dump): ... this one.
23345 (account_size_time): Update.
23346 (edge_set_predicate): Update.
23347 (set_hint_predicate): UPdate.
23348 (evaluate_conditions_for_known_args): Update.
23349 (evaluate_properties_for_edge): Update.
23350 (remap_predicate_after_duplication): Turn to...
23351 (predicate::remap_after_duplication): ... this one.
23352 (remap_hint_predicate_after_duplication): Update.
23353 (inline_summary_t::duplicate): UPdate.
23354 (dump_inline_edge_summary): Update.
23355 (dump_inline_summary): Update.
23356 (set_cond_stmt_execution_predicate): Update.
23357 (set_switch_stmt_execution_predicate): Update.
23358 (compute_bb_predicates): Update.
23359 (will_be_nonconstant_expr_predicate): Update.
23360 (will_be_nonconstant_predicate): Update.
23361 (phi_result_unknown_predicate): Update.
23362 (predicate_for_phi_result): Update.
23363 (array_index_predicate): Update.
23364 (estimate_function_body_sizes): Update.
23365 (estimate_node_size_and_time): Update.
23366 (estimate_ipcp_clone_size_and_time): Update.
23367 (remap_predicate): Rename to ...
23368 (predicate::remap_after_inlining): ... this one.
23369 (remap_hint_predicate): Update.
23370 (inline_merge_summary): Update.
23371 (inline_update_overall_summary): Update.
23372 (estimate_size_after_inlining): Update.
23373 (read_predicate): Rename to ...
23374 (predicate::stream_in): ... this one.
23375 (read_inline_edge_summary): Update.
23376 (write_predicate): Rename to ...
23377 (predicate::stream_out): ... this one.
23378 (write_inline_edge_summary): Update.
23379 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23380 (clause_t): Turn to uint32_t
23381 (predicate): Turn to class; implement constructor and operators
23383 (size_time_entry): Update.
23384 (inline_summary): Update.
23385 (inline_edge_summary): Update.
23387 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23389 * fold-const.c (fold_binary_loc): Move transformation...
23390 * match.pd (C - X CMP X): ... here.
23392 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23394 * config/sparc/sparc.c (sparc_option_override): Set function
23395 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23396 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23398 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23400 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23402 2017-05-18 Marek Polacek <polacek@redhat.com>
23405 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23406 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23407 gimple_assign_load_p.
23409 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23411 PR middle-end/80692
23412 * real.c (do_compare): Give decimal_do_compare preference over
23413 comparing just the signs.
23415 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23417 * doc/md.texi (Canonicalization of Instructions): Describe the
23418 canonical form of instructions that inherently set a condition
23421 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23423 PR middle-end/80775
23424 * tree-cfg.c: Move deletion of unreachable case statements to after
23425 the merging of consecutive case labels.
23427 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23429 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23430 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23431 restoring of callee-saved registers.
23433 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23435 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23436 * config/visium/visium.c (single_set_and_flags): Likewise.
23437 * config/visium/visium.md (Substitutions): Likewise.
23439 2017-05-17 Martin Liska <mliska@suse.cz>
23441 * cfg.c: Introduce dump_flags_t type and
23442 use it instead of int type.
23444 * cfghooks.c: Likewise.
23445 * cfghooks.h (struct cfg_hooks): Likewise.
23446 * cfgrtl.c: Likewise.
23447 * cfgrtl.h: Likewise.
23448 * cgraph.c (cgraph_node::get_body): Likewise.
23449 * coretypes.h: Likewise.
23450 * domwalk.c: Likewise.
23451 * domwalk.h: Likewise.
23452 * dumpfile.c (struct dump_option_value_info): Likewise.
23453 (dump_enable_all): Likewise.
23454 (dump_switch_p_1): Likewise.
23455 (opt_info_switch_p): Likewise.
23456 * dumpfile.h (enum tree_dump_index): Likewise.
23457 (struct dump_file_info): Likewise.
23458 * genemit.c: Likewise.
23459 * generic-match-head.c: Likewise.
23460 * gengtype.c (open_base_files): Likewise.
23461 * gimple-pretty-print.c: Likewise.
23462 * gimple-pretty-print.h: Likewise.
23463 * graph.c (print_graph_cfg): Likewise.
23464 * graphite-scop-detection.c (dot_all_sese): Likewise.
23465 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23466 * loop-unroll.c (report_unroll): Likewise.
23467 * passes.c (pass_manager::register_one_dump_file): Likewise.
23468 * print-tree.c: Likewise.
23469 * statistics.c: Likewise.
23470 * tree-cfg.c: Likewise.
23471 * tree-cfg.h: Likewise.
23472 * tree-dfa.c: Likewise.
23473 * tree-dfa.h: Likewise.
23474 * tree-dump.c (dump_function): Likewise.
23475 * tree-dump.h (struct dump_info): Likewise.
23476 * tree-pretty-print.c: Likewise.
23477 * tree-pretty-print.h: Likewise.
23478 * tree-ssa-live.c: Likewise.
23479 * tree-ssa-live.h: Likewise.
23480 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23481 * tree-vect-loop.c: Likewise.
23482 * tree-vect-slp.c: Likewise.
23484 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23485 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23487 PR tree-optimization/80457
23488 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23489 of all arguments to a statement as scalar_to_vec operations.
23490 (vectorizable_call): Adjust call to vect_model_simple_cost for
23492 (vectorizable_conversion): Likewise.
23493 (vectorizable_assignment): Likewise.
23494 (vectorizable_shift): Likewise.
23495 (vectorizable_operation): Likewise.
23496 (vectorizable_comparison): Likewise.
23497 (vect_is_simple_cond): Record the def types for operands.
23498 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23499 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23500 for statement argument count.
23502 2017-05-16 Carl Love <cel@us.ibm.com>
23504 * config/rs6000/rs6000-c: Add support for built-in functions
23505 vector unsigned long long vec_bperm (vector unsigned long long,
23506 vector unsigned char)
23507 vector signed long long vec_mule (vector signed int,
23509 vector unsigned long long vec_mule (vector unsigned int,
23510 vector unsigned int)
23511 vector signed long long vec_mulo (vector signed int,
23513 vector unsigned long long vec_mulo (vector unsigned int,
23514 vector unsigned int)
23515 vector signed char vec_sldw (vector signed char,
23516 vector signed char,
23518 vector unsigned char vec_sldw (vector unsigned char,
23519 vector unsigned char,
23521 vector signed short vec_sldw (vector signed short,
23522 vector signed short,
23524 vector unsigned short vec_sldw (vector unsigned short,
23525 vector unsigned short,
23527 vector signed int vec_sldw (vector signed int,
23530 vector unsigned int vec_sldw (vector unsigned int,
23531 vector unsigned int,
23533 vector signed long long vec_sldw (vector signed long long,
23534 vector signed long long,
23536 vector unsigned long long vec_sldw (vector unsigned long long,
23537 vector unsigned long long,
23539 * config/rs6000/rs6000-c: Add support for built-in functions
23540 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23541 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23542 * doc/extend.texi: Update the built-in documentation for the
23543 new built-in functions.
23545 2017-05-16 Marek Polacek <polacek@redhat.com>
23549 * tree.c (save_expr): Don't fold the expression.
23551 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23553 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23554 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23555 and (?*y,m). Update insn attributes.
23557 2017-05-16 Martin Liska <mliska@suse.cz>
23559 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23560 flags argument of print_gimple_stmt, print_gimple_expr,
23561 print_generic_stmt and print_generic_expr.
23562 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23563 * coretypes.h: Likewise.
23564 * except.c (dump_eh_tree): Likewise.
23565 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23566 * gimple-pretty-print.h: Likewise.
23567 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23568 (backprop::push_to_worklist): Likewise.
23569 (backprop::pop_from_worklist): Likewise.
23570 (backprop::process_use): Likewise.
23571 (backprop::intersect_uses): Likewise.
23572 (note_replacement): Likewise.
23573 * gimple-ssa-store-merging.c
23574 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23575 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23576 (pass_store_merging::execute): Likewise.
23577 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23578 (ssa_base_cand_dump_callback): Likewise.
23579 (dump_incr_vec): Likewise.
23580 (replace_refs): Likewise.
23581 (replace_mult_candidate): Likewise.
23582 (create_add_on_incoming_edge): Likewise.
23583 (create_phi_basis): Likewise.
23584 (insert_initializers): Likewise.
23585 (all_phi_incrs_profitable): Likewise.
23586 (introduce_cast_before_cand): Likewise.
23587 (replace_one_candidate): Likewise.
23588 * gimplify.c (gimplify_expr): Likewise.
23589 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23590 (set_rename): Likewise.
23591 (rename_uses): Likewise.
23592 (copy_loop_phi_nodes): Likewise.
23593 (add_close_phis_to_merge_points): Likewise.
23594 (copy_loop_close_phi_args): Likewise.
23595 (copy_cond_phi_args): Likewise.
23596 (graphite_copy_stmts_from_block): Likewise.
23597 (translate_pending_phi_nodes): Likewise.
23598 * graphite-poly.c (print_pdr): Likewise.
23599 (dump_gbb_cases): Likewise.
23600 (dump_gbb_conditions): Likewise.
23601 (print_scop_params): Likewise.
23602 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23603 (build_cross_bb_scalars_use): Likewise.
23604 (gather_bbs::before_dom_children): Likewise.
23605 * hsa-dump.c (dump_hsa_immed): Likewise.
23606 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23607 (get_replacement_map): Likewise.
23608 * ipa-inline-analysis.c (dump_condition): Likewise.
23609 (estimate_function_body_sizes): Likewise.
23610 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23611 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23612 * ipa-prop.c (ipa_dump_param): Likewise.
23613 (ipa_print_node_jump_functions_for_edge): Likewise.
23614 (ipa_modify_call_arguments): Likewise.
23615 (ipa_modify_expr): Likewise.
23616 (ipa_dump_param_adjustments): Likewise.
23617 (ipa_dump_agg_replacement_values): Likewise.
23618 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23619 * ipa-pure-const.c (check_stmt): Likewise.
23620 (pass_nothrow::execute): Likewise.
23621 * ipa-split.c (execute_split_functions): Likewise.
23622 * omp-offload.c (dump_oacc_loop_part): Likewise.
23623 (dump_oacc_loop): Likewise.
23624 * trans-mem.c (tm_log_emit): Likewise.
23625 (tm_memopt_accumulate_memops): Likewise.
23626 (dump_tm_memopt_set): Likewise.
23627 (dump_tm_memopt_transform): Likewise.
23628 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23629 (print_loop): Likewise.
23630 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23631 (chkp_gather_checks_info): Likewise.
23632 (chkp_get_check_result): Likewise.
23633 (chkp_remove_check_if_pass): Likewise.
23634 (chkp_use_outer_bounds_if_possible): Likewise.
23635 (chkp_reduce_bounds_lifetime): Likewise.
23636 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23637 (chkp_mark_completed_bounds): Likewise.
23638 (chkp_register_incomplete_bounds): Likewise.
23639 (chkp_mark_invalid_bounds): Likewise.
23640 (chkp_maybe_copy_and_register_bounds): Likewise.
23641 (chkp_build_returned_bound): Likewise.
23642 (chkp_get_bound_for_parm): Likewise.
23643 (chkp_build_bndldx): Likewise.
23644 (chkp_get_bounds_by_definition): Likewise.
23645 (chkp_generate_extern_var_bounds): Likewise.
23646 (chkp_get_bounds_for_decl_addr): Likewise.
23647 * tree-chrec.c (chrec_apply): Likewise.
23648 * tree-data-ref.c (dump_data_reference): Likewise.
23649 (dump_subscript): Likewise.
23650 (dump_data_dependence_relation): Likewise.
23651 (analyze_overlapping_iterations): Likewise.
23652 * tree-inline.c (expand_call_inline): Likewise.
23653 (tree_function_versioning): Likewise.
23654 * tree-into-ssa.c (dump_defs_stack): Likewise.
23655 (dump_currdefs): Likewise.
23656 (dump_names_replaced_by): Likewise.
23657 (dump_update_ssa): Likewise.
23658 (update_ssa): Likewise.
23659 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23660 * tree-parloops.c (build_new_reduction): Likewise.
23661 (try_create_reduction_list): Likewise.
23662 (ref_conflicts_with_region): Likewise.
23663 (oacc_entry_exit_ok_1): Likewise.
23664 (oacc_entry_exit_single_gang): Likewise.
23665 * tree-pretty-print.h: Likewise.
23666 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23667 (get_scalar_evolution): Likewise.
23668 (add_to_evolution): Likewise.
23669 (get_loop_exit_condition): Likewise.
23670 (analyze_evolution_in_loop): Likewise.
23671 (analyze_initial_condition): Likewise.
23672 (analyze_scalar_evolution): Likewise.
23673 (instantiate_scev): Likewise.
23674 (number_of_latch_executions): Likewise.
23675 (gather_chrec_stats): Likewise.
23676 (final_value_replacement_loop): Likewise.
23677 (scev_const_prop): Likewise.
23678 * tree-sra.c (dump_access): Likewise.
23679 (disqualify_candidate): Likewise.
23680 (create_access): Likewise.
23681 (reject): Likewise.
23682 (maybe_add_sra_candidate): Likewise.
23683 (create_access_replacement): Likewise.
23684 (analyze_access_subtree): Likewise.
23685 (analyze_all_variable_accesses): Likewise.
23686 (sra_modify_assign): Likewise.
23687 (initialize_constant_pool_replacements): Likewise.
23688 (find_param_candidates): Likewise.
23689 (decide_one_param_reduction): Likewise.
23690 (replace_removed_params_ssa_names): Likewise.
23691 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23692 * tree-ssa-copy.c (dump_copy_of): Likewise.
23693 (copy_prop_visit_cond_stmt): Likewise.
23694 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23695 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23696 (record_equivalences_from_stmt): Likewise.
23697 * tree-ssa-dse.c (compute_trims): Likewise.
23698 (delete_dead_call): Likewise.
23699 (delete_dead_assignment): Likewise.
23700 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23701 (forward_propagate_into_cond): Likewise.
23702 (pass_forwprop::execute): Likewise.
23703 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23704 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23706 (move_computations_worker): Likewise.
23707 (execute_sm): Likewise.
23708 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23709 (remove_exits_and_undefined_stmts): Likewise.
23710 (remove_redundant_iv_tests): Likewise.
23711 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23712 (adjust_iv_update_pos): Likewise.
23713 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23714 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23715 (value_replacement): Likewise.
23716 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23717 * tree-ssa-pre.c (print_pre_expr): Likewise.
23718 (get_representative_for): Likewise.
23719 (create_expression_by_pieces): Likewise.
23720 (insert_into_preds_of_block): Likewise.
23721 (eliminate_insert): Likewise.
23722 (eliminate_dom_walker::before_dom_children): Likewise.
23723 (eliminate): Likewise.
23724 (remove_dead_inserted_code): Likewise.
23725 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23726 * tree-ssa-reassoc.c (get_rank): Likewise.
23727 (eliminate_duplicate_pair): Likewise.
23728 (eliminate_plus_minus_pair): Likewise.
23729 (eliminate_not_pairs): Likewise.
23730 (undistribute_ops_list): Likewise.
23731 (eliminate_redundant_comparison): Likewise.
23732 (update_range_test): Likewise.
23733 (optimize_range_tests_var_bound): Likewise.
23734 (optimize_vec_cond_expr): Likewise.
23735 (rewrite_expr_tree): Likewise.
23736 (rewrite_expr_tree_parallel): Likewise.
23737 (linearize_expr): Likewise.
23738 (break_up_subtract): Likewise.
23739 (linearize_expr_tree): Likewise.
23740 (attempt_builtin_powi): Likewise.
23741 (attempt_builtin_copysign): Likewise.
23742 (transform_stmt_to_copy): Likewise.
23743 (transform_stmt_to_multiply): Likewise.
23744 (dump_ops_vector): Likewise.
23745 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23746 (print_scc): Likewise.
23747 (set_ssa_val_to): Likewise.
23748 (visit_reference_op_store): Likewise.
23749 (visit_use): Likewise.
23750 (sccvn_dom_walker::before_dom_children): Likewise.
23751 (run_scc_vn): Likewise.
23752 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23754 (expr_hash_elt::print): Likewise.
23755 (const_and_copies::pop_to_marker): Likewise.
23756 (const_and_copies::record_const_or_copy_raw): Likewise.
23757 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23758 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23759 (dump_predicates): Likewise.
23760 (find_uninit_use): Likewise.
23761 (warn_uninitialized_phi): Likewise.
23762 (pass_late_warn_uninitialized::execute): Likewise.
23763 * tree-ssa.c (verify_vssa): Likewise.
23764 (verify_ssa): Likewise.
23765 (maybe_optimize_var): Likewise.
23766 * tree-vrp.c (dump_value_range): Likewise.
23767 (dump_all_value_ranges): Likewise.
23768 (dump_asserts_for): Likewise.
23769 (register_edge_assert_for_2): Likewise.
23770 (vrp_visit_cond_stmt): Likewise.
23771 (vrp_visit_switch_stmt): Likewise.
23772 (vrp_visit_stmt): Likewise.
23773 (vrp_visit_phi_node): Likewise.
23774 (simplify_cond_using_ranges_1): Likewise.
23775 (fold_predicate_in): Likewise.
23776 (evrp_dom_walker::before_dom_children): Likewise.
23777 (evrp_dom_walker::push_value_range): Likewise.
23778 (evrp_dom_walker::pop_value_range): Likewise.
23779 (execute_early_vrp): Likewise.
23781 2017-05-16 Richard Biener <rguenther@suse.de>
23783 * dwarf2out.c (loc_list_from_tree_1): Do not create
23784 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23786 2017-05-16 Richard Biener <rguenther@suse.de>
23788 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23790 (note_variable_value_in_expr): If we resolved the decl ref
23791 do not push to the stack.
23793 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23795 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23796 operations in fast-math mode.
23797 (vaddq_f16): Likewise.
23798 (vmul_f16): Likewise.
23799 (vmulq_f16): Likewise.
23800 (vsub_f16): Likewise.
23801 (vsubq_f16): Likewise.
23802 * config/arm/neon.md (add<mode>3): New.
23804 (fma:<VH:mode>3): New. Also remove outdated comment.
23807 2017-05-16 Martin Liska <mliska@suse.cz>
23811 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23812 (odr_types_equivalent_p): Likewise.
23814 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23816 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23818 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23821 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23822 non-interunit SSE move alternatives with '?'.
23823 (zero-extendsidi peephole2): New peephole to skip intermediate
23824 general register in SSE zero-extend sequence.
23826 2017-05-15 Jeff Law <law@redhat.com>
23828 * reorg.c (relax_delay_slots): Create a new variable to hold
23829 the temporary target rather than clobbering TARGET_LABEL.
23831 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23832 missing argument to extract_bit_field call.
23833 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23835 2017-05-15 Martin Liska <mliska@suse.cz>
23838 * gcc.c (process_command): Do not allow empty argument of -o option.
23840 2017-05-15 Renlin Li <renlin.li@arm.com>
23842 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23843 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23844 * config/aarch64/constraints.md (Usf): Add long call check.
23845 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23846 (call_value): Likewise.
23847 (sibcall): Likewise.
23848 (sibcall_value): Likewise.
23850 (call_value_insn): New.
23851 (sibcall_insn): Update rtx pattern.
23852 (sibcall_value_insn): Likewise.
23853 (call_internal): Remove.
23854 (call_value_internal): Likewise.
23855 (sibcall_internal): Likewise.
23856 (sibcall_value_internal): Likewise.
23857 (call_reg): Likewise.
23858 (call_symbol): Likewise.
23859 (call_value_reg): Likewise.
23860 (call_value_symbol): Likewise.
23862 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23865 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23867 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23869 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23870 compatible with CCGOCmode and with CCZmode.
23872 2017-05-14 Martin Sebor <msebor@redhat.com>
23874 PR middle-end/77671
23875 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23876 (gimple_fold_builtin_snprintf): Same.
23877 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23878 (gimple_fold_builtin_snprintf): Same.
23879 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23880 of character types.
23881 (is_call_safe): New function.
23882 (try_substitute_return_value): Call it.
23883 (try_simplify_call): New function.
23884 (pass_sprintf_length::handle_gimple_call): Call it.
23886 2017-05-14 Martin Sebor <msebor@redhat.com>
23888 PR middle-end/80669
23889 * builtins.c (expand_builtin_stpncpy): Simplify.
23891 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23893 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23894 * config/i386/i386.h
23895 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23896 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23897 (struct machine_function): Add new members call_ms2sysv,
23898 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23899 (struct machine_frame_state): New fields sp_realigned and
23900 sp_realigned_offset.
23901 * config/i386/i386.c
23902 (enum xlogue_stub): New enum.
23903 (enum xlogue_stub_sets): New enum.
23904 (class xlogue_layout): New class.
23905 (struct ix86_frame): New fields stack_realign_allocate_offset,
23906 stack_realign_offset and outlined_save_offset. Modify comments to
23907 detail stack layout when using out-of-line stubs.
23908 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23909 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23910 -mcall-ms2sysv-xlogues.
23911 (stub_managed_regs): New static variable.
23912 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23913 registers managed by out-of-line stub.
23914 (disable_call_ms2sysv_xlogues): New function.
23915 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23916 m->call_ms2sysv when appropriate and compute frame layout for
23918 (sp_valid_at, fp_valid_at): New inline functions.
23919 (choose_basereg): New function.
23920 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23922 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23923 Use align parameter of choose_baseaddr to generated aligned SSE movs
23925 (pro_epilogue_adjust_stack): Modify to track
23926 machine_frame_state::sp_realigned.
23927 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23928 (ix86_nsaved_sseregs): Likewise.
23929 (ix86_emit_save_regs): Likewise.
23930 (ix86_emit_save_regs_using_mov): Likewise.
23931 (ix86_emit_save_sse_regs_using_mov): Likewise.
23932 (get_scratch_register_on_entry): Likewise.
23933 (gen_frame_set): New function.
23934 (gen_frame_load): Likewise.
23935 (gen_frame_store): Likewise.
23936 (emit_outlined_ms2sysv_save): Likewise.
23937 (emit_outlined_ms2sysv_restore): Likewise.
23938 (ix86_expand_prologue): Modify stack re-alignment code and call
23939 emit_outlined_ms2sysv_save when appropriate.
23940 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23941 parameter rtx_insn *insn, which allows the function to be used to only
23942 generate the notes.
23943 (ix86_expand_epilogue): Modify validity checks of frame and stack
23944 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23945 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23946 * config/i386/predicates.md
23947 (save_multiple): New predicate.
23948 (restore_multiple): Likewise.
23949 * config/i386/sse.md
23950 (save_multiple<mode>): New pattern.
23951 (save_multiple_realign<mode>): Likewise.
23952 (restore_multiple<mode>): Likewise.
23953 (restore_multiple_and_return<mode>): Likewise.
23954 (restore_multiple_leave_return<mode>): Likewise.
23955 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23957 2017-05-14 Julia Koval <julia.koval@intel.com>
23959 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23960 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23961 (__builtin_ia32_xsetbv): New builtins.
23962 * config/i386/i386.c (ix86_expand_special_args_builtin):
23964 (ix86_expand_builtin): Special expand for new intrinsics.
23965 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23966 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23967 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23969 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23971 * cfganal.c (inverted_post_order_compute): Change argument type
23973 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23974 * df-core.c (rest_of_handle_df_initialize): Adjust.
23975 (rest_of_handle_df_finish): Likewise.
23976 (df_analyze_1): Likewise.
23977 (df_analyze): Likewise.
23978 (loop_inverted_post_order_compute): Change argument to be a vec *.
23979 (df_analyze_loop): Adjust.
23980 (df_get_n_blocks): Likewise.
23981 (df_get_postorder): Likewise.
23982 * df.h (struct df_d): Change field to be a vec.
23983 * lcm.c (compute_laterin): Adjust.
23984 (compute_available): Likewise.
23985 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23986 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23987 * tree-ssa-pre.c (compute_antic): Likewise.
23989 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23991 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23992 (depth_first_search::depth_first_search): Change structure init
23993 function to this constructor.
23994 (depth_first_search::add_bb): Rename function to this member.
23995 (depth_first_search::execute): Likewise.
23996 (flow_dfs_compute_reverse_finish): Adjust.
23998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24000 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24001 (longest_simple_path): Likewise.
24002 * shrink-wrap.c (spread_components): Likewise.
24003 (disqualify_problematic_components): Likewise.
24004 (emit_common_heads_for_components): Likewise.
24005 (emit_common_tails_for_components): Likewise.
24006 (insert_prologue_epilogue_for_components): Likewise.
24008 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24010 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24015 * df-core.c (df_set_blocks): Start using auto_bitmap.
24016 (df_compact_blocks): Likewise.
24017 * df-problems.c (df_rd_confluence_n): Likewise.
24018 * df-scan.c (df_insn_rescan_all): Likewise.
24019 (df_process_deferred_rescans): Likewise.
24020 (df_update_entry_block_defs): Likewise.
24021 (df_update_exit_block_uses): Likewise.
24022 (df_entry_block_bitmap_verify): Likewise.
24023 (df_exit_block_bitmap_verify): Likewise.
24024 (df_scan_verify): Likewise.
24025 * lra-constraints.c (lra_constraints): Likewise.
24026 (undo_optional_reloads): Likewise.
24027 (lra_undo_inheritance): Likewise.
24028 * lra-remat.c (calculate_gen_cands): Likewise.
24029 (do_remat): Likewise.
24030 * lra-spills.c (assign_spill_hard_regs): Likewise.
24031 (spill_pseudos): Likewise.
24032 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24033 (bitmap_set_subtract_values): Likewise.
24035 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24037 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24038 management with auto_bitmap.
24039 (fix_inter_tick): Likewise.
24040 (fix_recovery_deps): Likewise.
24041 * ira.c (add_store_equivs): Likewise.
24042 (find_moveable_pseudos): Likewise.
24043 (split_live_ranges_for_shrink_wrap): Likewise.
24044 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24045 (rtx_reuse_manager::seen_def_p): Likewise.
24046 (rtx_reuse_manager::set_seen_def): Likewise.
24047 * print-rtl.h (class rtx_reuse_manager): Likewise.
24049 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24051 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24053 (migrate_btr_def): Likewise.
24054 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24055 * df-core.c (loop_post_order_compute): Likewise.
24056 (loop_inverted_post_order_compute): Likewise.
24057 * hsa-common.h: Likewise.
24058 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24059 * init-regs.c (initialize_uninitialized_regs): Likewise.
24060 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24061 (inline_small_functions): Likewise.
24062 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24063 * ira.c (combine_and_move_insns): Likewise.
24064 (build_insn_chain): Likewise.
24065 * loop-invariant.c (find_invariants): Likewise.
24066 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24067 * predict.c (tree_predict_by_opcode): Likewise.
24068 (predict_paths_leading_to): Likewise.
24069 (predict_paths_leading_to_edge): Likewise.
24070 (estimate_loops_at_level): Likewise.
24071 (estimate_loops): Likewise.
24072 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24073 (spread_components): Likewise.
24074 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24075 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24076 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24077 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24078 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24079 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24080 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24081 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24082 (mark_threaded_blocks): Likewise.
24083 (thread_through_all_blocks): Likewise.
24084 * tree-ssa.c (verify_ssa): Likewise.
24085 (execute_update_addresses_taken): Likewise.
24086 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24088 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24090 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24092 (post_order_compute): Likewise.
24093 (inverted_post_order_compute): Likewise.
24094 (pre_and_rev_post_order_compute_fn): Likewise.
24096 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24098 * genrecog.c (int_set::int_set): Explicitly construct our
24099 auto_vec base class.
24100 * vec.h (auto_vec::auto_vec): New constructor.
24102 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24104 * bitmap.h (class auto_bitmap): New constructor taking
24105 bitmap_obstack * argument.
24107 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24109 * bitmap.h (class auto_bitmap): Change type of m_bits to
24110 bitmap_head, and adjust ctor / dtor and member operators.
24112 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24114 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24115 when returned register mode doesn't match original mode.
24117 2017-05-12 Jeff Law <law@redhat.com>
24118 Jakub Jelinek <jakub@redhat.com>
24120 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24121 we look for cc setter after the compare-elim changes.
24122 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24123 within the vector to match what compare-elim now expects.
24124 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24125 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24127 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24128 after the compare-elim changes.
24129 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24130 the vector to match what compare-elim now expects.
24131 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24132 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24133 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24134 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24135 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24137 * config/visium/visium.c (single_set_and_flags): Fix where
24138 we look for cc setter after the compare-elim changes.
24139 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24140 with the vector to match what compare-elim now expects.
24141 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24142 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24143 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24144 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24145 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24146 (neg<mode>2_insn_set_overflow): Likewise.
24148 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24150 PR middle-end/79794
24151 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24152 maybe_expand_insn call, set ops[0].target. If still set after call,
24153 set alt_rtl. Add extra arg to recursive calls.
24154 (extract_bit_field): Add alt_rtl argument. Pass to
24156 * expmed.h (extract_bit_field): Fix prototype.
24157 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24158 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24159 to extract_bit_field_calls.
24160 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24161 Pass alt_rtl to extract_bit_field calls.
24162 * calls.c (store_unaligned_arguments_into_psuedos)
24163 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24164 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24166 * optabs.h (struct expand_operand): Add target bitfield.
24168 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24170 * compare-elim.c (try_eliminate_compare): Canonicalize
24171 operation with embedded compare to
24172 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24173 (set (reg) (operation)].
24175 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24177 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24180 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24181 cost of adding a carry flag for ADC instruction.
24182 [case MINUS]: Ignore the cost of subtracting a carry flag
24183 for SBB instruction.
24185 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24187 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24189 * config/rs6000/bmiintrin.h: New file.
24190 * config/rs6000/bmi2intrin.h: New file.
24191 * config/rs6000/x86intrin.h: New file.
24193 2017-05-12 Jeff Law <law@redhat.com>
24195 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24198 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24200 PR middle-end/80707
24201 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24203 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24205 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24206 early expansion of vector divide builtins.
24207 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24208 builtins identified as having unsigned arguments.
24210 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24212 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24213 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24214 expansion of vector logical operations (and, andc, or, xor,
24217 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24219 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24220 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24222 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24224 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24225 early GIMPLE expansion of vector multiplies.
24227 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24229 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24230 TARGET_HAVE_MOVT conditional.
24231 (movt splitter): Likewise.
24233 2017-05-12 Richard Biener <rguenther@suse.de>
24235 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24236 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24237 Fold all stmts not inplace.
24239 2017-05-12 Richard Biener <rguenther@suse.de>
24241 PR tree-optimization/80713
24242 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24243 inserted_exprs bit for not removed stmts.
24245 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24247 PR middle-end/69921
24248 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24249 parallelized" attribute for parallelized OpenACC kernels.
24250 * omp-offload.c (execute_oacc_device_lower): Use it.
24252 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24253 Set "oacc kernels" attribute.
24254 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24255 parameter. Adjust all users.
24256 (oacc_fn_attrib_kernels_p): Remove function.
24257 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24258 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24259 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24260 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24261 assert "oacc kernels" attribute is set.
24263 2017-05-11 Carl Love <cel@us.ibm.com>
24265 * config/rs6000/rs6000-c: Add support for built-in functions
24266 vector unsigned char vec_popcnt (vector signed char)
24267 vector unsigned char vec_popcnt (vector unsigned char)
24268 vector unsigned short vec_popcnt (vector signed short)
24269 vector unsigned short vec_popcnt (vector unsigned short)
24270 vector unsigned int vec_popcnt (vector signed int)
24271 vector unsigned int vec_popcnt (vector unsigned int)
24272 vector unsigned long long vec_popcnt (vector signed long long)
24273 vector unsigned long long vec_popcnt (vector unsigned long long)
24274 vector signed long long vec_slo (vector signed long long,
24275 vector signed char)
24276 vector signed long long vec_slo (vector signed long long,
24277 vector unsigned char)
24278 vector unsigned long long vec_slo (vector unsigned long long,
24279 vector signed char)
24280 vector unsigned long long vec_slo (vector unsigned long long,
24281 vector unsigned char)
24282 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24283 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24284 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24285 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24286 * doc/extend.texi: Update the built-in documentation file for the
24287 new built-in functions.
24289 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24291 * attribs.h (sorted_attr_string): Move machine independent
24292 functions for target clone support from the i386 port to common
24293 code. Rename ix86_function_versions to common_function_versions.
24294 Rename make_name to make_unique_name.
24295 (common_function_versions): Likewise.
24296 (make_unique_name): Likewise.
24297 (make_dispatcher_decl): Likewise.
24298 (is_function_default_version): Likewise.
24299 * attribs.c (attr_strcmp): Likewise.
24300 (sorted_attr_string): Likewise.
24301 (common_function_versions): Likewise.
24302 (make_unique_name): Likewise.
24303 (make_dispatcher_decl): Likewise.
24304 (is_function_default_version): Likewise.
24305 * config/i386/i386.c (attr_strcmp): Likewise.
24306 (sorted_attr_string): Likewise.
24307 (ix86_function_versions): Likewise.
24308 (make_name): Likewise.
24309 (make_dispatcher_decl): Likewise.
24310 (is_function_default_version): Likewise.
24311 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24313 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24316 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24317 Account for direct move costs for vec_construct of integer
24320 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24323 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24324 (UNSPEC_STX_ATOMIC): Ditto.
24325 (loaddi_via_sse): New insn.
24326 (storedi_via_sse): Ditto.
24327 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24328 Update corresponding peephole2 patterns.
24329 (atomic_storedi_fpu): Ditto.
24331 2017-05-11 Julia Koval <julia.koval@intel.com>
24333 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24334 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24336 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24337 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24338 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24340 2017-05-11 Nathan Sidwell <nathan@acm.org>
24342 * graphite-poly.c: Include dumpfile.h.
24344 * dumpfle.h (dump_function): Declare here ...
24345 * tree-dump.h (dump_function): ... not here.
24346 * dumpfile.c: #include tree-cfg.h.
24347 (dump_function): Move here from ...
24348 * tree-dump.c (dump_function): ... here.
24349 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24350 * graphite-poly.c: Don't include tree-dump.h.
24351 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24352 * print-tree.c: Likewise.
24353 * stor-layout.c: Likewise.
24354 * tree-nested.c: Likewise.
24356 * dumpfile.c (dump_start): Use TDF_FLAGS.
24357 (dump_enable_all): Fix TDF_KIND check thinko.
24359 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24361 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24362 array entries to represent two legal parameterizations of the
24363 overloaded __builtin_cmpb function, as represented by the
24364 P6_OV_BUILTIN_CMPB constant.
24365 (altivec_resolve_overloaded_builtin): Add special case handling
24366 for the __builtin_cmpb function, as represented by the
24367 P6_OV_BUILTIN_CMPB constant.
24368 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24369 (BU_P6_64BIT_2): New macro.
24370 (BU_P6_OVERLOAD_2): New macro
24371 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24372 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24373 (CMPB): Add overload support to represent both 32-bit and 64-bit
24374 compare-bytes function.
24375 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24376 support for TARGET_CMPB.
24377 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24378 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24379 documentation of the __builtin_cmpb overloaded built-in function.
24381 2017-05-11 Richard Biener <rguenther@suse.de>
24383 PR tree-optimization/80705
24384 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24385 bases are not vectorizable.
24387 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24389 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24390 when counting register pressure.
24392 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24394 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24395 (iv_ca_more_deps): Renamed to ...
24396 (iv_ca_compare_deps): ... this.
24397 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24399 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24401 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24403 (determine_group_iv_costs): ... here.
24404 (find_inv_vars_cb): Record inv var if it's not recorded before.
24406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24408 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24409 (get_shiftadd_cost): Ditto.
24411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24413 * tree-ssa-address.c: Include header file.
24414 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24416 (add_to_parts): Refactor.
24417 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24418 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24421 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24423 PR tree-optimization/53090
24424 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24426 (extract_cond_operands): Detect condition with IV on both sides
24427 and return COMP_IV_EXPR_2.
24428 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24429 (rewrite_use_compare): Simplify by removing call to function
24430 extract_cond_operands.
24432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24434 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24435 (extract_cond_operands): Detect condition comparing against non-
24436 invariant bound and return appropriate enum value.
24437 (find_interesting_uses_cond): Update use of extract_cond_operands.
24438 Handle its return value accordingly.
24439 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24443 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24444 nonlinear iv_use computation in loop invariant sensitive way.
24446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24448 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24449 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24453 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24454 (dump_cand): Support iv_cand.inv_exprs.
24455 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24457 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24462 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24464 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24465 as local function. Include necessary header files.
24466 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24468 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24470 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24474 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24475 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24476 RSHIFT_EXPR and BIT_NOT_EXPR.
24478 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24480 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24481 (adjust_setup_cost): New parameter supporting round up adjustment.
24482 (struct address_cost_data): Delete.
24483 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24484 (split_address_cost, ptr_difference_cost): Delete.
24485 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24486 (struct ainc_cost_data): New struct.
24487 (get_address_cost_ainc): New function.
24488 (get_address_cost, get_computation_cost): Reimplement.
24489 (determine_group_iv_cost_address): Record inv_expr for all uses of
24491 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24492 (iv_ca_has_deps): Reimplemented to ...
24493 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24495 (iv_ca_extend): Call iv_ca_more_deps.
24497 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24499 * tree-ssa-address.c (struct mem_address): Move to header file.
24500 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24501 * tree-ssa-address.h (struct mem_address): Move from C file.
24502 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24504 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24506 * tree-affine.h (aff_combination_type): New interface.
24507 (aff_combination_zero_p): Remove static.
24508 (aff_combination_const_p): New interface.
24509 (aff_combination_singleton_var_p): New interfaces.
24511 2017-05-11 Richard Biener <rguenther@suse.de>
24513 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24514 Skip unreachable blocks and destinations.
24515 (eliminate): Move stmt removal and fixup ...
24516 (fini_eliminate): ... here. Skip inserted exprs.
24517 (pass_pre::execute): Move fini_pre after fini_eliminate.
24518 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24519 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24520 PRE to get rid of dead code that has invalid SSA form and
24521 split critical edges again.
24523 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24525 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24527 2017-05-11 Richard Biener <rguenther@suse.de>
24529 * passes.c (execute_function_todo): Verify loops if they are
24530 said to be up-to-date.
24531 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24532 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24534 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24537 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24538 handle calling assemble_external ourself.
24541 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24542 modes with zero size. Enhance comment.
24544 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24546 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24547 built-ins for vec_xl and vec_xst with short and char pointer
24550 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24552 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24553 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24554 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24555 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24556 (_mm_maskz_min_round_ss): New intrinsics.
24557 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24558 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24559 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24560 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24561 (__builtin_ia32_minss_mask_round): New builtins.
24562 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24563 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24564 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24566 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24567 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24569 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24572 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24574 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24575 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24576 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24577 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24578 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24579 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24580 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24581 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24582 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24583 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24584 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24585 (__builtin_ia32_mulss_mask_round): New builtins.
24586 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24587 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24588 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24590 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24591 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24593 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24596 2017-05-10 Julia Koval <julia.koval@intel.com>
24598 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24599 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24600 (_mm256_setr_m128i): New intrinsics.
24602 2017-05-10 Julia Koval <julia.koval@intel.com>
24604 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24605 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24606 (_mm_maskz_rcp14_ss): New intrinsics.
24607 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24608 (__builtin_ia32_rcp14ss_mask): New builtins.
24609 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24611 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24613 PR tree-optimization/51513
24614 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24615 (assert_unreachable_fallthru_edge_p): Use it.
24616 (group_case_labels_stmt): Likewise.
24617 * tree-cfg.h: Prototype it.
24618 * stmt.c: Include cfghooks.h and tree-cfg.h.
24619 (emit_case_dispatch_table) <gap_label>: New local variable.
24620 Use it to fill dispatch table gaps.
24621 Test for default_label before updating probabilities.
24622 (expand_case) <default_label>: Remove unneeded initialization.
24623 Test for unreachable default case statement and remove its edge.
24624 Set default_label accordingly.
24625 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24627 2017-05-10 Carl Love <cel@us.ibm.com>
24629 * config/rs6000/rs6000-c: Add support for built-in functions
24630 vector signed char vec_neg (vector signed char)
24631 vector signed short int vec_neg (vector short int)
24632 vector signed int vec_neg (vector signed int)
24633 vector signed long long vec_neg (vector signed long long)
24634 vector float vec_neg (vector float)
24635 vector double vec_neg (vector double)
24636 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24638 * config/rs6000/altivec.h: Add define for vec_neg
24639 * doc/extend.texi: Update the built-in documentation for the
24640 new built-in functions.
24642 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24644 PR tree-optimization/77644
24645 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24647 2017-05-10 Nathan Sidwell <nathan@acm.org>
24649 * dumpfile.h (TDI_lang_all): New.
24650 (TDF_KIND): New. Renumber others
24651 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24653 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24655 (get_dump_file_name): Adjust suffix generation.
24656 (dump_enable_all): Use TDF_KIND.
24657 * doc/invoke.texi (-fdump-lang-all): Document.
24659 * dumpfile.h: Tabify.
24661 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24664 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24665 Move member access before delete.
24667 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24669 * tree-inline.c (expand_call_inline): Split block at stmt
24672 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24675 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24676 are now unused after splitting mov{sf,sd}_hardfloat.
24677 (f32_lr2): Likewise.
24678 (f32_lm): Likewise.
24679 (f32_lm2): Likewise.
24680 (f32_li): Likewise.
24681 (f32_li2): Likewise.
24682 (f32_lv): Likewise.
24683 (f32_sr): Likewise.
24684 (f32_sr2): Likewise.
24685 (f32_sm): Likewise.
24686 (f32_sm2): Likewise.
24687 (f32_si): Likewise.
24688 (f32_si2): Likewise.
24689 (f32_sv): Likewise.
24690 (f32_dm): Likewise.
24691 (f32_vsx): Likewise.
24692 (f32_av): Likewise.
24693 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24694 For movsf, order stores so the VSX stores occur before the GPR
24695 store which encourages the register allocator to use a traditional
24696 FPR instead of a GPR. For movsd, order the stores so that the GPR
24697 store comes before the VSX stores to allow the power6 to work.
24698 This is due to the power6 not having a 32-bit integer store
24699 instruction from a FPR.
24700 (movsf_hardfloat): Likewise.
24701 (movsd_hardfloat): Likewise.
24703 2017-05-09 Martin Sebor <msebor@redhat.com>
24705 PR translation/80280
24706 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24709 PR translation/80280
24710 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24711 data member added in r247778.
24712 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24714 2017-05-09 Nathan Sidwell <nathan@acm.org>
24716 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24718 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24721 2017-05-09 Marek Polacek <polacek@redhat.com>
24723 * doc/invoke.texi: Fix typo.
24725 2017-05-09 Richard Biener <rguenther@suse.de>
24727 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24728 (vrp_val_is_min): Likewise.
24729 (set_value_range_to_value): Likewise.
24730 (set_value_range_to_nonnegative): Likewise.
24731 (gimple_assign_nonzero_p): Likewise.
24732 (gimple_stmt_nonzero_p): Likewise.
24733 (vrp_int_const_binop): Likewise. Remove unreachable case.
24734 (adjust_range_with_scev): Adjust comments.
24735 (compare_range_with_value): Likewise.
24736 (extract_range_from_phi_node): Likewise.
24737 (test_for_singularity): Likewise.
24739 2017-05-09 Richard Biener <rguenther@suse.de>
24741 * tree-vrp.c (get_single_symbol): Add assert that we don't
24742 get overflowed constants as invariant part.
24743 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24744 checks. Use wi::cmp instead of recursing for integer constants.
24745 (compare_values): Just ignore whether we assumed undefined
24746 overflow instead of failing the compare.
24747 (extract_range_for_var_from_comparison_expr): Add comment before the
24748 TREE_NO_WARNING sets.
24749 (test_for_singularity): Likewise.
24750 (extract_range_from_comparison): Do not disable optimization
24751 when we assumed undefined overflow.
24752 (extract_range_basic): Remove init of unused var.
24754 2017-05-09 Richard Biener <rguenther@suse.de>
24756 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24757 (extract_range_from_multiplicative_op_1): Adjust.
24758 (extract_range_from_binary_expr_1): Use int_const_binop.
24760 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24763 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24764 rs6000_store_data_bypass_p in seven define_bypass directives and
24765 in several comments.
24766 * config/rs6000/rs6000-protos.h: Add prototype for
24767 rs6000_store_data_bypass_p function.
24768 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24769 function implements slightly different (rs6000-specific) semantics
24770 than store_data_bypass_p, returning false rather than aborting
24771 with assertion error when arguments do not satisfy the
24772 requirements of store data bypass.
24773 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24774 rs6000_store_data_bypass_p.
24776 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24778 * config/xtensa/xtensa-protos.h
24779 (xtensa_initial_elimination_offset): New declaration.
24780 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24781 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24782 macro definition, add case for FRAME_POINTER_REGNUM when
24783 FRAME_GROWS_DOWNWARD.
24784 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24785 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24786 xtensa_initial_elimination_offset.
24788 2017-05-08 Nathan Sidwell <nathan@acm.org>
24790 * doc/invoke.texi: Alphabetize -fdump options.
24792 2017-05-08 Martin Sebor <msebor@redhat.com>
24794 PR translation/80280
24795 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24797 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24799 * target.def (compute_frame_layout): New optional target hook.
24800 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24801 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24802 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24804 * reload1.c (verify_initial_elim_offsets): Likewise.
24805 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24806 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24807 (arm_get_frame_offsets): Split up into this ...
24808 (arm_compute_frame_layout): ... and this function.
24810 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24812 * config/aarch64/constraints.md (Usa): New constraint.
24813 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24815 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24817 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24818 with_multilib_list after it has been checked.
24820 2017-05-08 Richard Biener <rguenther@suse.de>
24822 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24823 (bitmap_set_subtract_values): Likewise.
24825 2017-05-08 Richard Biener <rguenther@suse.de>
24827 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24828 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24830 (gimple_stmt_nonzero_warnv_p): Rename to ...
24831 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24833 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24834 (extract_range_basic): Adjust, do not disable propagation on
24835 strict overflow sensitive simplification.
24836 (vrp_visit_cond_stmt): Likewise.
24838 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24840 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24841 body size unconditionally.
24843 2017-05-07 Jeff Law <law@redhat.com>
24846 2017-05-06 Jeff Law <law@redhat.com>
24847 PR tree-optimization/78496
24848 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24851 PR tree-optimization/78496
24852 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24853 (simplify_stmt_using_ranges): Call it.
24854 (vrp_dom_walker::before_dom_children): Extract equivalences
24855 from an ASSERT_EXPR with an equality comparison against a
24858 2017-05-06 Jeff Law <law@redhat.com>
24860 PR tree-optimization/78496
24861 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24864 PR tree-optimization/78496
24865 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24866 (simplify_stmt_using_ranges): Call it.
24867 (vrp_dom_walker::before_dom_children): Extract equivalences
24868 from an ASSERT_EXPR with an equality comparison against a
24871 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24873 * lra-constraints.c (lra_copy_reg_equiv): New function.
24874 (split_reg): Use it to copy equivalence information from the
24875 original register to the spill register.
24877 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24879 PR rtl-optimization/75964
24880 * simplify-rtx.c (simplify_const_relational_operation): Remove
24881 invalid handling of comparisons of integer ABS.
24883 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24885 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24886 initialize to zero.
24887 (init_regs): Remove declaration.
24888 (function_arg_advance_32): Initialize error_p as boolean variable.
24890 2017-05-05 Nathan Sidwell <nathan@acm.org>
24892 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24893 lines. Use for (;;).
24895 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24897 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24898 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24899 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24900 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24901 VF=2 that require versioning.
24903 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24905 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24908 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24910 * diagnostic.h (diagnostic_override_option_index): Convert from
24911 macro to inline function.
24913 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24915 * diagnostic.c (last_module_changed_p): New function.
24916 (set_last_module): New function.
24917 (diagnostic_report_current_module): Convert macro usage to
24918 the above functions.
24919 * diagnostic.h (diagnostic_context::last_module): Strengthen
24920 from const line_map * to const line_map_ordinary *.
24921 (diagnostic_last_module_changed): Delete macro.
24922 (diagnostic_set_last_module): Delete macro.
24924 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24926 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24927 with diagnostic_report_diagnostic.
24928 (diagnostic_n_impl_richloc): Likewise.
24929 * diagnostic.h (report_diagnostic): Delete macro.
24930 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24931 with diagnostic_report_diagnostic.
24932 * substring-locations.c (format_warning_va): Likewise.
24934 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24936 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24937 save/restor of format_spec. Move option-printing code to...
24938 (print_option_information): ...this new function, and
24939 reimplement by simply printing to the pretty_printer,
24940 rather than appending to the format string.
24942 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24944 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24945 handling logic into...
24946 (update_effective_level_from_pragmas): ...this new function.
24948 2017-05-04 Andrew Waterman <andrew@sifive.com>
24950 * config/riscv/riscv.opt (mstrict-align): New option.
24951 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24952 (SLOW_UNALIGNED_ACCESS): Define.
24953 (riscv_slow_unaligned_access): Declare.
24954 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24956 (riscv_slow_unaligned_access): New variable.
24957 (rocket_tune_info): Set slow_unaligned_access to true.
24958 (optimize_size_tune_info): Set slow_unaligned_access to false.
24959 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24960 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24961 (riscv_option_override): Set riscv_slow_unaligned_access.
24962 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24964 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24966 * config/riscv/riscv.md: Unify indentation.
24968 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24973 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24975 (extendsi<mode>2): Add support for doing sign extension via
24976 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24977 don't have ISA 3.0 instructions.
24978 (extendsi<mode>2 splitter): Likewise.
24979 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24980 generate the normal insns since SImode can now go in vector
24981 registers. Disallow the special UNSPECs needed for previous
24982 machines to hide SImode being used. Add new insns
24983 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24984 (fix_trunc<mode>si2_stfiwx): Likewise.
24985 (fix_trunc<mode>si2_internal): Likewise.
24986 (fixuns_trunc<mode>si2): Likewise.
24987 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24988 (fctiw<u>z_<mode>_smallint): Likewise.
24989 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24990 of floating point to 32-bit integer from doing a direct move to
24991 the GPR registers to do a store.
24992 (fctiwz_<mode>): Break long line.
24994 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24996 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24997 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24998 (addr_list, addr_offset_valid_p): New.
24999 (split_address_groups): Check offset validity with above function.
25000 (gt-tree-ssa-loop-ivopts.h): Include header file.
25002 2017-05-05 Nathan Sidwell <nathan@acm.org>
25004 * config.gcc (arm*-*-*): Add missing 'fi'.
25006 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25008 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25009 included in -fopt-info does not matter.
25010 * doc/optinfo.texi (-fopt-info): Fix description of default
25011 behavour. Explicitly say order of options included in -fopt-info
25014 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25016 * config.gcc: Allow combinations of aprofile and rmprofile values for
25017 --with-multilib-list.
25018 * config/arm/t-multilib: New file.
25019 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25020 variables. Remove setting of ISA and floating-point ABI in
25021 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25022 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25023 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25024 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25026 * config/arm/t-rmprofile: Likewise except for the matches changes.
25027 * doc/install.texi (--with-multilib-list): Document the combination of
25028 aprofile and rmprofile values and warn about pitfalls in doing that.
25030 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25032 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25033 (movdi_aarch64): Likewise.
25035 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25037 PR tree-optimization/80632
25038 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25040 (build_arrays): Initialize it for virtual phis.
25041 (fix_phi_nodes): Use it for virtual phis.
25043 PR tree-optimization/80558
25044 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25045 [x, y] op z into [x op, y op z] for op & or | if conditions
25048 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25049 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25052 * config/arm/arm.md (use_literal_pool): Remove.
25053 (64-bit immediate split): No longer takes cost into consideration
25054 if arm_disable_literal_pool is enabled.
25055 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25056 used when arm_disable_literal_pool is enabled.
25057 (arm_max_const_double_inline_cost): Remove use of
25058 arm_disable_literal_pool.
25059 (push_minipool_fix): Add assert.
25060 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25061 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25062 (no_literal_pool_sf_immediate): New.
25064 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25066 PR tree-optimization/80613
25067 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25068 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25070 2017-05-05 Richard Biener <rguenther@suse.de>
25072 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25074 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25076 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25077 of this flag from insn conditions due to removal from r247495.
25079 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25081 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25083 (arm_early_store_addr_dep_ptr): Likewise.
25084 * config/arm/aarch-common-protos.h
25085 (arm_early_load_addr_dep_ptr): Add prototype.
25086 (arm_early_store_addr_dep_ptr): Likewise.
25087 * config/arm/cortex-a53.md: Add new bypasses.
25089 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25091 * tree.c (next_type_uid): Change type to unsigned.
25092 (type_hash_canon): Decrement back next_type_uid if
25093 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25094 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25097 2017-05-04 Martin Sebor <msebor@redhat.com>
25099 * builtins.c: Fix a trivial typo in a comment.
25101 PR middle-end/79234
25102 * builtins.c (check_sizes): Adjust to handle reading past the end.
25103 Avoid printing excessive upper bound of ranges. Use %E to print
25104 tree nodes instead of converting them to %wu.
25105 (expand_builtin_memchr): New function.
25106 (compute_dest_size): Rename...
25107 (compute_objsize): ...to this.
25108 (expand_builtin_memcpy): Adjust.
25109 (expand_builtin_mempcpy): Adjust.
25110 (expand_builtin_strcat): Adjust.
25111 (expand_builtin_strcpy): Adjust.
25112 (check_strncat_sizes): Adjust.
25113 (expand_builtin_strncat): Adjust.
25114 (expand_builtin_strncpy): Adjust and simplify.
25115 (expand_builtin_memset): Adjust.
25116 (expand_builtin_bzero): Adjust.
25117 (expand_builtin_memcmp): Adjust.
25118 (expand_builtin): Handle memcmp.
25119 (maybe_emit_chk_warning): Check strncat just once.
25121 2017-05-04 Martin Sebor <msebor@redhat.com>
25123 PR preprocessor/79214
25124 PR middle-end/79222
25125 PR middle-end/79223
25126 * builtins.c (check_sizes): Add inlining context and issue
25127 warnings even when -Wno-system-headers is set.
25128 (check_strncat_sizes): Same.
25129 (expand_builtin_strncat): Same.
25130 (expand_builtin_memmove): New function.
25131 (expand_builtin_stpncpy): Same.
25132 (expand_builtin): Handle memmove and stpncpy.
25134 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25136 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25137 which is not used any more.
25139 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25141 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25143 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25145 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25146 (cortexa53_tunings): Likewise.
25147 (cortexa57_tunings): Likewise.
25148 (cortexa72_tunings): Likewise.
25149 (cortexa73_tunings): Likewise.
25151 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25153 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25154 Set loop alignment to 8.
25156 2017-05-04 Martin Sebor <msebor@redhat.com>
25158 PR translation/80280
25159 * builtins.c (expand_builtin_object_size): Add missing quoting to
25160 %D and like directives.
25161 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25162 (hsa_type_for_tree_type): Same.
25163 (verify_function_arguments): Same.
25164 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25165 * varasm.c (get_section): Same.
25168 2017-05-04 Martin Sebor <msebor@redhat.com>
25170 PR translation/80280
25171 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25173 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25175 * config/aarch64/aarch64.c (generic_addrcost_table):
25176 Change HI/TI mode setting.
25178 2017-05-04 Martin Jambor <mjambor@suse.cz>
25180 PR tree-optimization/80622
25181 * tree-sra.c (comes_initialized_p): New function.
25182 (build_accesses_from_assign): Only set write lazily when
25183 comes_initialized_p is false.
25184 (analyze_access_subtree): Use comes_initialized_p.
25185 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25186 instead of testing for PARM_DECL.
25188 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25190 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25191 constraint on operand 0 to allow more general addressing modes.
25192 Adjust output template.
25193 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25195 * config/aarch64/aarch64-protos.h
25196 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25197 * config/aarch64/constraints.md (Dp): New address constraint.
25198 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25201 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25203 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25204 update use of estimate_ipcp_clone_size_and_time.
25205 (estimate_local_effects): Update use of
25206 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25207 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25208 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25209 Return nonspecialized time.
25211 2017-05-04 Richard Biener <rguenther@suse.de>
25213 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25214 for the last VUSE which def dominates the PHI. Directly call
25216 (get_continuation_for_phi_1): Remove.
25218 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25220 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25221 to explain the use of truncating division. Cap the number of
25222 iterations to the maximum given by nb_iterations_upper_bound,
25225 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25227 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25228 * configure: Regenerate.
25229 * config.in: Regenerate.
25230 * config/i386/driver-mingw32.c: new file.
25231 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25232 * config.host: Link driver-mingw32.o on MinGW host.
25233 * doc/install.texi: Document new --enable-mingw-wildcard configure
25236 2017-05-04 Marek Polacek <polacek@redhat.com>
25238 PR tree-optimization/80612
25239 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25241 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25242 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25244 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25245 (movt splitter): Likewise.
25246 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25247 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25248 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25249 block for Thumb-1 with MOVT.
25250 (thumb2_legitimate_address_p): Move code block ...
25251 (can_avoid_literal_pool_for_label_p): ... into this new function.
25252 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25254 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25255 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25256 "M-profile targets with the MOVT instruction".
25258 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25260 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25261 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25262 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25264 2017-05-04 Martin Liska <mliska@suse.cz>
25266 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25267 variable cond_code.
25269 2017-05-04 Richard Biener <rguenther@suse.de>
25271 * tree.c (array_at_struct_end_p): Handle arrays at struct
25272 end with flexarrays more conservatively. Refactor and treat
25273 arrays of arrays or aggregates more strict. Fix
25274 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25275 * tree.c (array_at_struct_end_p): Adjust prototype.
25276 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25277 * gimple-fold.c (get_range_strlen): Likewise.
25278 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25280 2017-05-04 Richard Biener <rguenther@suse.de>
25282 PR tree-optimization/31130
25283 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25285 (supports_overflow_infinity): Likewise.
25286 (is_negative_overflow_infinity): Likewise.
25287 (is_positive_overflow_infinity): Likewise.
25288 (is_overflow_infinity): Likewise.
25289 (stmt_overflow_infinity): Likewise.
25290 (overflow_infinity_range_p): Likewise.
25291 (usable_range_p): Remove as always returning true.
25292 (make_overflow_infinity): Remove.
25293 (negative_overflow_infinity): Likewise.
25294 (positive_overflow_infinity): Likewise.
25295 (avoid_overflow_infinity): Likewise.
25296 (set_value_range): Adjust accordingly.
25297 (set_value_range_to_nonnegative): Likewise, remove now unused
25298 overflow_infinity arg.
25299 (vrp_operand_equal_p): Adjust.
25300 (update_value_range): Likewise.
25301 (range_int_cst_singleton_p): Likewise.
25302 (operand_less_p): Likewise.
25303 (compare_values_warnv): Likewise.
25304 (extract_range_for_var_from_comparison_expr): Likewise.
25305 (vrp_int_const_binop): Likewise.
25306 (zero_nonzero_bits_from_vr): Likewise.
25307 (extract_range_from_multiplicative_op_1): Likewise.
25308 (extract_range_from_binary_expr_1): Likewise.
25309 (extract_range_from_unary_expr): Likewise.
25310 (extract_range_from_comparison): Likewise.
25311 (extract_range_basic): Likewise.
25312 (adjust_range_with_scev): Likewise.
25313 (compare_ranges): Likewise.
25314 (compare_range_with_value): Likewise.
25315 (dump_value_range): Likewise.
25316 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25318 (simplify_cond_using_ranges): Adjust.
25320 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25322 * brig-builtins.def: Added a builtin for class_f64.
25323 * builtin-types.def: Added a builtin type needed by class_f64.
25325 2017-05-03 Jason Merrill <jason@redhat.com>
25327 * timevar.def: Add TV_CONSTEXPR.
25329 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25331 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25333 2017-05-03 Martin Jambor <mjambor@suse.cz>
25335 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25336 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25337 * ipa-cp.c (ipcp_propagate_stage): Do not call
25338 ipa_update_after_lto_read.
25339 * ipa-inline.c (ipa_inline): Likewise.
25341 2017-05-03 Martin Jambor <mjambor@suse.cz>
25343 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25344 tag. Added a default constructor and a destructor.
25345 (ipa_edge_args_sum_t): New class;
25346 (ipa_edge_args_sum): Declare.
25347 (ipa_edge_args_vector): Remove declaration.
25348 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25349 (ipa_free_edge_args_substructures): Remove declaration.
25350 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25351 (ipa_edge_args_info_available_for_edge_p): Likewise.
25352 * ipa-prop.c (ipa_edge_args_vector): Removed.
25353 (edge_removal_hook_holder): Likewise.
25354 (edge_duplication_hook_holder): Likewise.
25355 (ipa_edge_args_sum): New variable.
25356 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25357 ipa_edge_args_vector.
25358 (ipa_free_edge_args_substructures): Likewise.
25359 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25360 ipa_edge_args_vector.
25361 (ipa_edge_removal_hook): Turned into method
25362 ipa_edge_args_sum_t::remove.
25363 (ipa_edge_duplication_hook): Turned into method
25364 ipa_edge_args_sum_t::duplicate.
25365 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25366 registering edge hooks.
25367 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25368 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25369 ipa_edge_args_sum instead of ipa_edge_args_vector.
25370 * ipa-profile.c (ipa_profile): Likewise.
25372 2017-05-03 Martin Jambor <mjambor@suse.cz>
25374 * symbol-summary.h (function_summary): New method exists.
25375 (function_summary::symtab_removal): Deallocate through release.
25376 (call_summary): New class.
25377 (gt_ggc_mx): New overload.
25378 (gt_pch_nx): Likewise.
25379 (gt_pch_nx): Likewise.
25381 2017-05-03 Jeff Law <law@redhat.com>
25383 PR tree-optimization/78496
25384 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25385 from simplify_cond_using_ranges. Split off code to walk
25386 backwards through casts into ...
25387 (simplify_cond_using_ranges_2): New function.
25388 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25389 (execute_vrp): After identifying jump threads, call
25390 simplify_cond_using_ranges_2.
25392 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25395 * ipa-inline.h (inline_summary): Add ctor.
25396 (create_ggc): Do not use ggc_cleared_alloc.
25398 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25399 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25401 * gcc.c (handle_braces): Support escaping in switch matching
25403 * doc/invoke.texi (Spec Files): Document it.
25404 Remove superfluous @code markup in items.
25406 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25408 * diagnostic-show-locus.c (struct column_range): New struct.
25409 (get_affected_columns): New function.
25410 (get_printed_columns): New function.
25411 (struct correction): New struct.
25412 (correction::ensure_capacity): New function.
25413 (correction::ensure_terminated): New function.
25414 (struct line_corrections): New struct.
25415 (line_corrections::~line_corrections): New dtor.
25416 (line_corrections::add_hint): New function.
25417 (layout::print_trailing_fixits): Reimplement in terms of the new
25419 (selftest::test_overlapped_fixit_printing): New function.
25420 (selftest::diagnostic_show_locus_c_tests): Call it.
25422 2017-05-03 Nathan Sidwell <nathan@acm.org>
25424 Canonicalize canonical type hashing
25425 * tree.h (type_hash_canon_hash): Declare.
25426 * tree.c (type_hash_list, attribute_hash_list): Move into
25427 type_hash_canon_hash.
25428 (build_type_attribute_qual_variant): Break out hash code calc into
25429 type_hash_canon_hash.
25430 (type_hash_canon_hash): New. Generic type hash computation.
25431 (build_range_type_1, build_array_type_1, build_function_type,
25432 build_method_type_directly, build_offset_type, build_complex_type,
25433 make_vector_type): Call it.
25435 2017-05-03 Richard Biener <rguenther@suse.de>
25437 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25438 When all DRs have unknown misaligned do not always peel
25439 when there is a store but apply the same costing model as if
25440 there were only loads.
25442 2017-05-03 Richard Biener <rguenther@suse.de>
25445 PR tree-optimization/80492
25446 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25447 compare_base_decls returning dont-know properly.
25449 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25451 * config/arm/iterators.md (CCSI): New mode iterator.
25452 (arch): New mode attribute.
25453 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25454 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25455 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25456 code iterator for success result mode.
25457 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25458 the corresponding new insn generators.
25460 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25463 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25464 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25466 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25468 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25469 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25470 (DDR_A): Wrap DDR argument in brackets.
25471 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25472 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25473 (DDR_REVERSED_P): Likewise.
25475 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25477 PR tree-optimization/79472
25478 * tree-switch-conversion.c (struct switch_conv_info): Add
25479 contiguous_range and default_case_nonstandard fields.
25480 (collect_switch_conv_info): Compute contiguous_range and
25481 default_case_nonstandard fields, don't clear final_bb if
25482 contiguous_range and only the default case doesn't have the required
25484 (check_all_empty_except_final): Set default_case_nonstandard instead
25485 of failing if contiguous_range and the default case doesn't have empty
25487 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25488 and only the default case doesn't have the required constants. Skip
25490 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25491 if default_case_nonstandard.
25492 (build_constructors): Build constant 1 just once. Assert that default
25493 values aren't inserted in between cases if contiguous_range. Skip
25495 (build_arrays): Skip virtual phis.
25496 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25497 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25498 Handle virtual phis.
25499 (gen_inbound_check): Handle default_case_nonstandard case.
25500 (process_switch): Adjust check_final_bb caller. Call
25501 gather_default_values with the first non-default case instead of
25502 default case if default_case_nonstandard.
25504 2017-05-02 Nathan Sidwell <nathan@acm.org>
25506 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25507 check. Fix formatting.
25509 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25511 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25512 errors when comparing specialized and unspecialized times.
25514 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25516 * diagnostic-show-locus.c
25517 (layout::should_print_annotation_line_p): Make private.
25518 (layout::print_annotation_line): Make private.
25519 (layout::annotation_line_showed_range_p): Make private.
25520 (layout::show_ruler): Make private.
25521 (layout::print_source_line): Make private. Pass in line and
25522 line_width, rather than calling location_get_source_line. Drop
25524 (layout::print_leading_fixits): New method.
25525 (layout::print_any_fixits): Rename to...
25526 (layout::print_trailing_fixits): ...this, and make private.
25527 Don't print newline fixits.
25528 (diagnostic_show_locus): Move logic for printing one row into...
25529 (layout::print_line): ...this new function. Move the
25530 location_get_source_line call and error-handling from
25531 print_source_line to here. Call print_leading_fixits, and rename
25532 print_any_fixits to print_trailing_fixits.
25533 (selftest::test_fixit_insert_containing_newline): Update now that
25534 newlines are partially supported.
25535 (selftest::test_fixit_insert_containing_newline_2): New test.
25536 (selftest::test_fixit_replace_containing_newline): Update comments.
25537 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25538 * edit-context.c (class added_line): New class.
25539 (class edited_line): Describe newline handling in comment.
25540 (edited_line::actually_edited_p): New method.
25541 (edited_line::print_content): Delete redundant decl.
25542 (edited_line::m_predecessors): New field.
25543 (edited_file::print_content): Call edited_line::print_content.
25544 (edited_file::print_diff): Update to support newlines.
25545 (edited_file::print_diff_hunk): Likewise.
25546 (edited_file::print_run_of_changed_lines): New function.
25547 (edited_file::print_diff_line): Convert to...
25548 (print_diff_line): ...this.
25549 (edited_file::get_effective_line_count): New function.
25550 (edited_line::edited_line): Initialize new field m_predecessors.
25551 (edited_line::~edited_line): Clean up m_predecessors.
25552 (edited_line::apply_fixit): Handle newlines.
25553 (edited_line::get_effective_line_count): New function.
25554 (edited_line::print_content): New function.
25555 (edited_line::print_diff_lines): New function.
25556 (selftest::test_applying_fixits_insert_containing_newline): New
25558 (selftest::test_applying_fixits_replace_containing_newline): New
25560 (selftest::insert_line): New function.
25561 (selftest::test_applying_fixits_multiple_lines): Add example of
25563 (selftest::edit_context_c_tests): Call the new tests.
25565 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25567 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25568 parameter cand. Update dump information.
25569 (get_computation_cost): Update uses.
25571 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25573 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25574 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25575 (get_computation_at, rewrite_use_address): Update use of
25576 get_computation_aff.
25578 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25580 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25581 (get_computation): Delete.
25582 (get_computation_cost): Implement like get_computation_cost_at.
25583 Use get_computation_at.
25584 (get_computation_cost_at): Delete.
25585 (rewrite_use_nonlinear_expr): Use get_computation_at.
25586 (rewrite_use_compare, remove_unused_ivs): Ditto.
25588 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25590 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25592 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25594 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25595 (ivopts_global_cost_for_size): Rename parameter and update uses.
25596 (iv_ca_recount_cost): Update uses.
25597 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25598 candidates seperately in n_invs and n_cands.
25599 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25601 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25603 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25604 (find_inv_vars_cb): New.
25605 (find_depends): Renamed to ...
25606 (find_inv_vars): ... this.
25607 (add_candidate_1, force_var_cost): Call find_inv_vars.
25608 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25610 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25612 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25613 inv_vars. Add inv_exprs.
25614 (struct iv_cand): Rename depends_on to inv_vars.
25615 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25616 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25617 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25618 (dump_cand): Dump inv_vars.
25619 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25620 (record_invariant, find_depends, add_candidate_1): Ditto.
25621 (set_group_iv_cost, force_var_cost): Ditto.
25622 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25623 (get_computation_cost_at, get_computation_cost): Ditto.
25624 (determine_group_iv_cost_generic): Ditto.
25625 (determine_group_iv_cost_address): Ditto.
25626 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25627 (determine_group_iv_costs): Ditto.
25628 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25629 (iv_ca_set_remove_invariants): Renamed to ...
25630 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25631 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25632 (iv_ca_set_add_invariants): Renamed to ...
25633 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25634 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25635 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25636 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25637 (create_new_ivs): Remove useless dump.
25639 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25641 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25643 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25644 (iv_ca_set_no_cp, create_new_iv): Ditto.
25646 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25648 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25650 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25652 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25653 function tree_check2.
25655 2017-05-02 Martin Liska <mliska@suse.cz>
25657 * doc/gcov.texi: Add missing preposition.
25658 * gcov.c (function_info::function_info): Properly fill up
25659 all member variables.
25661 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25663 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25665 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25667 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25669 2017-05-02 Martin Liska <mliska@suse.cz>
25672 * lto-streamer-in.c (lto_read_tree_1): Remove
25673 LTO_STREAMER_DEBUG.
25674 * lto-streamer.c (struct tree_hash_entry): Likewise.
25675 (struct tree_entry_hasher): Likewise.
25676 (tree_entry_hasher::hash): Likewise.
25677 (tree_entry_hasher::equal): Likewise.
25678 (lto_streamer_init): Likewise.
25679 (lto_orig_address_map): Likewise.
25680 (lto_orig_address_get): Likewise.
25681 (lto_orig_address_remove): Likewise.
25682 * lto-streamer.h: Likewise.
25683 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25684 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25686 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25688 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25689 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25690 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25691 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25692 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25693 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25694 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25695 (mm_maskz_sub_ss): New intrinsics.
25696 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25697 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25698 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25699 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25700 (__builtin_ia32_subss_mask_round): New builtins.
25701 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25702 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25703 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25705 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25706 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25708 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25711 2017-05-02 Martin Jambor <mjambor@suse.cz>
25713 PR tree-optimization/78687
25714 * tree-sra.c (access): New field parent.
25715 (process_subtree_disqualification): New function.
25716 (disqualify_candidate): Call it.
25717 (build_accesses_from_assign): Reset write flag if creating an
25719 (build_access_subtree): Fill in parent field and also prpagate
25720 down grp_write flag.
25721 (create_artificial_child_access): New parameter set_grp_write, set
25722 grp_write to its value.
25723 (propagate_subaccesses_across_link): Also propagate grp_write flag
25725 (propagate_all_subaccesses): Push the closest parent back to work
25726 queue if add_access_to_work_queue returned true.
25728 2017-05-02 Richard Biener <rguenther@suse.de>
25730 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25731 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25732 -fstrict-overflow documentation.
25733 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25734 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25735 flag_strict_overflow.
25736 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25737 * lto-opts.c (lto_write_options): Do not stream it.
25738 * lto-wrapper.c (merge_and_complain): Do not handle it.
25739 * opts.c (default_options_table): Do not set -fstrict-overflow.
25740 (finish_options): Likewise do not clear it when sanitizing.
25741 * simplify-rtx.c (simplify_const_relational_operation): Do not
25742 test flag_strict_overflow.
25744 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25746 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25747 using enabled attribute.
25748 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25749 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25750 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25751 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25752 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25753 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25754 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25755 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25756 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25757 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25759 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25761 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25763 2017-05-02 Richard Biener <rguenther@suse.de>
25765 PR tree-optimization/80591
25767 2017-04-10 Richard Biener <rguenther@suse.de>
25769 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25772 2017-05-02 Richard Biener <rguenther@suse.de>
25774 PR tree-optimization/80549
25775 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25776 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25777 headers if they do not have a preheader.
25779 2017-05-02 Martin Liska <mliska@suse.cz>
25782 * common.opt: Fix typo.
25783 * doc/invoke.texi: Likewise.
25785 2017-05-01 Jan Beulich <jbeulich@suse.com>
25787 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25788 swapping, add (x,x,m,x,n) alternative.
25790 2017-05-01 Nathan Sidwell <nathan@acm.org>
25792 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25793 unnecessary unadjusted_alignment check.
25795 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25798 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25800 * gimplify.c (gimplify_cilk_detach): New function.
25801 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25802 * tree-core.h: Document EXPR_CILK_SPAWN.
25803 * tree.h (EXPR_CILK_SPAWN): Define.
25805 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25807 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25808 to use new fixit_hint representation, using the "replace" logic.
25809 (get_line_span_for_fixit_hint): Likewise.
25810 (layout::print_any_fixits): Likewise.
25811 (selftest::test_one_liner_many_fixits): Rename to...
25812 (selftest::test_one_liner_many_fixits_1): ...this, and update
25813 comment and expected output to reflect that the multiple fix-it
25814 hints are now consolidated into one insertion.
25815 (selftest::test_one_liner_many_fixits_2): New test.
25816 (selftest::test_diagnostic_show_locus_one_liner): Update for
25818 (selftest::test_fixit_consolidation): Update for fix-it API
25820 * diagnostic.c (print_parseable_fixits): Likewise.
25821 * edit-context.c (edited_line::m_line_events): Convert from
25822 auto_vec <line_event *> to auto_vec <line_event>.
25823 (class line_event): Convert from abstract base class to a concrete
25824 class, taking over the role of replace_event.
25825 (class insert_event): Delete.
25826 (class replace_event): Rename to class line_event. Convert to
25828 (edit_context::add_fixits): Reimplement.
25829 (edit_context::apply_insert): Delete.
25830 (edit_context::apply_replace): Rename to...
25831 (edit_context::apply_fixit): ...this. Convert to half-open range.
25832 (edited_file::apply_insert): Delete.
25833 (edited_file::apply_replace): Rename to...
25834 (edited_file::apply_fixit): ...this.
25835 (edited_line::~edited_line): Drop deletion of events.
25836 (edited_line::apply_insert): Delete.
25837 (edited_line::apply_replace): Rename to...
25838 (edited_line::apply_fixit): ...this. Convert to half-open range.
25839 Update for change to type of m_line_events.
25840 * edit-context.h (edit_context::apply_insert): Delete.
25841 (edit_context::apply_replace): Rename to...
25842 (edit_context::apply_fixit): ...this.
25844 2017-05-01 Martin Sebor <msebor@redhat.com>
25846 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25849 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25852 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25853 __get_cpuid_max returns 0.
25854 (__get_cpuid_count): Ditto.
25856 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25858 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25859 replacement expression is another instance of one of its arguments.
25861 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25864 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25865 check for stack push/pop autoinc.
25866 * config/i386/i386.c (ix86_agi_dependent): Return false
25867 if the only reason why modified_in_p returned true is that
25868 addr is SP based and set_insn is a push or pop.
25870 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25872 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25875 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25878 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25879 (account_size_time): Use two predicates - exec_pred and
25881 (evaluate_conditions_for_known_args): Compute both clause and
25883 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25884 (inline_summary_t::duplicate): Update.
25885 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25887 (compute_inline_parameters): Likewise.
25888 (estimate_edge_size_and_time): Update caluclation of time.
25889 (estimate_node_size_and_time): Compute both time and nonspecialized
25891 (estimate_ipcp_clone_size_and_time): Update.
25892 (inline_merge_summary): Update.
25893 (do_estimate_edge_time): Update.
25894 (do_estimate_edge_size): Update.
25895 (do_estimate_edge_hints): Update.
25896 (inline_read_section, inline_write_summary): Stream both new predicates.
25897 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25899 (compute_inlined_call_time): Cleanup.
25900 (big_speedup_p): Update.
25901 (edge_badness): Update.
25902 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25903 (size_time_entry): Replace predicate by exec_predicate and
25904 nonconst_predicate.
25905 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25906 (estimate_edge_time): Return also nonspec_time.
25907 (reset_edge_growth_cache): Update.
25909 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25911 PR rtl-optimization/80491
25912 * ifcvt.c (noce_process_if_block): When looking for x setter
25913 with missing else_bb, don't check only the insn right before
25914 cond_earliest, but look for the last insn that x is modified in
25915 within the same bb.
25917 PR rtl-optimization/80491
25918 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25920 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25922 PR tree-optimization/80487
25923 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25925 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25927 PR tree-optimization/79697
25928 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25929 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25930 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25932 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25933 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25935 2017-04-28 Martin Sebor <msebor@redhat.com>
25937 PR tree-optimization/80523
25938 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25939 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25941 (maybe_warn, format_directive, parse_directive): Use new functions.
25942 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25944 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25946 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25948 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25950 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25951 target_header_dir): Set correctly.
25952 * configure: Regenerated.
25953 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25954 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25955 instead of SYSTEM_HEADER_DIR.
25957 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25959 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25960 (estimate_local_effects): Likewise.
25961 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25962 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25963 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25964 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25965 do_estimate_edge_time, estimate_edge_time): Likewise.
25966 * ipa-inline-analysis.c (estimate_node_size_and_time,
25967 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25968 (estimate_time_after_inlining): Remove.
25970 2017-04-28 Martin Liska <mliska@suse.cz>
25972 * doc/gcov.texi: Enhance documentation of gcov.
25974 2017-04-28 Martin Liska <mliska@suse.cz>
25976 * doc/gcov.texi: Sort options in alphabetic order.
25977 * doc/gcov-dump.texi: Likewise.
25978 * doc/gcov-tool.texi: Likewise.
25979 * gcov.c (print_usage): Likewise.
25980 * gcov-dump.c (print_usage): Likewise.
25981 * gcov-tool.c (print_merge_usage_message): Likewise.
25982 (print_rewrite_usage_message): Likewise.
25983 (print_overlap_usage_message): Likewise.
25985 2017-04-28 Martin Liska <mliska@suse.cz>
25987 PR gcov-profile/53915
25988 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25990 2017-04-28 Martin Liska <mliska@suse.cz>
25992 PR gcov-profile/79891
25993 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25994 is marked by compiler as living on a line.
25995 (get_cycles_count): Remove usage of the union.
25996 (output_intermediate_file): Likewise.
25997 (find_source): Fix GNU coding style.
25998 (accumulate_line_counts): Remove old non-all block mode.
25999 (output_lines): Remove usage of the union.
26000 * profile.c (output_location): Include all BBs, even if
26001 belonging to a same line (and file) as a previous BB.
26003 2017-04-28 Martin Liska <mliska@suse.cz>
26005 * gcov.c (process_args): Handle new argument 'w'.
26006 (read_graph_file): Assign ID to BBs.
26007 (output_branch_count): Display BB # if verbose flag is set.
26008 (output_lines): Likewise for arcs.
26009 (print_usage): Add '--verbose' option help.
26010 * doc/gcov.texi: Document --verbose (-w) option.
26012 2017-04-28 Martin Liska <mliska@suse.cz>
26014 * gcov.c (struct block_location_info): New struct.
26015 (process_file): Fill up the new structure.
26016 (read_graph_file): Replace usage of encoding by the newly added
26018 (add_line_counts): Likewise.
26019 (accumulate_line_counts): Remove usage of the union.
26020 (function_info::function_info): New function.
26021 (function_info::~function_info): Likewise.
26022 (process_file): Call delete instead of release_function.
26023 (release_function): Release the function.
26024 (release_structures): Call delete instead of release_function.
26025 (solve_flow_graph): Replace usage of num_blocks.
26026 (find_exception_blocks): Likewise.
26027 (output_lines): Fix GNU coding style.
26029 2017-04-28 Martin Liska <mliska@suse.cz>
26032 * coverage.c (coverage_remove_note_file): New function.
26033 * coverage.h: Declare the function.
26034 * toplev.c (finalize): Clean if an error has been seen.
26036 2017-04-28 Martin Liska <mliska@suse.cz>
26038 PR gcov-profile/80031
26039 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26040 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26041 * gcov.c (read_graph_file): Read just number of blocks.
26042 * profile.c (branch_prob): Do not stream 0 flags per a basic
26045 2017-04-28 Martin Liska <mliska@suse.cz>
26047 * gcov-dump.c (tag_*): Add new argument to declarations.
26048 (dump_gcov_file): Likewise.
26049 (tag_blocks): Add and use new argument depth.
26050 (tag_arcs): Likewise.
26051 (tag_lines): Likewise.
26052 (tag_counters): Likewise.
26053 (tag_summary): Likewise.
26054 (dump_working_sets): Use depth to do a proper indentation.
26056 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26059 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26060 * symtab.c (symtab_node::debug_symtab): Move definition here.
26062 2017-04-28 Richard Biener <rguenther@suse.de>
26064 * lto-streamer.h (LTO_major_version): Bump to 7.
26066 2017-04-28 Richard Biener <rguenther@suse.de>
26068 * tree-vrp.c (assert_info): New struct.
26069 (add_assert_info): New helper.
26070 (register_edge_assert_for_2): Refactor to add asserts to a vector
26072 (register_edge_assert_for_1): Likewise.
26073 (register_edge_assert_for): Likewise.
26074 (finish_register_edge_assert_for): New helper actually registering
26075 asserts where live on edge.
26076 (find_conditional_asserts): Adjust.
26077 (find_switch_asserts): Likewise.
26078 (evrp_dom_walker::try_find_new_range): Generalize.
26079 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26081 2017-04-27 Marek Polacek <polacek@redhat.com>
26084 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26085 arg10 and arg11 to itype.
26087 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26089 * doc/extend.texi (Object Size Checking): Improve grammar.
26091 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26094 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26095 that the logic for permitting reciprocal estimates matches that
26098 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26101 * tree.c (type_cache_hasher::equal): Only compare
26102 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26103 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26104 non-aggregate element types.
26105 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26106 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26108 2017-04-27 Richard Biener <rguenther@suse.de>
26110 PR middle-end/80533
26111 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26112 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26113 keeping a reference to a trailing array.
26115 2017-04-27 Richard Biener <rguenther@suse.de>
26117 PR middle-end/80539
26118 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26119 being in loop-closed SSA form conservatively.
26120 (chrec_fold_multiply_poly_poly): Likewise.
26122 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26124 PR middle-end/79665
26125 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26126 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26128 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26131 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26132 (aarch64_function_arg_alignment): Return unsigned int again, but still
26133 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26134 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26135 Don't emit -Wpsabi note.
26136 (aarch64_function_arg_boundary): Likewise.
26137 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26140 2017-04-26 Nathan Sidwell <nathan@acm.org>
26142 * tree.h (crc32_unsigned_n): Declare.
26143 (crc32_unsigned, crc32_unsigned): Make inline.
26144 * tree.c (crc32_unsigned_bits): Replace with ...
26145 (crc32_unsigned_n): ... this.
26146 (crc32_unsigned, crc32_byte): Remove.
26147 (crc32_string): Remove unnecessary braces.
26149 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26151 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26152 * ipa-inline-analysis.c (MAX_TIME): Remove.
26153 (account_size_time): Use sreal for time.
26154 (dump_inline_summary): Update.
26155 (estimate_function_body_sizes): Update.
26156 (estimate_edge_size_and_time): Update.
26157 (estimate_calls_size_and_time): Update.
26158 (estimate_node_size_and_time): Update.
26159 (inline_merge_summary): Update.
26160 (inline_update_overall_summary): Update.
26161 (estimate_time_after_inlining): Update.
26162 (inline_read_section): Update.
26163 (inline_write_summary): Update.
26164 * ipa-inline.c (compute_uninlined_call_time): Update.
26165 (compute_inlined_call_time): Update.
26166 (recursive_inlining): Update.
26167 (inline_small_functions): Update.
26168 (dump_overall_stats): Update.
26169 * ipa-inline.h: Include sreal.h.
26170 (size_time_entry): Turn time to sreal.
26171 (inline_summary): Turn self_time nad time to sreal.
26173 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26175 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26177 (sreal::stream_out, sreal::stream_in): New.
26178 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26180 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26182 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26185 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26188 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26190 (dimode_scalar_chain::compute_convert_gain): Ditto.
26191 (dimode_scalar_chain::make_vector_copies): Ditto.
26192 (dimode_scalar_chain::convert_reg): Ditto.
26193 (dimode_scalar_chain::convert_insn): Ditto.
26194 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26195 (VI248_AVX512BW_1): New mode iterator.
26196 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26197 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26200 2017-04-25 Martin Sebor <msebor@redhat.com>
26202 PR tree-optimization/80497
26203 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26204 constants are representable in HOST_WIDE_INT.
26205 (parse_directive): Ditto.
26207 2017-04-25 Martin Sebor <msebor@redhat.com>
26210 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26211 (new_zero_array): Adjust signature.
26212 (dom_info::dom_init): Used unsigned rather that size_t.
26213 (dom_info::dom_info): Same.
26215 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26216 Jakub Jelinek <jakub@redhat.com>
26219 * config/arm/arm.c: Include gimple.h.
26220 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26221 returns negative, increment ncrn only if it returned positive.
26222 (arm_needs_doubleword_align): Return int instead of bool,
26223 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26224 members, but if there is any such non-FIELD_DECL
26225 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26226 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26227 returns negative, increment nregs only if it returned positive.
26228 (arm_setup_incoming_varargs): Likewise.
26229 (arm_function_arg_boundary): Emit -Wpsabi note if
26230 arm_needs_doubleword_align returns negative, return
26231 DOUBLEWORD_ALIGNMENT only if it returned positive.
26233 2017-04-25 Marek Polacek <polacek@redhat.com>
26236 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26237 first argument to type.
26239 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26242 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26243 type checks to test for compatibility instead of equality.
26245 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26246 Jakub Jelinek <jakub@redhat.com>
26249 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26251 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26252 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26253 the alignment computation, but return their maximum in warn_alignment.
26254 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26255 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26257 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26258 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26261 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26263 * config/arc/simdext.md (dmpyh): Fix typo.
26265 2017-04-25 Richard Biener <rguenther@suse.de>
26267 PR tree-optimization/80492
26268 * alias.c (compare_base_decls): Handle registers with asm
26269 specification conservatively.
26270 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26271 compare_base_decls returning dont-know properly.
26273 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26275 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26276 (legitimate_offset_address_p): New function.
26277 (arc_legitimate_address_p): Use above function.
26279 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26281 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26285 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26286 ACCH registers whenever they are available.
26288 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26290 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26291 double regs fix when not used.
26293 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26295 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26297 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26298 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26300 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26302 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26303 when emitting short ADD instructions.
26305 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26307 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26309 (cmpsi_cc_c_insn): Likewise.
26310 (cbranchsi4_scratch): Compute proper instruction length using
26311 compact_hreg_operand.
26312 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26314 2017-04-25 Richard Biener <rguenther@suse.de>
26316 PR middle-end/80509
26317 * passes.c (pass_manager::pass_manager): Initialize
26318 m_name_to_pass_map.
26320 2017-04-25 Richard Biener <rguenther@suse.de>
26322 PR tree-optimization/79201
26323 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26325 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26328 * config/s390/vector.md: Split MEM->GPR vector moves for
26329 non-s_operand addresses.
26331 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26334 * config/s390/predicates.md (reload_const_wide_int_operand): New
26336 * config/s390/s390.md ("movti"): Remove d/P alternative.
26337 ("movti_bigconst"): New pattern definition.
26339 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26342 * s390-protos.h (s390_expand_cs_hqi): Removed.
26343 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26344 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26345 modes as well as CCZ1mode and CCZmode.
26346 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26347 signature of s390_emit_compare_and_swap.
26348 (s390_expand_cs_hqi): Likewise, make static.
26349 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26350 compare-and-swap, in some cases.
26351 (s390_expand_cs): Wrapper function.
26352 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26354 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26355 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26356 patterns for small and large integers. Forbid symref memory operands.
26357 Move expander to s390.c. Require cc register.
26358 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26359 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26360 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26361 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26362 symref memory operands. Remove CC mode and call s390_match_ccmode
26364 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26366 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26368 * config/s390/s390.md (define_peephole2): New peephole to help
26369 combining the load-and-test pattern with volatile memory.
26371 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26373 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26374 with CCZmode for TARGET_Z196.
26376 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26378 PR rtl-optimization/80501
26379 * combine.c (make_compound_operation_int): Set subreg_code to SET
26380 even for AND with mask of the sign bit of mode.
26382 PR rtl-optimization/80500
26383 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26384 sum's initial value.
26386 2017-04-25 Julian Brown <julian@codesourcery.com>
26387 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26389 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26391 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26393 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26395 2017-04-25 Julian Brown <julian@codesourcery.com>
26396 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26398 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26399 (thunderx2t99_sha): New Reservation.
26401 2017-04-25 Julian Brown <julian@codesourcery.com>
26402 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26404 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26405 type for 1-element load.
26407 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26409 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26411 2017-04-24 Martin Jambor <mjambor@suse.cz>
26413 PR tree-optimization/80293
26414 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26415 char arrays not totally scalarizable if it is false.
26416 (analyze_all_variable_accesses): Pass correct value in the new
26417 parameter. Add a statistics counter.
26419 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26421 PR middle-end/79931
26422 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26424 2017-04-24 Richard Biener <rguenther@suse.de>
26426 PR tree-optimization/80494
26427 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26428 out for complex types.
26430 2017-04-24 Richard Biener <rguenther@suse.de>
26432 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26433 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26434 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26435 (DFS): Adjust and never fail.
26436 (sccvn_dom_walker::fail): Remove.
26437 (sccvn_dom_walker::before_dom_children): Adjust.
26438 (run_scc_vn): Likewise and never fail.
26439 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26440 (pass_fre::execute): Likewise.
26442 2017-04-24 Richard Biener <rguenther@suse.de>
26444 PR tree-optimization/79725
26445 * tree-ssa-sink.c (statement_sink_location): Return whether
26446 failure reason was zero uses. Move that check later.
26447 (sink_code_in_bb): Deal with zero uses by removing the stmt
26450 2017-04-24 Richard Biener <rguenther@suse.de>
26453 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26454 pointer-based references.
26456 2017-04-24 Richard Biener <rguenther@suse.de>
26459 * pass_manager.h (pass_manager::operator new): Remove.
26460 (pass_manager::operator delete): Likewise.
26461 * passes.c (pass_manager::operator new): Remove.
26462 (pass_manager::operator delete): Likewise.
26463 (pass_manager::pass_manager): Zero individual pass members.
26465 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26468 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26469 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26470 Check "XEXP (src, 1)" operand here.
26471 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26472 Check "XEXP (src, 1)" operand here.
26473 (dimode_scalar_chain::make_vector_copies): Detect count register
26474 of a shift instruction. Zero extend count register from QImode
26475 to DImode to satisfy vector shift pattern count operand predicate.
26476 Substitute vector shift count operand with a DImode copy.
26477 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26480 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26482 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26483 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26484 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26485 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26486 (UNSPEC_NOREX_MEM): Remove definition.
26488 2017-04-21 Richard Biener <rguenther@suse.de>
26490 PR tree-optimization/79547
26491 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26492 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26493 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26494 without any constraints.
26496 2017-04-21 Richard Biener <rguenther@suse.de>
26498 PR tree-optimization/78847
26499 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26501 2017-04-21 Richard Biener <rguenther@suse.de>
26503 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26504 (build_distinct_type_copy): Likewise.
26505 (build_variant_type_copy): Likewise.
26506 * tree.c (build_qualified_type): Pass down mem-stat info.
26507 (build_distinct_type_copy): Likewise.
26508 (build_variant_type_copy): Likewise.
26510 2017-04-21 Richard Biener <rguenther@suse.de>
26512 PR tree-optimization/80237
26513 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26515 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26516 for a simplified result.
26518 2016-04-21 Richard Biener <rguenther@suse.de>
26520 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26521 sth as strict as a simple_iv but a chrec without symbols and an
26522 operand defined in the loop we are peeling (and not some subloop).
26523 (propagate_constants_for_unrolling): Propagate all constants.
26525 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26528 * config/i386/i386.c (print_reg): Remove assert for disalowed
26529 regno values, call output_operand_lossage instead.
26531 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26534 * config/i386/constraints.md (Yc): New register constraint.
26535 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26536 Use Yc constraint for alternative 2 of operand 0. Remove
26537 preferred_for_speed attribute.
26539 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26541 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26542 lastprivate clauses in SIMT case.
26544 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26546 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26548 2017-04-20 Richard Biener <rguenther@suse.de>
26550 PR tree-optimization/57796
26551 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26552 as N scalar stores.
26553 (vect_model_load_cost): Cost gathers as N scalar loads.
26555 2017-04-20 Richard Biener <rguenther@suse.de>
26557 * ggc-page.c (ggc_allocated_p): Rename to ...
26558 (safe_lookup_page_table_entry): ... this and return the lookup
26560 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26562 2017-04-20 Richard Biener <rguenther@suse.de>
26564 PR tree-optimization/80453
26565 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26566 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26567 from the conditions.
26568 (vn_phi_eq): Pass them down.
26569 (vn_phi_lookup): Record them.
26570 (vn_phi_insert): Likewise.
26572 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26574 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26575 uninitialized variable warning to avoid buffer overrun.
26577 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26580 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26581 is suppressed for '{ 0 }' in C.
26583 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26585 * BASE-VER: Set to 8.0.0.
26587 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26589 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26590 priority .init_array and .fini_array section with SECTION_NOTYPE
26593 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26595 PR middle-end/80423
26596 * tree.h (build_array_type): Add typeless_storage default argument.
26597 * tree.c (type_cache_hasher::equal): Also compare
26598 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26599 (build_array_type): Add typeless_storage argument, set
26600 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26602 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26603 (build_array_type): Likewise. Add typeless_storage argument.
26605 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26606 Jakub Jelinek <jakub@redhat.com>
26608 PR tree-optimization/80426
26609 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26610 operation on symbolic operands, also compute the overflow for the
26611 invariant part when the operation degenerates into a negation.
26613 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26616 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26617 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26620 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26622 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26625 * config/avr/avr.c (tree.h): Include it.
26626 (cgraph.h): Include it.
26627 (avr_encode_section_info): Don't warn for uninitialized progmem
26628 variable if it's just an alias.
26630 2017-04-19 Richard Biener <rguenther@suse.de>
26633 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26634 when needed by AutoPGO.
26636 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26639 * doc/lto.texi: Remove an extra 'that'.
26641 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26643 PR rtl-optimization/80429
26644 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26645 are only used in debug insns.
26647 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26648 Vladimir Makarov <vmakarov@redhat.com>
26650 * config/sparc/predicates.md (input_operand): Add comment. Return
26651 true for any memory operand when LRA is in progress.
26652 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26654 2017-04-18 Jeff Law <law@redhat.com>
26657 * mips.md ({return,simple_return}_internal): Do not overwrite
26660 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26662 PR tree-optimization/80443
26663 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26664 instead of adding 1, subtract -1 and similarly instead of subtracting
26667 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26669 PR rtl-optimization/80357
26670 * haifa-sched.c (tmp_bitmap): New variable.
26671 (model_recompute): Handle duplicate use records.
26672 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26673 (free_global_sched_pressure_data): Free it.
26675 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26678 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26679 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26680 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26681 instead of SYSTEM_HEADER_DIR.
26683 2017-04-18 Jeff Law <law@redhat.com>
26685 PR middle-end/80422
26686 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26687 predecessors after walking up the insn chain.
26689 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26692 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26693 sizetype type into debug info.
26695 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26698 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26699 unneeded test for TARGET_UPPER_REGS_SF.
26700 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26702 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26705 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26706 instead of gsi_after_labels.
26708 2017-04-18 Jeff Law <law@redhat.com>
26710 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26714 2017-04-13 Jeff Law <law@redhat.com>
26715 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26716 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26718 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26721 * config/avr/avr.c (intl.h): Include it.
26722 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26724 2017-04-18 Martin Liska <mliska@suse.cz>
26726 PR gcov-profile/78783
26727 * gcov-tool.c (gcov_output_files): Validate that destination
26728 file is either removed by the tool or by a user.
26730 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26731 Guy Benyei <guybe@mellanox.com>
26733 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26734 block, and do not negate it, the stored id is already negative.
26736 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26738 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26740 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26743 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26744 masks of options that should be turned off if the VSX vector
26745 options are turned off.
26746 (OTHER_P8_VECTOR_MASKS): Likewise.
26747 (OTHER_VSX_VECTOR_MASKS): Likewise.
26748 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26749 rs6000_disable_incompatible_switches to validate no type switches
26751 (rs6000_incompatible_switch): New function to disallow turning on
26752 other vector options if -mno-vsx, -mno-power8-vector, or
26753 -mno-power9-vector are specified.
26755 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26757 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26759 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26761 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26762 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26763 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26764 (ARG_POINTER_CFA_OFFSET): Likewise.
26766 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26768 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26769 conditions to take advantage of various optimizations.
26771 2017-04-13 Jeff Law <law@redhat.com>
26773 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26774 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26775 (zero_extendsidi2_dext): Likewise.
26777 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26780 * fold-const.c (fold_ternary_loc): Revert
26781 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26784 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26786 PR rtl-optimization/80343
26787 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26788 new scratch pseudo.
26790 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26793 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26794 to ubsan_encode_value.
26796 2017-04-13 Jeff Law <law@redhat.com>
26798 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26799 appearing in DEBUG_INSNs.
26801 2017-04-13 Martin Liska <mliska@suse.cz>
26803 PR gcov-profile/80413
26804 * gcov-io.c (gcov_write_string): Copy to buffer just when
26805 allocated size is greater than zero.
26807 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26810 * dwarf2out.c (decls_for_scope): Ignore declarations of
26811 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26813 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26816 * ipa-visibility.c (non_local_p): Fix typos.
26817 (localize_node): When localizing symbol in same comdat group,
26818 dissolve the group only when we know external symbols are going
26820 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26822 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26824 PR tree-optimization/79390
26825 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26826 order does not result in usable sequence, retry with reversed operand
26832 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26833 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26834 op0 instead of fold_convert_loc (loc, type, arg0).
26836 2017-04-12 Jeff Law <law@redhat.com>
26838 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26839 has a delay slot in the generated code.
26841 * config/cris/cris.md (cris_preferred_reload_class): Return
26842 GENNONACR_REGS rather than GENERAL_REGS.
26844 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26847 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26848 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26849 signedness of the result type.
26851 2017-04-12 Richard Biener <rguenther@suse.de>
26852 Jeff Law <law@redhat.com>
26854 PR tree-optimization/80359
26855 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26856 trim stores to TARGET_MEM_REFs.
26858 2017-04-12 Richard Biener <rguenther@suse.de>
26860 PR tree-optimization/79390
26861 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26862 threading case even more.
26864 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26867 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26868 for quad_address_p for TImode, instead of just not indexed_address.
26870 2017-04-12 Richard Biener <rguenther@suse.de>
26871 Bernd Edlinger <bernd.edlinger@hotmail.de>
26873 PR middle-end/79671
26874 * alias.c (component_uses_parent_alias_set_from): Handle
26875 TYPE_TYPELESS_STORAGE.
26876 (get_alias_set): Likewise.
26877 * tree-core.h (tree_type_common): Add typeless_storage flag.
26878 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26879 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26880 for types containing members with TYPE_TYPELESS_STORAGE.
26881 (place_field): Likewise.
26882 (layout_type): Likewise for ARRAY_TYPE.
26883 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26884 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26885 TYPE_TYPELESS_STORAGE.
26886 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26888 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26891 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26892 first argument to type.
26894 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26898 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26899 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26900 (rs6000_expand_binop_builtin): Likewise.
26901 (rs6000_expand_ternop_builtin): Likewise; also add missing
26902 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26903 vshasigma built-ins.
26904 * doc/extend.texi: Document that vec_xxpermdi's third argument
26905 must be a constant.
26907 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26909 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26910 Use shift_const cost parameter when calculating gain of STV shifts.
26912 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26914 PR rtl-optimization/70478
26915 * lra-constraints.c (process_alt_operands): Check memory for
26916 disfavoring memory insn operand.
26918 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26920 PR middle-end/80100
26921 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26922 left shift in unsigned HOST_WIDE_INT type.
26924 PR rtl-optimization/80385
26925 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26926 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26929 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26930 if they have any depend clauses.
26932 2017-04-11 Martin Liska <mliska@suse.cz>
26935 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26936 * ipa-split.c (split_function): Create a local comdat symbol
26937 if caller is in a comdat group.
26939 2017-04-11 Martin Liska <mliska@suse.cz>
26942 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26945 2017-04-11 Martin Sebor <msebor@redhat.com>
26947 PR middle-end/80364
26948 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26949 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26951 (directive::set_width, directive::set_precision, format_character):
26953 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26956 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26959 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26960 conflict, set target->arch_name instead of target->cpu_name.
26962 2017-04-11 Richard Biener <rguenther@suse.de>
26964 PR tree-optimization/80374
26965 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26966 build_zero_cst, remove fold_convertible_p check again.
26968 2017-04-11 Martin Liska <mliska@suse.cz>
26971 * ubsan.c (instrument_object_size): Do not instrument register
26974 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26977 * config/i386/i386-builtin-types.def
26978 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26979 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26980 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26981 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26982 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26983 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26984 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26985 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26986 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26987 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26988 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26989 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26990 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26991 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26992 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26993 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26994 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26995 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26996 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26997 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26998 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26999 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27000 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27001 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27002 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27003 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27004 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27005 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27006 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27007 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27008 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27009 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27010 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27011 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27012 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27013 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27014 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27015 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27016 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27017 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27018 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27019 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27020 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27021 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27022 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27023 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27025 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27026 flag to second_arg_count, handle 4 argument function type _COUNT
27027 aliases, handle second_arg_count on second argument rather than last.
27029 2017-04-10 Jeff Law <law@redhat.com>
27031 PR tree-optimization/80374
27032 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27033 record anything if we can not convert integer_zero_node to the
27036 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27039 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27040 Enhance special handling given to the TARGET_P9_MINMAX option in
27041 relation to certain other options.
27043 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27045 PR tree-optimization/80153
27046 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27047 remove POINTER_PLUS_EXPR's base part directly, rather than through
27050 2017-04-10 Richard Biener <rguenther@suse.de>
27051 Bin Cheng <bin.cheng@arm.com>
27053 PR tree-optimization/80153
27054 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27055 the first element of pointer type aff_tree. Build result expr in
27057 (add_elt_to_tree): Convert to type unconditionally. Remove other
27058 fold_convert calls.
27059 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27060 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27062 2017-04-10 Richard Biener <rguenther@suse.de>
27064 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27067 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27069 PR rtl-optimization/70478
27070 * lra-constraints.c (curr_small_class_check): New.
27071 (update_and_check_small_class_inputs): New.
27072 (process_alt_operands): Update curr_small_class_check. Disfavor
27073 alternative insn memory operands. Check available regs for small
27076 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27079 * config/mips/mips.opt (-mvirt): Update description.
27080 * doc/invoke.texi (-mvirt): Likewise.
27082 2017-04-10 Richard Biener <rguenther@suse.de>
27084 PR middle-end/80362
27085 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27086 looking for NEGATE_EXPR in -A / -B to A / B folding.
27088 2017-04-10 Martin Liska <mliska@suse.cz>
27090 PR gcov-profile/80224
27091 * gcov.c (print_usage): Fix usage string.
27092 (get_gcov_intermediate_filename): Remove.
27093 (output_gcov_file): Use both for normal and intermediate format.
27094 (generate_results): Do not initialize special file for
27095 intermediate format.
27097 2017-04-10 Richard Biener <rguenther@suse.de>
27099 PR tree-optimization/80304
27100 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27103 2017-04-10 Nathan Sidwell <nathan@acm.org>
27106 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27107 (rs6000_init_builtins): Use it.
27109 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27111 * config/arm/arm.md (<mrc>): Add mode to SET source.
27112 (<mrrc>): Likewise.
27114 2017-04-10 Richard Biener <rguenther@suse.de>
27116 PR middle-end/80344
27117 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27119 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27122 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27123 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27124 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27125 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27126 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27127 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27128 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27129 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27130 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27131 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27132 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27133 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27134 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27135 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27136 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27137 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27138 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27139 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27140 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27141 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27142 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27143 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27144 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27146 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27148 PR rtl-optimization/70478
27149 * lra-constraints.c: Reverse the last patch.
27151 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27153 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27154 Add comment for WCHAR_T.
27156 2017-04-08 Martin Liska <mliska@suse.cz>
27159 2017-04-07 Martin Liska <mliska@suse.cz>
27162 * ipa-split.c (split_function): Add function part to a same comdat
27165 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27168 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27170 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27172 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27173 (vec_pairing): ...to this.
27174 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27175 (rs6000_sched_init): Adjust for name change.
27176 (struct rs6000_sched_context): Likewise.
27177 (rs6000_init_sched_context): Likewise.
27178 (rs6000_set_sched_context): Likewise.
27180 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27186 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27188 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27189 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27190 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27192 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27194 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27196 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27198 PR rtl-optimization/70703
27199 * ira-color.c (update_conflict_hard_regno_costs): Use
27200 int64_t instead of HOST_WIDE_INT.
27202 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27204 PR rtl-optimization/70478
27205 * lra-constraints.c (process_alt_operands): Disfavor alternative
27206 insn memory operands.
27208 2017-04-07 Jeff Law <law@redhat.com>
27210 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27211 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27213 2017-04-07 Martin Liska <mliska@suse.cz>
27216 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27217 Show error message instead of an ICE.
27219 2017-04-07 Martin Liska <mliska@suse.cz>
27222 * ipa-split.c (split_function): Add function part to a same comdat
27225 2017-04-07 Richard Biener <rguenther@suse.de>
27227 PR middle-end/80341
27228 * tree.c (get_unwidened): Also handle ! for_type case for
27230 * convert.c (do_narrow): Split out from ...
27231 (convert_to_integer_1): ... here. Do not pass final truncation
27232 type to get_unwidened for TRUNC_DIV_EXPR.
27234 2017-04-07 Richard Biener <rguenther@suse.de>
27236 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27238 (aff_combination_const): Adjust.
27239 (aff_combination_scale): Likewise.
27240 (aff_combination_add_elt): Likewise.
27241 (aff_combination_add_cst): Likewise.
27242 (aff_combination_convert): Likewise.
27243 (add_elt_to_tree): Likewise. Remove unused argument.
27244 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27246 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27248 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27250 * config/arm/arm.c (arm_default_short_enums): Use
27251 ARM_DEFAULT_SHORT_ENUMS.
27252 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27254 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27257 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27258 members with redundant out-of-class redeclaration.
27260 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27263 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27264 * config/i386/i386.md (*zero_extendsidi2):
27265 Add (?*x,*x) and (?*v,*v) alternatives.
27267 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27270 * config/i386/i386.c (ix86_expand_builtin)
27271 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27272 mode from insn data. Convert operands to insn operand mode.
27273 Copy operands that don't satisfy insn predicate to a register.
27275 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27277 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27280 2017-04-06 Richard Biener <rguenther@suse.de>
27282 PR tree-optimization/80334
27283 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27284 preserve alignment of accesses.
27286 2017-04-06 Richard Biener <rguenther@suse.de>
27288 PR tree-optimization/80262
27289 * tree-sra.c (build_ref_for_offset): Preserve address-space
27291 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27292 Drop useless address-space information on MEM_REF offsets.
27294 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27296 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27298 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27300 PR rtl-optimization/70703
27301 * ira-color.c (update_conflict_hard_regno_costs): Use
27302 HOST_WIDE_INT instead of long.
27304 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27307 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27308 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27310 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27311 for x86_64 target. Handle -m3dnowa option.
27313 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27315 PR rtl-optimization/70703
27316 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27317 (update_conflict_hard_regno_costs): Use long instead of unsigned
27318 arithmetic for cost calculation.
27320 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27321 Bernd Edlinger <bernd.edlinger@hotmail.de>
27324 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27327 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27330 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27331 ptr_mode with Pmode throughout.
27332 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27333 into probe_stack_range and use DImode.
27335 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27338 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27339 call_eh_return is true.
27341 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27343 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27344 Initialize last_match_fntype_index.
27346 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27349 * tree-nvr.c: Include internal-fn.h.
27350 (pass_return_slot::execute): Ignore internal calls without
27353 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27354 Richard Biener <rguenther@suse.de>
27357 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27358 captures used multiple times, except for the last use.
27359 * generic-match-head.c: Include gimplify.h.
27361 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27363 PR tree-optimization/79390
27364 * target.h (struct noce_if_info): Declare.
27365 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27366 * target.def (noce_conversion_profitable_p): New target hook.
27367 * ifcvt.h (struct noce_if_info): New type, moved from ...
27368 * ifcvt.c (struct noce_if_info): ... here.
27369 (noce_conversion_profitable_p): Renamed to ...
27370 (default_noce_conversion_profitable_p): ... this. No longer
27372 (noce_try_store_flag_constants, noce_try_addcc,
27373 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27374 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27375 instead of noce_conversion_profitable_p.
27376 * config/i386/i386.c: Include ifcvt.h.
27377 (ix86_option_override_internal): Don't override
27378 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27379 (ix86_noce_conversion_profitable_p): New function.
27380 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27381 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27382 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27383 * doc/tm.texi: Regenerated.
27385 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27387 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27390 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27393 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27394 instructions for small multiply cores.
27396 2017-04-04 Jeff Law <law@redhat.com>
27398 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27400 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27401 that are not set by the loop over the elements.
27403 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27406 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27407 int mode, convert_modes it to mode as unsigned, otherwise use
27408 lowpart_subreg to mode rather than SImode.
27409 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27410 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27411 Use DImode instead of SImode for the shift count operand.
27412 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27415 2017-04-04 Richard Biener <rguenther@suse.de>
27417 PR middle-end/80281
27418 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27419 arithmetic done for the negate or the plus. Simplify.
27420 (A - (-B) -> A + B): Likewise.
27421 * fold-const.c (split_tree): Make sure to not negate pointers.
27423 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27425 PR rtl-optimization/60818
27426 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27427 a compare of comparisons with the thing compared if this results
27428 in a different machine mode.
27430 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27432 * alias.c (base_alias_check): Fix typo in comment.
27433 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27434 * cgraphunit.c (symbol_table::compile): Likewise.
27435 * collect2.c (maybe_run_lto_and_relink): Likewise.
27436 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27437 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27438 * config/avr/avr.c (avr_map_op_t): Likewise.
27439 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27440 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27441 * config/epiphany/epiphany.md (movcc): Likewise.
27442 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27443 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27445 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27446 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27447 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27448 * config/sh/sh.c (sh_rtx_costs): Likewise.
27449 * fold-const.c (fold_truth_andor): Likewise.
27450 * genautomata.c (collapse_flag): Likewise.
27451 * gengtype.h (struct type::u::s): Likewise.
27452 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27453 * input.c (FORMAT_AMOUNT): Likewise.
27454 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27455 (known_aggs_to_agg_replacement_list): Likewise.
27456 * ipa-inline-analysis.c: Likewise.
27457 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27458 * ipa-polymorphic-call.c
27459 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27460 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27461 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27463 * modulo-sched.c (apply_reg_moves): Likewise.
27464 * omp-expand.c (build_omp_regions_1): Likewise.
27465 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27466 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27467 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27468 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27469 * value-prof.c: Likewise.
27470 * var-tracking.c (val_reset): Likewise.
27472 2017-04-03 Richard Biener <rguenther@suse.de>
27474 PR tree-optimization/80275
27475 * fold-const.c (split_address_to_core_and_offset): Handle
27478 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27480 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27481 descriptors is at least equal to that of functions.
27483 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27485 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27487 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27490 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27491 (mov<IMOD4:mode>): New expander.
27492 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27494 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27496 PR rtl-optimization/79405
27497 * fwprop.c (propagations_left): New variable.
27498 (forward_propagate_into): Decrement it.
27499 (fwprop_init): Initialize it.
27500 (fw_prop): If the variable has reached zero, stop propagating.
27501 (fwprop_addr): Ditto.
27503 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27506 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27507 a FUNCTION_DECL, pass it as decl instead of origin to
27510 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27512 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27515 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27518 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27519 TARGET_VSX_SMALL_INTEGER.
27521 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27523 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27524 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27526 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27528 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27529 extraction from odd-numbered MSA register.
27531 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27533 PR middle-end/80173
27534 * expmed.c (store_bit_field_1): Don't attempt to create
27535 a word subreg out of hard registers wider than word if they
27536 have HARD_REGNO_NREGS of 1 for their mode.
27538 PR middle-end/80163
27539 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27540 conversions to integer types wider than word and pointer.
27543 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27544 (rtx_equal_for_cselib_p): Pass 0 to it.
27545 * cselib.c (cselib_hasher::equal): Likewise.
27546 (rtx_equal_for_cselib_1): Add depth argument. If depth
27547 is 128, don't look up VALUE locs and punt. Increment
27548 depth in recursive calls when walking VALUE locs.
27550 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27552 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27553 (make_gcov_file_name): Use the canonical path name for generating
27555 (read_line): Fix handling of files with ascii null bytes.
27557 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27559 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27560 to initialise a vector register instead
27561 of using a const_int.
27563 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27565 PR translation/80189
27566 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27567 diagnostic messages.
27569 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27572 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27573 (dfp_diex_<mode>): Update mode of operand 1.
27574 * doc/extend.texi (dxex, dxexq): Document change to return type.
27575 (diex, diexq): Document change to argument type.
27577 2017-03-30 Martin Jambor <mjambor@suse.cz>
27580 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27581 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27582 it reflects the signature changes performed at the callee side.
27583 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27584 to cgraph_build_function_type_skip_args.
27585 (build_function_decl_skip_args): Adjust call to the above function.
27587 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27590 * config/i386/sse.md
27591 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27592 register as dest whenever it is a MEM not rtx_equal_p to the
27593 corresponding dup operand, and when forcing into reg move the
27594 reg into the memory afterwards.
27595 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27596 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27597 for the force_reg mode.
27598 (avx512vl_vextractf128<mode>): Use register as dest either
27599 always when a MEM, or when it is a MEM not rtx_equal_p to the
27600 corresponding dup operand, or even not when it is a CONST_VECTOR
27601 depending on the mode and lo vs. hi.
27602 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27604 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27605 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27606 Likewise. Require that operands[2] is even.
27607 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27608 Remove extraneous parens. Require that operands[2] is a multiple
27610 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27611 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27612 disallow memory then.
27614 2017-03-30 Richard Biener <rguenther@suse.de>
27616 PR tree-optimization/77498
27617 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27618 to non-constants over backedges.
27620 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27622 PR rtl-optimization/80233
27623 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27624 as last_combined_insn. Do not test for BARRIER_P separately.
27626 2017-03-29 Andreas Schwab <schwab@suse.de>
27629 * calls.c (prepare_call_address): Convert funexp to Pmode before
27630 copying to temp reg.
27632 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27634 PR tree-optimization/80158
27635 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27636 Handle possible future case of more than one alternate
27638 (replace_rhs_if_not_dup): Likewise.
27639 (replace_one_candidate): Likewise.
27641 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27643 PR rtl-optimization/80193
27644 * ira.c (ira): Do not check allocation for LRA.
27646 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27648 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27649 (nvptx_output_simt_exit): Declare.
27650 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27651 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27652 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27653 (nvptx_declare_function_name): ...here. Emit declaration of local
27654 memory space buffer for omp_simt_enter insn.
27655 (nvptx_output_unisimt_switch): New.
27656 (nvptx_output_softstack_switch): New.
27657 (nvptx_output_simt_enter): New.
27658 (nvptx_output_simt_exit): New.
27659 * config/nvptx/nvptx.h (struct machine_function): New fields
27660 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27661 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27662 (UNSPECV_SIMT_EXIT): Ditto.
27663 (omp_simt_enter_insn): New insn.
27664 (omp_simt_enter): New expansion.
27665 (omp_simt_exit): New insn.
27666 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27668 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27669 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27670 (expand_GOMP_SIMT_EXIT): New.
27671 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27672 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27673 (GOMP_SIMT_EXIT): Ditto.
27674 * target-insns.def (omp_simt_enter): New insn.
27675 (omp_simt_exit): Ditto.
27676 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27678 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27679 (lower_rec_input_clauses): Likewise.
27680 (lower_lastprivate_clauses): Handle SIMT privatization.
27682 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27683 (ompdevlow_adjust_simt_enter): New.
27684 (find_simtpriv_var_op): New.
27685 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27686 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27688 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27689 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27690 (copy_decl_for_dup_finish): Ditto.
27692 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27694 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27697 * config/i386/i386.c (ix86_option_override_internal): Always
27698 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27700 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27702 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27704 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27706 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27707 mark new edge's irreducible flag accordign to it.
27708 (vect_do_peeling): Check loop preheader edge's irreducible flag
27709 and pass it to function slpeel_add_loop_guard.
27711 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27713 PR tree-optimization/80218
27714 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27715 Update block frequencies and counts.
27717 2017-03-28 Richard Biener <rguenther@suse.de>
27719 PR tree-optimization/78644
27720 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27721 of a simplification result we may not use it at all.
27723 2017-03-28 Richard Biener <rguenther@suse.de>
27726 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27727 without arguments, generate default definition of a SSA name.
27729 2017-03-28 Richard Biener <rguenther@suse.de>
27731 PR middle-end/80222
27732 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27733 TYPE_REF_CAN_ALIAS_ALL references.
27734 * fold-const.c (fold_indirect_ref_1): Likewise.
27736 2017-03-28 Martin Liska <mliska@suse.cz>
27739 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27740 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27742 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27743 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27745 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27746 (EXTRA_SPECS): Define.
27747 (SUBTARGET_EXTRA_SPECS): Likewise.
27748 (SUBTARGET_CPP_SPEC): Likewise.
27749 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27750 SUBTARGET_EXTRA_SPECS.
27751 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27753 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27755 * config/arc/simdext.md (vst64_insn): Update pattern.
27756 (vld32wh_insn): Likewise.
27757 (vld32wl_insn): Likewise.
27758 (vld64_insn): Likewise.
27759 (vld32_insn): Likewise.
27761 2017-03-28 Marek Polacek <polacek@redhat.com>
27764 * fold-const.c (fold_comparison): Use protected_set_expr_location
27765 instead of SET_EXPR_LOCATION.
27767 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27769 * tree.c (add_expr): Avoid name lookup warning.
27771 2017-03-27 Jeff Law <law@redhat.com>
27773 PR tree-optimization/80216
27774 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27775 function name. Limit recursion depth.
27776 (record_temporary_equivalences): Corresponding changes.
27778 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27780 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27783 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27786 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27788 * cfgcleanup.c (reg_note_cfa_p): New array.
27789 (insns_have_identical_cfa_notes): New function.
27790 (old_insns_match_p): Don't cross-jump in between /f
27791 and non-/f instructions. If both i1 and i2 are frame related,
27792 verify all CFA notes, their order and content.
27794 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27797 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27798 HImode and SImode with zero extend to DImode to one insn.
27799 (bswap<mode>2_extenddi): Likewise.
27800 (bswapsi2_extenddi): Likewise.
27801 (bswaphi2_extendsi): Likewise.
27802 (bswaphi2): Combine bswap HImode and SImode into one insn.
27803 Separate memory insns from swapping register.
27804 (bswapsi2): Likewise.
27805 (bswap<mode>2): Likewise.
27806 (bswaphi2_internal): Delete, no longer used.
27807 (bswapsi2_internal): Likewise.
27808 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27809 store, and gpr<-gpr swap insns.
27810 (bswap<mode>2_store): Likewise.
27811 (bswaphi2_reg): Register only splitter, combine with the splitter.
27812 (bswaphi2 splitter): Likewise.
27813 (bswapsi2_reg): Likewise.
27814 (bswapsi2 splitter): Likewise.
27815 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27816 the insns into load, store, and register/register insns.
27817 (bswapdi2_ldbrx): Likewise.
27818 (bswapdi2_load): Likewise.
27819 (bswapdi2_store): Likewise.
27820 (bswapdi2_reg): Likewise.
27822 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27824 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27825 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27827 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27830 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27832 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27833 special handling for target option conflicts between dform
27834 options (-mpower9-dform, -mpower9-dform-vector,
27835 -mpower9-dform-scalar) and -mno-direct-move.
27837 2017-03-27 Richard Biener <rguenther@suse.de>
27839 PR tree-optimization/80181
27840 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27842 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27844 * config/arc/predicates.md (move_double_src_operand): Replace the
27845 call to move_double_src_operand with a call to address_operand.
27847 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27849 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27850 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27851 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27853 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27855 * config/arc/predicates.md (long_immediate_loadstore_operand):
27856 Consider scaled addresses cases.
27858 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27860 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27861 restored when in interrupt.
27862 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27863 doesn't have delay slot.
27865 2017-03-27 Richard Biener <rguenther@suse.de>
27868 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27869 inlined thunk clones.
27871 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27874 * asan.c (instrument_derefs): Copy over last operand from
27875 original COMPONENT_REF to the new COMPONENT_REF with
27876 DECL_BIT_FIELD_REPRESENTATIVE.
27877 * ubsan.c (instrument_object_size): Likewise.
27879 2017-03-27 Richard Biener <rguenther@suse.de>
27881 PR tree-optimization/80170
27882 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27883 sure DR/SCEV didnt fold in constants we do not see when looking
27884 at the reference base alignment.
27886 2017-03-27 Richard Biener <rguenther@suse.de>
27888 PR middle-end/80171
27889 * gimple-fold.c (fold_ctor_reference): Properly guard against
27890 NULL return value from canonicalize_constructor_val.
27892 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27895 * config/i386/i386.c (ix86_expand_builtin)
27896 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27897 flags reg setting and flags reg using instructions.
27898 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27899 clobbering instructions to zero extend op2.
27901 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27903 * doc/install.texi (Configuration) <--with-aix-soname>:
27904 Update link to AIX ld.
27906 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27908 PR rtl-optimization/80160
27909 PR rtl-optimization/80159
27910 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27911 reg_alternate_class into account.
27913 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27916 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27917 to consider in curr_insn_transform.
27919 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27921 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27922 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27923 and emit_mode_inner.
27925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27927 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27928 argument to the overloaded builtin variants. Use the new flag to
27929 deprecate certain builtin variants.
27930 * config/s390/s390-builtin-types.def: Add new builtin types.
27931 * config/s390/s390-builtins.h: Support new flags field for
27932 overloaded builtins.
27933 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27934 (s390_macro_to_expand): Enable vector float data type.
27935 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27936 builtins by incrementing the __VEC__ version number.
27937 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27939 (s390_resolve_overloaded_builtin): Emit error messages depending
27940 on the builtin flags.
27941 * config/s390/s390.c (s390_expand_builtin): Support additional
27942 flags argument. Change error message to match the messages
27943 emitted in s390-c.c.
27944 * config/s390/s390.md: New UNSPEC_* constants.
27945 (op_type): Add new instruction types.
27946 * config/s390/vecintrin.h: Add new builtins and test data class
27948 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27949 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27950 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27951 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27952 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27953 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27955 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27956 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27957 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27958 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27960 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27961 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27962 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27963 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27965 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27966 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27967 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27968 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27969 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27970 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27971 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27973 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27974 ("vec_scatter_element<V_HW_4:mode>_DI")
27975 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27976 ("vec_fpint<mode>", "vflls")
27977 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27978 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27979 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27980 ("*vec_cmphe<mode>_cc"): ... these.
27982 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27983 mode constant instead of magic value.
27985 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27987 * config/s390/s390.c (s390_expand_vec_compare): Support other
27988 vector floating point modes than just V2DF.
27989 (s390_expand_vcond): Likewise.
27990 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27991 (s390_cannot_change_mode_class): Prevent mode changes between TF
27992 and V1TF in vector registers.
27993 * config/s390/s390.md (DF, SF): New mode attributes.
27994 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27995 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27996 SFmode support for VRs.
27997 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27999 (VFT, VF_HW): New mode iterators.
28000 (vw, sdx): New mode attributes.
28001 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28002 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28003 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28004 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28005 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28006 also the new vector floating point modes. Renaming to ...
28008 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28009 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28010 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28011 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28012 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28013 ("vec_unordered<mode>"): ... these.
28015 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28016 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28017 ("*vec_extendv2df"): New insn definitions.
28019 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28021 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28022 ("mulditi3_2", "*muldi3_sign"): New patterns.
28023 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28024 rename the pattern definition.
28026 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28028 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28030 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28032 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28034 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28035 instruction if possible.
28036 * config/s390/vector.md (vec_halfnumelts): New mode
28038 ("*vec_vllezlf<mode>"): New pattern.
28040 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28042 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28043 ("popcountv4si2", "popcountv2di2"): Rename to ...
28044 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28045 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28047 ("popcount<mode>2_vxe"): New pattern.
28049 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28051 * common/config/s390/s390-common.c (processor_flags_table): Add
28053 * config.gcc: Add arch12.
28054 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28055 Default to arch12 for unknown CPU model numbers.
28056 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28057 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28058 PROCESSOR_max sanity check.
28059 * config/s390/s390-opts.h (enum processor_type): Add
28061 * config/s390/s390.c (processor_table): Add arch12.
28062 (s390_expand_builtin): Add check for B_VXE flag.
28063 (s390_issue_rate): Add PROCESSOR_ARCH12.
28064 (s390_get_sched_attrmask): Likewise.
28065 (s390_get_unit_mask): Likewise.
28066 (s390_sched_score): Enable z13 scheduling for arch12.
28067 (s390_sched_reorder): Likewise.
28068 (s390_sched_variable_issue): Likewise.
28069 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28071 (s390_tune_attr): Use z13 scheduling also for arch12.
28072 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28073 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28074 (TARGET_VXE_P): New macros.
28075 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28076 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28077 * config/s390/s390.opt: Add arch12 as processor_type.
28079 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28081 * config/s390/s390.md
28082 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28083 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28084 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28086 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28087 Rename expanders to ...
28089 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28090 ("fixuns_truncdddi2_emu"): ... these.
28092 ("fixuns_trunc<mode>si2_emu"): New expander.
28094 ("*fixuns_truncdfdi2_z13"): Rename to ...
28095 ("*fixuns_truncdfdi2_vx"): ... this.
28097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28099 * config/s390/2964.md: Remove the single element vector compare
28100 instructions which are no longer used.
28101 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28103 (s390_canonicalize_comparison): Remove handling of DFmode
28105 (s390_expand_vec_compare_scalar): Remove function.
28106 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28107 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28109 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28111 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28113 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28115 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28116 will anyway by matched by mov<mode>_64dfp.
28118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28120 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28121 vlef/vstef. Add missing operand to vleif.
28123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28125 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28126 pair for all vector types with 64 bit elements.
28127 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28128 * config/s390/vector.md (V_HW_64): ... here.
28129 (V_128_NOSINGLE): New mode iterator.
28130 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28131 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28132 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28133 ("*vec_load_pairv2di"): Change to ...
28134 ("*vec_load_pair<mode>"): ... this one.
28136 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28138 * config/s390/constraints.md: Add comments.
28139 (jKK): Reject element sizes > 8 bytes.
28140 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28142 * config/s390/s390.md: Add the s_operand checks formerly in
28143 s390_split_ok_p to various splitters where they are still
28145 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28146 for 128 bit vectors. Plus two splitters.
28148 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28150 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28156 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28157 error if the boundary argument is not constant.
28159 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28161 PR rtl-optimization/80112
28162 * loop-doloop.c (doloop_condition_get): Don't check condition
28163 if cmp isn't SET with IF_THEN_ELSE src.
28165 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28167 PR tree-optimization/80158
28168 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28169 replacing a candidate statement, also replace it for the
28170 candidate's alternate interpretation.
28171 (replace_rhs_if_not_dup): Likewise.
28172 (replace_one_candidate): Likewise.
28174 2017-03-24 Richard Biener <rguenther@suse.de>
28176 PR tree-optimization/80167
28177 * graphite-isl-ast-to-gimple.c
28178 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28180 (translate_isl_ast_to_gimple::get_rename): Likewise.
28182 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28184 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28185 handling of certain combinations of target options, including the
28186 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28187 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28189 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28192 * config/arm/arm.md (*load_multiple): Add reload_completed to
28193 matching condition.
28195 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28196 Richard Biener <rguenth@suse.de>
28198 PR tree-optimization/79908
28199 PR tree-optimization/80136
28200 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28201 been cast away, gimplify_and_add suffices.
28203 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28205 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28207 2017-03-23 Richard Biener <rguenther@suse.de>
28209 PR tree-optimization/80032
28210 * gimplify.c (gimple_push_cleanup): Forced unconditional
28211 cleanups still have to go to the conditional_cleanups
28214 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28216 PR tree-optimization/80072
28217 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28219 (next_operand_entry_id): Change type to unsigned int.
28220 (sort_by_operand_rank): Make sure to return the right return value
28221 even if unsigned fields are bigger than INT_MAX.
28222 (struct oecount): Change cnt and id type to unsigned int.
28223 (oecount_hasher::equal): Formatting fix.
28224 (oecount_cmp): Make sure to return the right return value
28225 even if unsigned fields are bigger than INT_MAX.
28226 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28229 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28230 TREE_READONLY on result if writing it more than once.
28233 * doc/invoke.texi (-fsanitize=thread): Document that with
28234 -fnon-call-exceptions atomics are not able to throw
28238 * tsan.c: Include tree-eh.h.
28239 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28240 maybe_clean_or_replace_eh_stmt where needed.
28241 (instrument_memory_accesses): Add cfg_changed argument.
28242 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28243 if it returned true.
28244 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28246 PR rtl-optimization/63191
28247 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28248 wrapper function, moved the whole old content into ...
28249 (ix86_delegitimize_address_1): ... this. New inline function.
28250 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28251 true as last argument instead of ix86_delegitimize_address.
28253 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28255 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28256 cortexa57_branch_cost.
28258 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28260 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28262 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28265 * doc/md.texi (Constraints): Document wA constraint.
28266 * config/rs6000/constraints.md (wA): New.
28267 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28268 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28269 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28270 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28272 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28275 * gimplify.c (is_oacc_declared): New function.
28276 (oacc_default_clause): Use it to set default flags for acc declared
28277 variables inside parallel regions.
28278 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28279 declared variables.
28280 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28281 declare attribute to any decl as necessary.
28283 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28286 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28287 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28288 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28289 (arm_arch_lpae): This.
28290 * config/arm/arm.c (arm_arch7ve): Rename into ...
28291 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28292 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28295 2017-03-22 Martin Liska <mliska@suse.cz>
28298 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28299 error message instead of an ICE.
28301 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28303 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28305 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28307 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28309 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28312 2017-03-21 Martin Sebor <msebor@redhat.com>
28314 * doc/extend.texi: Use "cannot" instead of "can't."
28315 * doc/hostconfig.texi: Same.
28316 * doc/install.texi: Same.
28317 * doc/invoke.texi: Same.
28318 * doc/loop.texi: Same.
28319 * doc/md.texi: Same.
28320 * doc/objc.texi: Same.
28321 * doc/rtl.texi: Same.
28322 * doc/tm.texi: Same.
28323 * doc/tm.texi.in: Same.
28324 * doc/trouble.texi: Same.
28326 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28329 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28330 (collect_checksum_attributes): Set it.
28331 (die_checksum_ordered): Use it.
28333 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28335 PR tree-optimization/79908
28336 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28337 change: For a VA_ARG whose LHS has been cast away, use
28338 force_gimple_operand to construct the side effects.
28340 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28342 PR translation/80001
28343 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28344 more amenable to translation.
28345 (oacc_loop_auto_partitions): Likewise.
28347 2017-03-21 Marek Polacek <polacek@redhat.com>
28348 Martin Sebor <msebor@redhat.com>
28350 PR tree-optimization/80109
28351 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28352 on INTEGRAL_TYPE_P.
28354 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28355 Segher Boessenkool <segher@kernel.crashing.org>
28358 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28359 check reg_used_between_p between insn and one of succ or succ2
28360 depending on if succ is artificial insn not inserted into insn
28363 2017-03-21 Martin Liska <mliska@suse.cz>
28365 PR gcov-profile/80081
28366 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28367 * doc/gcc.texi: Include gcov-dump stuff.
28368 * doc/gcov-dump.texi: New file.
28370 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28372 PR rtl-optimization/79150
28373 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28374 conditional jump, if the jump is the last insn of the loop.
28376 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28377 Richard Biener <rguenth@suse.de>
28379 PR tree-optimization/79908
28380 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28381 been cast away, use force_gimple_operand to construct the side
28384 2017-03-21 Martin Liska <mliska@suse.cz>
28386 PR libfortran/79956
28387 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28390 2017-03-21 Brad Spengler <spender@grsecurity.net>
28393 * plugin.c (htab_hash_plugin): New function.
28394 (add_new_plugin): Use it and adjust.
28395 (parse_plugin_arg_opt): Adjust.
28396 (init_one_plugin): Likewise.
28398 2017-03-21 Richard Biener <rguenther@suse.de>
28400 PR tree-optimization/80032
28401 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28402 if set force the cleanup to happen unconditionally.
28403 (gimplify_target_expr): Push inserted clobbers with force_uncond
28404 to avoid them being removed by control-dependent DCE.
28406 2017-03-21 Richard Biener <rguenther@suse.de>
28408 PR tree-optimization/80122
28409 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28410 va_arg_pack_len when the inlined call stmt requires pack
28412 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28414 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28417 * tsan.c (instrument_builtin_call): If the memory model argument
28418 is not a constant, assume it is valid.
28421 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28424 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28426 PR rtl-optimization/79910
28427 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28428 if its dest is used by an insn before I2 (other than the combined
28429 insns themselves, which are properly handled already).
28431 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28434 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28436 * combine.c (record_used_regs): New static function.
28437 (try_combine): Handle situations where there is an additional
28438 instruction between I2 and I3 which needs to have a LOG_LINK
28442 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28444 * combine.c (try_combine): Delete redundant i1 test. Call
28445 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28447 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28450 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28451 alternatives 13/14.
28453 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28455 PR tree-optimization/80054
28456 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28457 the optimization if a PHI or any of its arguments is not dominated
28458 by the candidate's basis. Use gphi* rather than gimple* as
28460 (replace_profitable_candidates): Clean up a gimple* variable that
28461 should be a gphi* variable.
28463 2017-03-20 Martin Sebor <msebor@redhat.com>
28466 * doc/extend.texi (attribute constructor): Document present limitation.
28468 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28471 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28472 __POWER9_VECTOR__ #ifdef control, change template definition to
28473 use Power9-specific built-in function.
28474 (vec_any_eq): Likewise.
28475 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28476 to control outcomes from this test.
28477 (vector_ae_<mode>p): For VEC_F modes, likewise.
28479 2017-03-20 Ian Lance Taylor <iant@google.com>
28481 * config/i386/i386.c (ix86_function_regparm): Save an extra
28482 register for -fsplit-stack with DECL_STATIC_CHAIN.
28484 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28487 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28488 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28490 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28492 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28494 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28497 2017-03-20 Marek Polacek <polacek@redhat.com>
28500 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28502 2017-03-20 Richard Biener <rguenther@suse.de>
28504 PR tree-optimization/80113
28505 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28506 allocate extra SSA name for PHI def.
28507 (add_close_phis_to_outer_loops): Likewise.
28508 (add_close_phis_to_merge_points): Likewise.
28509 (copy_loop_close_phi_args): Likewise.
28510 (copy_cond_phi_nodes): Likewise.
28512 2017-03-20 Martin Liska <mliska@suse.cz>
28514 PR middle-end/79753
28515 * tree-chkp.c (chkp_build_returned_bound): Do not build
28516 returned bounds for a LHS that's not a BOUNDED_P type.
28518 2017-03-20 Martin Liska <mliska@suse.cz>
28522 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28523 COMPLEX_CST and VECTOR_CST.
28525 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28528 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28529 target operand. A new splitter adds the clobber statement in case
28530 the target operand is dead anyway.
28532 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28534 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28535 to age-old versions of binutils and glibc.
28537 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28539 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28541 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28543 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28545 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28547 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28548 requirement for binutils 2.13.
28550 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28552 * combine.c (try_combine): Delete redundant i1 test. Call
28553 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28555 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28557 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28558 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28560 <riscv64-*-elf>: Re-arrange section
28561 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28562 <riscv32-*-linux>: Likewise.
28563 <riscv64-*-elf>: Likewise
28564 <riscv64-*-linux>: Likewise.
28566 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28569 * aarch64.opt(verbose-cost-dump): Fix typo.
28571 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28574 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28575 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28577 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28579 * reload.c (find_reloads): When reloading a nonoffsettable address,
28580 use RELOAD_OTHER for it and its address reloads.
28582 PR rtl-optimization/79910
28583 * combine.c (record_used_regs): New static function.
28584 (try_combine): Handle situations where there is an additional
28585 instruction between I2 and I3 which needs to have a LOG_LINK
28588 2017-03-17 Jeff Law <law@redhat.com>
28590 PR tree-optimization/71437
28591 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28592 conditional in the hash table first.
28593 (vrp_dom_walker::before_dom_children): Extract condition from
28594 ASSERT_EXPR. Record condition, its inverion and any implied
28595 conditions as well.
28597 2017-03-17 Marek Polacek <polacek@redhat.com>
28598 Markus Trippelsdorf <markus@trippelsdorf.de>
28600 PR tree-optimization/80079
28601 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28604 2017-03-17 Richard Biener <rguenther@suse.de>
28606 PR middle-end/80075
28607 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28608 Properly verify the LHS before the RHS possibly claims to be
28610 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28613 2017-03-17 Martin Jambor <mjambor@suse.cz>
28615 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28616 (List of -O2 options): Likewise.
28617 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28620 2017-03-17 Tom de Vries <tom@codesourcery.com>
28622 * gcov-dump.c (print_usage): Print bug_report_url.
28624 2017-03-17 Richard Biener <rguenther@suse.de>
28626 PR middle-end/80050
28627 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28628 (parser::peek): Likewise.
28630 2017-03-17 Richard Biener <rguenther@suse.de>
28632 PR tree-optimization/80048
28633 * sese.c (free_sese_info): Properly release rename_map and
28634 copied_bb_map elements.
28636 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28638 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28639 Add linked-list forward and backlinks. Insert on
28640 construction, remove on destruction.
28641 (class pass_store_merging): Add m_stores_head field.
28642 (pass_store_merging::terminate_and_process_all_chains):
28643 Iterate over m_stores_head list.
28644 (pass_store_merging::terminate_all_aliasing_chains):
28646 (pass_store_merging::execute): Check for debug stmts first.
28647 Push new chains onto the m_stores_head stack.
28649 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28652 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28653 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28654 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28656 2017-03-16 Jeff Law <law@redhat.com>
28658 PR tree-optimization/71437
28659 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28660 member function. Implementation moved into after_dom_children
28661 member function and into the threader's thread_outgoing_edges
28663 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28664 some code into new thread_outgoing_edges.
28665 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28666 definition. Simplify marker handling (do it here). Assume we always
28667 have the available expression and the const/copies tables.
28668 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28670 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28671 * tree-vrp.c (equiv_stack): No longer file scoped.
28672 (vrp_dom_walker): New class.
28673 (vrp_dom_walker::before_dom_children): New member function.
28674 (vrp_dom_walker::after_dom_children): Likewise.
28675 (identify_jump_threads): Setup domwalker. Use it rather than
28676 walking edges in a random order by hand. Simplify setup/finalization.
28677 (finalize_jump_threads): Remove.
28678 (vrp_finalize): Do not call identify_jump_threads here.
28679 (execute_vrp): Do it here instead and call thread_through_all_blocks
28682 PR tree-optimization/71437
28683 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28685 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28687 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28688 (dom_opt_dom_walker::thread_across_edge): Remove
28689 handle_dominating_asserts argument. All callers changed.
28690 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28691 changes. Remove calls to lhs_of_dominating_assert. Other
28692 uses of handle_dominating_asserts turn into unconditional code
28693 (simplify_control_stmt_condition_1): Likewise.
28694 (simplify_control_stmt_condition): Likewise.
28695 (thread_through_normal_block, thread_across_edge): Likewise.
28696 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28697 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28698 object if it is not an SSA_NAME.
28699 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28700 before calling into the VRP specific simplifiers.
28701 (identify_jump_threads): Remove handle_dominating_asserts
28704 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28707 * tree-diagnostic.c (default_tree_printer): No longer static.
28708 * tree-diagnostic.h (default_tree_printer): New prototype.
28710 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28712 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28713 Change ins into fmov.
28715 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28717 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28718 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28719 Use h_con constraint for operand 1.
28720 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28721 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28723 2017-03-15 Jeff Law <law@redhat.com>
28725 PR tree-optimization/71437
28726 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28727 (record_temporary_equivalences): Use it.
28729 PR tree-optimization/71437
28730 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28731 tree-ssa-scopedtables.
28732 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28733 (record_conditions, record_cond, vuse_eq): Likewise.
28734 (record_edge_info): Adjust to API tweak of record_conditions.
28735 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28736 (record_temporary_equivalences, optimize_stmt): Likewise.
28737 (eliminate_redundant_computations): Likewise.
28738 (record_equivalences_from_stmt): Likewise.
28739 * tree-ssa-scopedtables.c: Include options.h and params.h.
28740 (vuse_eq): New function, moved from tree-ssa-dom.c
28741 (build_and_record_new_cond): Likewise.
28742 (record_conditions): Likewise. Accept vector of conditions rather
28743 than edge_equivalence structure for first argument.
28744 for the first argument.
28745 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28746 from tree-ssa-dom.c.
28747 (avail_exprs_stack::record_cond): Likewise.
28748 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28749 from tree-ssa-dom.c.
28750 (avail_exprs_stack): Add new member functions lookup_avail_expr
28752 (record_conditions): Declare.
28754 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28757 * lra-constraints.c (process_alt_operands): Increase reject for
28758 reloading an input/output operand.
28760 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28763 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28764 insns to convert from signed/unsigned char/short to IEEE 128-bit
28766 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28768 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28771 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28772 subreg of inner mode for values already in registers.
28774 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28776 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28777 iteration reg is used after the loop.
28779 2017-03-14 Martin Sebor <msebor@redhat.com>
28781 PR tree-optimization/79800
28782 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28783 precision in negative-positive range.
28784 (format_floating): Call non-const overload with adjusted precision.
28786 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28789 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28792 2017-03-14 Martin Sebor <msebor@redhat.com>
28794 PR middle-end/80020
28795 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28796 * builtins.def (aligned_alloc): Use it.
28799 * Makefile.in (GTFILES): Add calls.c.
28800 * calls.c: Include "gt-calls.h".
28802 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28804 PR rtl-optimization/79728
28805 * regs.h (struct target_regs): New field
28806 x_contains_allocatable_regs_of_mode.
28807 (contains_allocatable_regs_of_mode): New macro.
28808 * reginfo.c (init_reg_sets_1): Initialize it, and change
28809 contains_reg_of_mode so it includes global regs as well.
28810 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28811 rather than contains_regs_of_mode.
28813 2017-03-14 Martin Liska <mliska@suse.cz>
28815 * doc/invoke.texi: Document options that can't be combined with
28816 -fcheck-pointer-bounds.
28818 2017-03-14 Martin Liska <mliska@suse.cz>
28820 PR middle-end/79831
28821 * doc/invoke.texi (-Wchkp): Document the option.
28823 2017-03-14 Martin Liska <mliska@suse.cz>
28825 * Makefile.in: Install gcov-dump.
28827 2017-03-14 Martin Liska <mliska@suse.cz>
28829 * multiple_target.c (expand_target_clones): Bail out for
28830 an invalid attribute.
28832 2017-03-14 Richard Biener <rguenther@suse.de>
28834 * alias.c (struct alias_set_entry): Pack properly.
28835 * cfgloop.h (struct loop): Likewise.
28836 * cse.c (struct set): Likewise.
28837 * ipa-utils.c (struct searchc_env): Likewise.
28838 * loop-invariant.c (struct invariant): Likewise.
28839 * lra-remat.c (struct cand): Likewise.
28840 * recog.c (struct change_t): Likewise.
28841 * rtl.h (struct address_info): Likewise.
28842 * symbol-summary.h (function_summary): Likewise.
28843 * tree-loop-distribution.c (struct partition): Likewise.
28844 * tree-object-size.c (struct object_size_info): Likewise.
28845 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28846 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28847 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28848 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28849 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28850 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28851 (struct _stmt_vec_info): Likewise.
28853 2017-03-14 Martin Liska <mliska@suse.cz>
28856 * multiple_target.c (create_dispatcher_calls): Check that
28857 a target can create a function dispatcher.
28859 2017-03-14 Martin Liska <mliska@suse.cz>
28862 * multiple_target.c (expand_target_clones): Drop local.local
28863 flag for default implementation.
28865 2017-03-14 Richard Biener <rguenther@suse.de>
28867 PR tree-optimization/80030
28868 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28870 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28872 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28873 gcc_fallthrough() instead of __attribute__((fallthrough));
28875 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28877 * doc/gcc.texi: Remove "up" link to (DIR).
28878 * doc/gccint.texi: Ditto.
28880 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28882 * doc/install.texi (Specific) <avr>: Remove reference to
28885 2017-03-13 Jeff Law <law@redhat.com>
28887 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28888 attribute rather than comments.
28890 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28891 match_scratch operand is highest.
28893 2017-03-13 Martin Liska <mliska@suse.cz>
28895 PR middle-end/78339
28896 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28897 is a CHKP clone, use original declaration.
28899 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28901 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28902 (arc_conditional_register_usage): Use a different allocation order
28903 when optimizing for size.
28904 * common/config/arc/arc-common.c (arc_option_optimization_table):
28905 Section anchors default on when optimizing for size.
28907 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28909 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28911 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28913 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28914 * config/arc/arc.md (cpu_facility): Add cd variant.
28915 (*movqi_insn): Add code density variant.
28916 (*movhi_insn): Likewise.
28917 (*movqi_insn): Likewise.
28918 (*addsi3_mixed): Likewise.
28919 (subsi3_insn): Likewise.
28921 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28923 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28925 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28927 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28928 expressions with MINUS and UNARY ops.
28930 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28933 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28935 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28936 between vec_select and vector argument.
28937 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28938 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28939 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28940 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28941 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28942 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28944 2017-03-13 Richard Biener <rguenther@suse.de>
28947 * params.def (vect-max-peeling-for-alignment): Fix typo.
28949 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28951 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28952 issue that only occurred with binutils below 2.18.
28954 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28956 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28957 refer to binutils 2.11/2.12 minimum.
28959 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28961 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28962 ftp.kernel.org and simplify binutils requirement.
28964 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28966 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28968 (Optimize Options): Ditto. Also remove redundancy.
28970 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28972 PR translation/79848
28973 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28975 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28976 to G_ to avoid double translation.
28978 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28980 PR translation/79923
28981 * auto-profile.c (get_combined_location): Convert leading
28982 character of diagnostics to lower case and remove trailing period.
28983 (read_profile): Likewise for various diagnostics.
28984 * config/arm/arm.c (arm_option_override): Remove trailing period
28985 from various diagnostics.
28986 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28987 (msp430_expand_delay_cycles): Likewise.
28989 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28992 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28993 full command-line argument, rather than just "str".
28994 (aarch64_validate_march): Likewise.
28995 (aarch64_validate_mtune): Likewise.
28997 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28999 PR rtl-optimization/78911
29000 * lra-assigns.c (must_not_spill_p): New function.
29001 (spill_for): Use it.
29003 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29005 PR tree-optimization/79981
29006 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29007 ATOMIC_COMPARE_EXCHANGE ifn result.
29008 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29009 IFN_ATOMIC_COMPARE_EXCHANGE.
29011 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29014 * opts.c (parse_sanitizer_options): Add missing question mark to
29015 "did you mean" message.
29017 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29019 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29021 (VMULEUH_UNS): Likewise.
29022 (VMULOUB_UNS): Likewise.
29023 (VMULOUH_UNS): Likewise.
29024 * config/rs6000/rs6000.c (builtin_function_type): Remove
29025 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29027 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29030 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29031 x with result of extra_parsing_for_operand_code_0.
29032 (function_reader::extra_parsing_for_operand_code_0): Convert
29033 return type from void to rtx, returning x. When reading
29034 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29035 larger size containing struct block_symbol.
29037 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29040 -mfloat128-hardware without -m64.
29042 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29045 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29046 entries to the case statement that marks unsigned arguments to
29047 overloaded functions.
29049 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29052 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29054 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29057 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29058 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29060 2017-03-10 Martin Liska <mliska@suse.cz>
29064 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29065 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29068 2017-03-10 Olivier Hainque <hainque@adacore.com>
29070 * tree-switch-conversion (array_value_type): Start by resetting
29071 candidate type to it's main variant.
29073 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29075 PR rtl-optimization/79909
29076 * combine.c (try_combine): Use simplify_replace_rtx on individual
29077 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29078 of the whole CALL_INSN_FUNCTION_USAGE.
29080 PR tree-optimization/79972
29081 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29082 get_range_info on SSA_NAMEs. Formatting fixes.
29084 2017-03-10 Richard Biener <rguenther@suse.de>
29085 Jakub Jelinek <jakub@redhat.com>
29087 PR tree-optimization/77975
29088 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29089 edge to be constant.
29090 (get_val_for): For constant x return it. Formatting fix.
29091 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29092 would use the same bases as the current one.
29094 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29096 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29097 instead of vec_select for V1TImode.
29098 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29100 (VSX_LE_128): Add V1TI to this mode iterator.
29101 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29102 (*vsx_le_perm_store_<mode>): Likewise.
29103 (pre-reload splitter for VSX stores): Likewise.
29104 (post-reload splitter for VSX stores): Likewise.
29105 (*vsx_xxpermdi2_le_<mode>): Likewise.
29106 (*vsx_lxvd2x2_le_<mode>): Likewise.
29107 (*vsx_stxvd2x2_le_<mode>): Likewise.
29109 2017-03-09 Michael Eager <eager@eagercon.com>
29111 Correct failures with --enable-checking=yes,rtl.
29113 * config/microblaze/microblaze.c (microblaze_expand_shift):
29114 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29115 test for const0_rtx.
29116 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29117 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29119 2017-03-09 Richard Biener <rguenther@suse.de>
29121 PR tree-optimization/79977
29122 * graphite-scop-detection.c (scop_detection::merge_sese):
29123 Handle the case of extra exits to blocks dominating the entry.
29125 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29127 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29130 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29132 PR rtl-optimization/79949
29133 * lra-constraints.c (process_alt_operands): Check memory when
29134 trying to predict a cycle. Print about the overall increase.
29136 2017-03-09 Richard Biener <rguenther@suse.de>
29138 PR middle-end/79971
29139 * gimple-expr.c (useless_type_conversion_p): Preserve
29140 TYPE_SATURATING for fixed-point types.
29142 2017-03-09 Richard Biener <rguenther@suse.de>
29145 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29146 alignment of BLKmode params.
29148 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29151 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29152 (VALL_NO_V2Q): Likewise.
29154 * config/aarch64/aarch64-simd.md
29155 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29157 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29158 VALL_NO_V2Q mode iterator.
29159 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29161 2017-03-09 Martin Liska <mliska@suse.cz>
29163 PR tree-optimization/79631
29164 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29165 tree_int_cst_sign_bit just for INTEGER constants.
29167 2017-03-09 Martin Liska <mliska@suse.cz>
29171 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29174 2017-03-09 Marek Polacek <polacek@redhat.com>
29177 * tree.c (inchash::add_expr): Handle TREE_VEC.
29179 2017-03-09 Martin Liska <mliska@suse.cz>
29182 (chkp_narrow_size_and_offset): New function.
29183 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29184 (void chkp_parse_bit_field_ref): New function.
29185 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29186 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29188 2017-03-09 Martin Liska <mliska@suse.cz>
29191 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29192 (chkp_find_bounds_1): Remove gcc_unreachable.
29194 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29197 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29198 BUILT_IN_SYNC*, determine the access type from the size suffix and
29199 always build a MEM_REF with that type. Handle forgotten
29200 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29203 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29204 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29205 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29206 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29207 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29208 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29209 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29210 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29211 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29212 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29213 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29214 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29215 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29216 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29217 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29218 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29219 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29220 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29221 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29222 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29223 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29224 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29225 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29226 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29227 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29228 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29229 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29230 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29231 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29232 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29233 definitions outside of __OPTIMIZE__ guarded section.
29236 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29237 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29238 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29239 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29242 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29244 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29245 ("vfenez<mode>"): Add missing constraints.
29247 2017-03-08 Martin Sebor <msebor@redhat.com>
29250 * config/nds32/nds32.c (nds32_option_override):
29251 Fix misspelled diagnostic.
29253 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29256 * gimplify.c (gimplify_omp_for): Replace index var in outer
29257 taskloop statement with an artificial variable and add
29258 OMP_CLAUSE_PRIVATE clause for it.
29260 2017-03-08 Richard Biener <rguenther@suse.de>
29262 PR tree-optimization/79955
29263 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29264 for accesses that are completely outside of the variable.
29266 2017-03-08 Andrew Haley <aph@redhat.com>
29268 PR tree-optimization/79943
29269 * tree-ssa-loop-split.c (compute_new_first_bound): When
29270 calculating the new upper bound, (END-BEG) should be added, not
29273 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29275 * config/avr/avr.md (setmemhi): Make sure match_dup
29276 operand number comes before match_scratch.
29278 2017-03-08 Richard Biener <rguenther@suse.de>
29280 PR tree-optimization/79920
29281 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29282 with ncopies == 1 to ...
29283 (vect_transform_slp_perm_load): ... here. Properly compute
29284 all element loads by iterating VF times over the group. Do
29285 not handle ncopies (computed in a broken way) in
29286 vect_create_mask_and_perm.
29288 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29291 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29292 is a uniform vector, use uniform_vector_p return value instead of
29293 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29295 2017-03-07 Marek Polacek <polacek@redhat.com>
29297 PR middle-end/79809
29298 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29299 (alloca_call_type): Likewise.
29301 2017-03-07 Martin Liska <mliska@suse.cz>
29303 * gcov.c (process_args): Put comment to correct location.
29305 2017-03-07 Martin Liska <mliska@suse.cz>
29307 PR middle-end/68270
29308 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29309 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29310 (chkp_narrow_bounds_for_field): Likewise.
29311 (chkp_parse_array_and_component_ref): Pass one more argument to
29314 2017-03-07 Richard Biener <rguenther@suse.de>
29316 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29319 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29321 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29322 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29324 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29327 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29328 to end of description.
29329 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29331 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29333 PR rtl-optimization/79901
29334 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29336 (*avx512f_<code><mode>3<mask_name>): ... this.
29337 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29338 iterator instead of VI8_AVX2_AVX512BW.
29340 PR rtl-optimization/79901
29341 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29342 min/max expander, expand it using expand_vec_cond_expr.
29345 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29348 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29351 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29352 to void * for PCH reasons.
29353 * dwarf2out.c (output_loc_operands, output_die): Cast
29354 v.val_vec.array to unsigned char *.
29356 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29359 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29362 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29364 PR rtl-optimization/79571
29365 * lra-constraints.c (process_alt_operands): Calculate static
29366 reject and subtract it from overall when only addresses will be
29369 2017-03-06 Julia Koval <julia.koval@intel.com>
29372 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29373 incoming stack boundary to 128 for 64-bit targets.
29375 2017-03-06 Richard Biener <rguenther@suse.de>
29377 PR tree-optimization/79894
29378 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29379 to NULL after folding it.
29381 2017-03-06 Richard Biener <rguenther@suse.de>
29383 PR tree-optimization/79824
29384 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29385 check disabling peeling for gaps.
29387 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29389 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29390 attributes): Document gettimeofday.
29392 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29394 * config/s390/s390.c (s390_option_override_internal): Set
29395 PARAM_MIN_VECT_LOOP_BOUND
29397 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29399 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29400 * config/s390/s390.md: Likewise.
29402 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29405 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29406 (<avx2_avx512>_perm<mode>): Rename to ...
29407 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29409 (<avx512>_perm<mode>_mask): Rename to ...
29410 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29412 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29413 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29414 instead of VI8F_256_512.
29415 (avx512f_perm<mode>): New define_expand.
29416 (avx512f_perm<mode>_mask): Likewise.
29417 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29418 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29420 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29422 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29423 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29425 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29427 2017-03-06 Martin Liska <mliska@suse.cz>
29430 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29431 when having a SSA NAME w/o VAR_DECL assigned to it.
29433 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29435 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29436 msa_dpsub_<su>_d): Fix MODE for vec_select.
29438 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29440 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29442 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29444 2017-03-06 Richard Biener <rguenther@suse.de>
29446 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29447 * plugin.c (register_plugin_info): Likewise.
29448 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29450 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29452 * config/i386/sse.md (sse_storehps, sse_storelps,
29453 avx_<castmode><avxsizesuffix>_<castmode>,
29454 avx512f_<castmode><avxsizesuffix>_<castmode>,
29455 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29456 in condition that at least one operand is not a MEM.
29458 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29460 PR middle-end/79805
29461 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29462 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29464 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29465 gimple_call_nothrow_p flag based on whether original builtin can throw.
29466 If it can, emit following stmts on the fallthrough edge.
29467 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29468 don't create new bb if inserting just debug stmts on the edge, try to
29469 insert them on the fallthru bb or just reset debug stmts.
29471 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29474 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29475 restore recog_data (including the operand rtxes inside it) around
29476 the call to get_insn_template.
29478 2017-03-03 Martin Sebor <msebor@redhat.com>
29480 PR tree-optimization/79699
29481 * context.c (context::~context): Free MPFR caches to avoid
29482 a memory leak on program exit.
29484 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29486 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29487 Use wide_int::ulow () instead of .elt (0).
29489 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29491 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29492 (*pushxf): Limit oF constraint to 32bit targets and add oC
29493 constraint for 64bit targets.
29494 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29495 (*pushdf): Change rmF constraint to rmC.
29497 2017-03-03 Martin Liska <mliska@suse.cz>
29499 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29500 Remove unused variable.
29502 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29505 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29506 is a memory operand, increase num_memory.
29507 (ix86_expand_args_builtin): Likewise.
29509 2017-03-03 Jan Hubicka <jh@suse.cz>
29512 * ipa-devirt.c (maybe_record_node): Properly handle
29513 __cxa_pure_virtual visibility.
29515 2017-03-03 Martin Liska <mliska@suse.cz>
29517 PR tree-optimization/79803
29518 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29520 (pass_loop_prefetch::execute): Disabled optimization if an
29521 assumption about L1 cache size is not met.
29523 2017-03-03 Martin Liska <mliska@suse.cz>
29525 PR rtl-optimization/79574
29526 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29527 (hash_scan_set): Likewise.
29528 (dump_hash_table): Likewise.
29529 (hoist_code): Likewise.
29531 2017-03-03 Richard Biener <rguenther@suse.de>
29533 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29534 in place of uhigh ().
29535 (fixed_convert_from_real): Likewise.
29537 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29540 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29542 2017-03-03 Richard Biener <rguenther@suse.de>
29544 PR middle-end/79818
29545 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29546 TYPE_OVERFLOW_UNDEFINED check.
29548 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29550 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29552 (vector_ae_<mode>_p): Likewise.
29553 (vector_nez_<mode>_p): Likewise.
29554 (vector_ne_v2di_p): Likewise.
29555 (vector_ae_v2di_p): Likewise.
29556 (vector_ne_<mode>_p): Likewise.
29557 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29559 (vsx_tsqrt<mode>2_fe): Likewise.
29561 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29564 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29566 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29568 PR rtl-optimization/79780
29569 * cprop.c (one_cprop_pass): When second and further conditional trap
29570 in a single basic block is turned into an unconditional trap, turn it
29571 into a deleted note to avoid RTL verification failures.
29573 2017-03-02 Richard Biener <rguenther@suse.de>
29575 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29577 2017-03-02 Richard Biener <rguenther@suse.de>
29579 PR tree-optimization/79345
29581 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29582 param and abort the walk, returning -1 if it is hit.
29583 (walk_aliased_vdefs): Take a limit param and pass it on.
29584 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29585 defaulting to 0 and return a signed int.
29586 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29587 (check_defs): New helper.
29588 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29589 about uninitialized memory.
29590 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29591 bogus uninitialized warning.
29592 (fixed_convert_from_real): Likewise.
29594 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29596 PR tree-optimization/66768
29597 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29598 iv_use if base object can't be determined.
29600 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29602 PR tree-optimization/79345
29603 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29604 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29605 (get_pattern_stats): Initialize it.
29606 * genemit.c (gen_expand): Verify match_scratch numbers come after
29607 match_operand/match_dup numbers.
29608 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29609 match_scratch numbers.
29610 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29612 * config/s390/s390.md (trunctdsd2): Likewise.
29614 2017-03-02 Richard Biener <rguenther@suse.de>
29616 * wide-int.h (wide_int_storage::operator=): Implement in terms
29619 2017-03-02 Richard Biener <rguenther@suse.de>
29621 PR tree-optimization/79777
29622 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29623 the to insert expression to sth existing.
29625 2017-03-01 Martin Sebor <msebor@redhat.com>
29627 PR middle-end/79692
29628 * gimple-ssa-sprintf.c
29629 (directive::known_width_and_precision): New function.
29630 (format_integer): Use it.
29631 (get_mpfr_format_length): Consider the full range of precision
29632 when computing %g output with the # flag. Set the likely byte
29633 count to 3 rather than 1 when precision is indeterminate.
29634 (format_floating): Correct the lower bound of precision.
29636 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29638 * doc/invoke.texi: Document default code model for 64-bit Linux.
29640 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29643 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29644 udiv rather than div since input pattern is unsigned.
29646 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29648 * config/i386/i386.c (print_reg): Warn for values of
29649 unsupported size in integer register.
29651 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29654 * config/rs6000/predicates.md (current_file_function_operand): Do
29655 not allow self calls to be local if the function is replaceable.
29657 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29660 * config/rs6000/altivec.h (vec_ctz and others): Change the
29661 preprocessor macro that controls conditional compilation from
29662 _ARCH_PWR9 to __POWER9_VECTOR__.
29663 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29664 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29665 control (instead of _ARCH_PWR9 control) so that template
29666 definition uses power9-specific function.
29667 (vec_any_eq): Likewise.
29668 (vec_all_ne): Change macro definition to use a power9-specific
29669 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29670 _ARCH_PWR9 control).
29671 (vec_any_eq) Likewise.
29672 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29673 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29674 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29675 support for xvcmpnedp instruction.
29676 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29677 macro expansion so that Power9 implementation of vec_all_ne does
29678 not use the AltiVec predicate framework.
29679 (VCMPNEH_P): Likewise.
29680 (VCMPNEW_P): Likewise.
29681 (VCMPNED_P): Likewise.
29682 (VCMPNEFP_P): Likewise.
29683 (VCMPNEDP_P): Likewise.
29684 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29685 implementation of vec_any_eq to not use AltiVec predicate
29687 (VCMPAEH_P): Likewise.
29688 (VCMPAEW_P): Likewise.
29689 (VCMPAED_P): Likewise.
29690 (VCMPAEFP_P): Likewise.
29691 (VCMPAEDP_P): Likewise.
29692 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29693 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29694 not use the AltiVec predicate framework.
29695 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29696 of vec_any_eq to not use AltiVec predicate framework.
29697 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29698 support for predefined __POWER9_VECTOR__ macro to indicate that
29699 Power9 instruction selection is enabled.
29700 (altivec_overloaded_builtins): Remove extraneous
29701 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29702 function argument types RS6000_BTI_bool_V16QI and
29703 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29704 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29705 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29706 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29707 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29708 Power9 for implementations of vec_cmpne. Change the signature for
29709 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29710 (representing vec_all_ne) to remove the previously described first
29711 argument of type RS6000_BTI_INTSI, as this was an artifact of
29712 reliance on the AltiVec predicate framework, which is no longer
29713 used in the implementation of these functions. Add
29714 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29715 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29716 since, unlike the AltiVec predicate framework implementation, we
29717 do not share function descriptors between vec_alle and vec_anyeq.
29718 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29719 set of modes that receive special treatment even when
29720 TARGET_P9_VECTOR is true. The special treatment emits code that
29721 does not depend on Power9 instructions.
29722 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29723 define_expand to not rely on AltiVec predicate framework.
29724 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29726 (vector_ne_v2di_p): Change this define_expand to not rely on
29727 AltiVec predicate framework.
29728 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29730 (vector_ne_<mode>_p): Change this define_expand to not rely on
29731 AltiVec predicate framework.
29732 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29734 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29735 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29736 define_insn pattern.
29737 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29738 define_insn pattern because the xvcmpne<VSs>. instruction is not
29740 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29741 instruction is not supported.
29743 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29745 * config/nvptx/nvptx.c: Include intl.h.
29747 2017-03-01 Martin Jambor <mjambor@suse.cz>
29750 * ipa-prop.h (ipa_bits): Removed field known.
29751 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29752 to pointers. Adjusted their comments to warn about their sharing.
29753 (ipcp_transformation_summary): Change bits to a vector of pointers.
29754 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29755 (ipa_get_ipa_bits_for_value): Declare.
29756 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29757 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29758 (ipa_bits_hash_table): Likewise.
29759 (ipa_vr_ggc_hash_traits): Likewise.
29760 (ipa_vr_hash_table): Likewise.
29761 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29762 being pointers and vr_known being removed.
29763 (ipa_set_jf_unknown): Likewise.
29764 (ipa_get_ipa_bits_for_value): New function.
29765 (ipa_set_jfunc_bits): Likewise.
29766 (ipa_get_value_range): New overloaded functions.
29767 (ipa_set_jfunc_vr): Likewise.
29768 (ipa_compute_jump_functions_for_edge): Use the above functions to
29769 construct bits and vr parts of jump functions.
29770 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29771 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29773 (ipcp_grow_transformations_if_necessary): Also allocate
29774 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29776 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29777 them. Fix too long lines.
29778 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29779 vr_known being removed.
29780 (ipa_read_jump_function): Use new setter functions to construct bits
29781 and vr parts of jump functions or set them to NULL.
29782 (write_ipcp_transformation_info): Adjust for bits being pointers.
29783 (read_ipcp_transformation_info): Likewise.
29784 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29786 Include gt-ipa-prop.h.
29787 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29789 (ipcp_store_bits_results): Likewise.
29790 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29791 Do not write to existing jump functions but use a temporary instead.
29793 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29796 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29797 attempt to use its first operand as BIT_FIELD_REF base.
29799 2017-03-01 Richard Biener <rguenther@suse.de>
29801 PR middle-end/79721
29802 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29803 interpolating formula in wrapping arithmetic.
29804 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29806 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29808 PR tree-optimization/79734
29809 * tree-vect-generic.c (expand_vector_condition): Optimize
29810 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29811 Handle VEC_COND_EXPR where comparison has different inner width from
29812 type's inner width.
29814 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29816 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29817 markup, and similar issues. Remove @opindex entries for things
29818 that aren't options. Add missing -mmpy-option entries.
29820 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29822 PR tree-optimization/79737
29823 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29824 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29825 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29826 instead of byte_size. Formatting fix.
29827 (shift_bytes_in_array_right): Formatting fix.
29829 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29832 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29833 condition on optimize for the leaf function test.
29835 2017-02-28 Martin Liska <mliska@suse.cz>
29838 * read-rtl-function.c (function_reader::handle_unknown_directive):
29839 Bail out when one uses -flto.
29841 2017-02-28 Martin Liska <mliska@suse.cz>
29843 * common.opt: Replace space with tabular for options of <number>
29845 * config/i386/i386.opt: Show <number> value for
29846 -mlarge-data-threshold.
29847 * opts.c (print_filtered_help): Do not display number in hexadecimal
29850 2017-02-28 Martin Liska <mliska@suse.cz>
29852 * common.opt: Fix --help=option -Q for options which are of
29855 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29857 * config/i386/i386.c (print_reg): Error out for values
29858 of 8-bit size in invalid integer register.
29860 2017-02-28 Martin Sebor <msebor@redhat.com>
29862 PR tree-optimization/79691
29863 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29865 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29868 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29869 gcc_unreachable with output_operand_lossage.
29871 2017-02-28 Richard Biener <rguenther@suse.de>
29873 PR tree-optimization/79740
29874 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29876 (visit_nary_op): Insert the nary into the hashtable if we
29877 pattern-matched sth.
29878 * tree-ssa-pre.c (eliminate_insert): Robustify.
29880 2017-02-28 Richard Biener <rguenther@suse.de>
29882 PR middle-end/79731
29883 * fold-const.c (decode_field_reference): Reject out-of-bound
29886 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29888 * config/i386/i386.c: Include intl.h.
29889 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29890 instead of just cond ? "..." : "...".
29891 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29892 * coverage.c (read_counts_file): Likewise.
29893 * omp-offload.c: Include intl.h.
29894 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29895 of just cond ? "..." : "...".
29896 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29897 of just cond ? "..." : "...".
29899 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29902 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29904 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29905 'tune for' CPU name.
29906 * config/arm/arm-cpu-data.h: Regenerated.
29908 2017-02-28 Richard Biener <rguenther@suse.de>
29910 PR tree-optimization/79732
29911 * tree-inline.c (expand_call_inline): Do not shadow var.
29913 2017-02-28 Richard Biener <rguenther@suse.de>
29915 PR tree-optimization/79723
29916 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29917 address-space properly.
29919 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29921 * doc/optinfo.texi (Optimization groups): Fix option used for
29923 * doc/invoke.texi (-fopt-info): Document "omp".
29924 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29925 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29926 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29927 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29928 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29930 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29932 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29933 OPTGROUP_OMP with "omp".
29935 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29938 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29939 for arithmetic shift of unsigned V2DI.
29941 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29943 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29944 arc/linux.h headers.
29945 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29946 (LINK_SPEC): Likewise.
29947 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29948 (EXTRA_SPECS): Likewise.
29949 (STARTFILE_SPEC): Likewise.
29950 (ENDFILE_SPEC): Likewise.
29951 (LIB_SPEC): Likewise.
29952 (TARGET_SDATA_DEFAULT): Likewise.
29953 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29954 (MULTILIB_DEFAULTS): Likewise.
29955 (DWARF2_UNWIND_INFO): Likewise.
29956 * config/arc/big.h: New file.
29957 * config/arc/elf.h: Likewise.
29958 * config/arc/linux.h: Likewise.
29959 * config/arc/t-uClibc: Remove.
29961 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29963 PR tree-optimization/77536
29964 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29965 (tree_transform_and_unroll_loop): Use above function to compute the
29966 estimated niter of unrolled loop and use it when scaling profile.
29967 Also use count info rather than frequency if it's non-zero.
29968 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29969 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29970 (vect_transform_loop): Call above function.
29972 2017-02-27 Richard Biener <rguenther@suse.de>
29974 PR tree-optimization/45397
29975 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29976 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29977 (visit_nary_op): Add pattern matching for CSEing sign-changed
29978 or truncated operations with wider ones.
29980 2017-02-27 Richard Biener <rguenther@suse.de>
29982 PR tree-optimization/79690
29983 * tree-vect-stmts.c (vectorizable_store): Use vector type
29984 built from the DR with address-space.
29986 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29988 * doc/invoke.texi (Optimize Options): Refine the description
29989 of asan-use-after-return.
29991 2017-02-25 Alan Modra <amodra@gmail.com>
29993 PR rtl-optimization/79584
29994 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29995 base, not ad->base_term, the reg within base. Remove assertion
29996 that ad->base == ad->base_term. Replace gen_int_mode using
29997 bogus mode with const0_rtx.
29999 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30001 PR middle-end/79396
30002 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30003 FMA_EXPR like tcc_binary or tcc_unary.
30005 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30008 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30010 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30011 (output_loc_operands): Handle DW_OP_call_ref and
30012 DW_OP_GNU_variable_value.
30013 (struct variable_value_struct): New type.
30014 (struct variable_value_hasher): Likewise.
30015 (variable_value_hash): New variable.
30016 (string_types): Remove.
30017 (copy_loc_descr): New function.
30018 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30019 (prepend_loc_descr_to_each): New function.
30020 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30021 instead of add_loc_descr_to_each if the first argument is single
30022 location list and the second has multiple.
30023 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30024 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30025 when looking for variable value which doesn't have other location info.
30026 (loc_list_from_tree): Formatting fix.
30027 (gen_array_type_die): Simplify DW_AT_string_length handling.
30028 (adjust_string_types): Remove.
30029 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30030 string_types. Call resolve_variable_values.
30031 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30032 (resolve_addr_in_expr): Likewise. Add A argument.
30033 (copy_deref_exprloc): Remove deref argument. Adjust for the
30034 original expression being DW_OP_GNU_variable_value with optionally
30035 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30036 optionally after it.
30037 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30038 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30039 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30040 (variable_value_hasher::hash, variable_value_hasher::equal): New
30042 (resolve_variable_value_in_expr, resolve_variable_value,
30043 resolve_variable_values, note_variable_value_in_expr,
30044 note_variable_value): New functions.
30045 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30048 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30051 * opts.h (handle_generated_option): Add GENERATED_P argument.
30052 * opts-common.c (handle_option): Adjust function comment.
30053 (handle_generated_option): Add GENERATED_P argument, pass it to
30055 (control_warning_option): Pass false to handle_generated_option
30057 * opts.c (maybe_default_option): Pass true to handle_generated_option
30059 * optc-gen.awk: Likewise.
30061 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30063 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30064 a REG, look at the REG it is a SUBREG of.
30065 (splitter for cmpeqsi_t): Ditto.
30067 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30069 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30070 the special USEs with the pattern of the insn, not the insn itself.
30072 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30075 * doc/invoke.texi: Document -mload-store-pairs.
30077 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30078 Sandra Loosemore <sandra@codesourcery.com>
30080 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30081 argument isn't a CONST_INT.
30082 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30083 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30084 (nios2_validate_compare): Bypass alternate compare logic if *op2
30085 is not a CONST_INT.
30086 (ldstwm_operation_p): Return false if first_base is not a REG or
30087 if first_offset is not a CONST_INT.
30089 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30091 * config/cris/cris.md: Use correct operand in a define_peephole2.
30093 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30095 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30097 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30099 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30100 this_insn if it is an INSN or JUMP_INSN.
30101 (force_offsettable): Look at base, not at addr.
30102 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30103 on things that aren't necessarily CONST_INTs.
30105 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30107 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30108 -mfpmath=sse is the default also for x86-32 targets with SSE2
30109 instruction set when @option{-ffast-math} is enabled
30111 2017-02-24 Jeff Law <law@redhat.com>
30113 PR rtl-optimizatoin/79286
30114 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30117 2017-02-24 Richard Biener <rguenther@suse.de>
30119 PR tree-optimization/79389
30120 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30123 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30125 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30126 function comment to reflect reality.
30127 (loop_exits_before_overflow): Fix typo in function description.
30129 2017-02-24 Richard Biener <rguenther@suse.de>
30131 PR tree-optimization/79389
30132 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30133 properly that a threading opportunity exists. Detect conditional
30134 copy/constant propagation opportunities.
30136 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30138 * config/visium/visium.md (type): Add trap.
30139 (b): New mode attribute.
30140 (*btst): Rename into...
30141 (*btst<mode>): ...this and adjust.
30142 (*cbranchsi4_btst_insn): Rename into...
30143 (*cbranch<mode>4_btst_insn): ...this and adjust.
30144 (trap): New define_insn.
30146 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30148 PR tree-optimization/79389
30149 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30150 (noce_reversed_cond_code): New function.
30151 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30152 reversed_comparison_code. Formatting fix.
30153 (noce_try_store_flag): Test rev_cond != NULL in addition to
30154 reversed_comparison_code.
30155 (noce_try_store_flag_constants): Likewise.
30156 (noce_try_store_flag_mask): Likewise.
30157 (noce_try_addcc): Use rev_cond if non-NULL instead of
30158 reversed_comparison_code.
30159 (noce_try_cmove_arith): Likewise. Formatting fixes.
30160 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30161 (noce_find_if_block): Initialize rev_cond.
30162 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30163 instead of false as last argument never attempt to reverse it
30166 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30168 PR tree-optimization/79663
30169 * tree-predcom.c (combine_chains): Process refs in reverse order
30170 only for ZERO length chains, and add explaining comment.
30172 2017-02-23 Jeff Law <law@redhat.com>
30174 PR tree-optimization/79578
30175 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30176 in call to operand_equal_p.
30178 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30181 * config/i386/cpuid.h: Fix another undefined behavior.
30183 2017-02-23 Richard Biener <rguenther@suse.de>
30185 PR tree-optimization/79683
30186 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30187 vector types for data-refs.
30189 2017-02-23 Martin Liska <mliska@suse.cz>
30191 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30193 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30195 PR middle-end/79665
30196 * internal-fn.c (get_range_pos_neg): Moved to ...
30197 * tree.c (get_range_pos_neg): ... here. No longer static.
30198 * tree.h (get_range_pos_neg): New prototype.
30199 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30200 are known to be in between 0 and signed maximum inclusive, try to
30201 expand both unsigned and signed divmod and use the cheaper one from
30204 2017-02-22 Jeff Law <law@redhat.com>
30206 PR tree-optimization/79578
30207 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30208 to compare base operands.
30210 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30213 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30214 gpc_reg_operand instead of fpr_reg_operand.
30216 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30218 * config/mips/mips.c (mips_return_in_memory): Force FP
30219 vector types to be returned in memory for o32 ABI.
30221 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30223 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30224 instead of DW_TAG_member for static data member declarations and don't
30225 set no_linkage_name for static inline data members.
30226 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30229 2017-02-22 Martin Liska <mliska@suse.cz>
30231 * doc/invoke.texi: Replace inequality signs with square brackets
30234 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30237 * lra-constraints.c (simplify_operand_subreg): Handle
30238 WORD_REGISTER_OPERATIONS targets.
30240 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30243 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30244 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30245 elimination by swapping fld*.
30247 2017-02-22 Richard Biener <rguenther@suse.de>
30249 PR tree-optimization/79673
30250 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30251 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30252 irrelevant address-space qualifiers and avoiding a
30253 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30255 2017-02-22 Richard Biener <rguenther@suse.de>
30257 PR tree-optimization/79666
30258 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30259 to not symbolically negate if that may introduce undefined
30262 2017-02-22 Martin Liska <mliska@suse.cz>
30265 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30266 * data-streamer-out.c (streamer_write_gcov_count_stream):
30268 * value-prof.c (stream_out_histogram_value): Make assert more
30269 precise based on type of counter.
30271 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30274 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30275 Use nonimmediate_operand instead of memory_operand for operand 1.
30276 (float-extend standard_x87sse_constant_load splitter): Ditto.
30278 2017-02-21 Jeff Law <law@redhat.com>
30280 PR tree-optimization/79621
30281 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30282 blocks with edges to themselves.
30284 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30287 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30288 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30289 Use gimple_call_builtin_p.
30292 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30293 on temporarily removed DEBUG_INSNs.
30295 PR tree-optimization/79649
30296 * tree-loop-distribution.c (classify_partition): Give up on
30297 non-generic address space loads/stores.
30299 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30301 * doc/loop.texi (Loop manipulation): Remove nonexistent
30302 tree_ssa_loop_version from the documentation.
30303 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30305 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30308 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30309 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30310 * config/rs6000/rs6000.c: Include except.h.
30311 (rs6000_expand_split_stack_prologue): Call
30312 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30314 2017-02-21 Martin Jambor <mjambor@suse.cz>
30317 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30318 have been analyzed.
30320 2017-02-21 Martin Jambor <mjambor@suse.cz>
30322 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30323 for backward compatibility only.
30324 * doc/invoke.texi (Option Summary): Remove all references to
30325 -fipa-cp-alignment.
30327 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30331 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30333 * lra-constraints.c (curr_insn_transform): Handle
30334 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30336 2017-02-21 Martin Liska <mliska@suse.cz>
30338 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30340 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30342 PR translation/79638
30343 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30345 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30348 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30349 (arm_function_ok_for_sibcall): Return false for an indirect call by
30350 descriptor if all the argument registers are used.
30351 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30352 alignment of the function.
30354 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30356 PR tree-optimization/61441
30357 * simplify-rtx.c (simplify_const_unary_operation): For
30358 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30359 the sNaN unmodified.
30361 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30363 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30364 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30365 instead of SYSTEM_HEADER_DIR.
30367 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30368 Martin Liška <mliska@suse.cz>
30370 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30371 Fix typos and grammar, use active voice, and clarify.
30373 2017-02-20 Marek Polacek <polacek@redhat.com>
30375 PR middle-end/79537
30376 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30379 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30381 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30384 * config/i386/i386.c (ix86_expand_builtin): Handle
30385 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30386 ix86_builtins_isa[fcode].isa as a requirement of those
30387 flags and any other flag in the bitmask.
30388 (ix86_init_mmx_sse_builtins): Use 0 instead of
30389 ~OPTION_MASK_ISA_64BIT as mask.
30390 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30391 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30392 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30393 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30395 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30398 * lra-constraints.c (split_reg): Check requested split mode
30399 is supported by the register.
30401 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30403 * lra-constraints.c (simplify_operand_subreg): Remove early
30406 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30409 * lra-constraints.c (curr_insn_transform): Tighten condition
30410 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30412 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30415 * lra-constraints.c (curr_insn_transform): Handle
30416 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30418 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30421 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30423 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30425 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30428 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30431 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30433 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30436 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30438 * final.c (last_columnnum, override_columnnum): New variables.
30439 (final_start_function): Set last_columnnum, pass it to begin_prologue
30440 hook and pass 0 to dwarf2out_begin_prologue.
30441 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30442 to source_line debug hook.
30443 (notice_source_line): Compute last_columnnum and for debug_column_info
30444 return true on column changes.
30445 * debug.h (struct gcc_debug_hooks): Add column argument to
30446 source_line and begin_prologue hooks.
30447 (debug_nothing_int_charstar_int_bool): Remove prototype.
30448 (debug_nothing_int_int_charstar,
30449 debug_nothing_int_int_charstar_int_bool): New prototypes.
30450 (dwarf2out_begin_prologue): Add column argument.
30451 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30452 begin_prologue hooks.
30453 (debug_nothing_int_charstar_int_bool): Remove.
30454 (debug_nothing_int_int_charstar,
30455 debug_nothing_int_int_charstar_int_bool): New functions.
30456 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30457 through to dwarf2out_source_line.
30458 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30459 (dwarf2out_source_line): Add column argument, emit it if requested.
30460 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30462 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30463 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30464 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30465 through to dwarf2out_begin_prologue.
30466 (vmsdbgout_source_line): Add column argument, pass it through to
30467 dwarf2out_source_line.
30468 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30469 dbxout_source_line caller.
30470 (dbxout_source_line): Add column argument.
30472 * common.opt (gno-column-info, gcolumn-info): New options.
30473 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30474 (check_die): Also test for multiple DW_AT_decl_column attributes.
30475 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30476 DW_AT_decl_column if requested.
30477 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30479 (gen_variable_die): Likewise.
30480 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30481 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30484 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30485 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30486 (ix86_handle_option): Handle OPT_m3dnowa.
30487 * doc/invoke.texi (-m3dnowa): Document.
30488 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30489 -m3dnowa instead of -m3dnow -march=athlon.
30492 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30493 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30495 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30498 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30499 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30500 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30501 generator for vsx_xxpermdi_<mode>_be.
30502 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30503 force big-endian semantics.
30504 (vsx_xxpermdi_<mode>_be): New define_expand with same
30505 implementation as previous version of vsx_xxpermdi_<mode>.
30507 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30509 PR tree-optimization/79327
30510 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30511 variable, its initialization and use.
30513 2017-02-17 Julia Koval <julia.koval@intel.com>
30515 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30516 (OPTION_MASK_ISA_PKU_UNSET): New.
30517 (ix86_handle_option): Handle -mrdpid.
30518 * config/i386/cpuid.h (bit_RDPID): New.
30519 * config/i386/driver-i386.c (host_detect_local_cpu):
30520 Detect RDPID feature.
30521 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30522 * config/i386/i386-c.c (ix86_target_macros_internal):
30524 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30525 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30526 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30527 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30528 * config/i386/i386.md (define_insn "rdpid"): New.
30529 * config/i386/i386.opt Add -mrdpid.
30530 * config/i386/immintrin.h (_rdpid_u32): New.
30532 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30534 PR rtl-optimization/79541
30535 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30536 instead of transforming it into USE.
30538 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30540 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30541 If HONOR_SNANS (SFmode) force the input to a register.
30542 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30543 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30544 an frsp or similar insn.
30546 2017-02-17 Martin Liska <mliska@suse.cz>
30548 PR rtl-optimization/79577
30549 * params.def (selsched-max-sched-times): Increase minimum to 1.
30551 2017-02-17 Martin Liska <mliska@suse.cz>
30553 PR rtl-optimization/79574
30554 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30556 2017-02-17 Martin Liska <mliska@suse.cz>
30558 PR tree-optimization/79529
30559 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30560 ssa_defined_default_def_p to handle cases which are implicitly
30562 * tree-ssa.c (ssa_defined_default_def_p): New function.
30563 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30564 which are implicitly defined.
30565 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30567 2017-02-17 Richard Biener <rguenther@suse.de>
30569 PR middle-end/79576
30570 * params.def (max-ssa-name-query-depth): Limit to 10.
30572 2017-02-17 Richard Biener <rguenther@suse.de>
30574 PR tree-optimization/79552
30575 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30578 2017-02-17 Richard Biener <rguenther@suse.de>
30581 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30583 2017-02-17 Marek Polacek <polacek@redhat.com>
30585 PR middle-end/79536
30586 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30587 (fold_negate_expr): New wrapper.
30589 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30591 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30592 Correct terminology and de-emphasize pre-standard behavior.
30594 2017-02-16 Alan Modra <amodra@gmail.com>
30596 PR rtl-optimization/79286
30597 * ira.c (def_dominates_uses): New function.
30598 (update_equiv_regs): Don't create an equivalence for insns that
30599 may trap where the register def does not dominate the use.
30601 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30603 PR rtl-optimization/78127
30604 * lra.c (lra): Call lra_eliminate before finish the loop after
30607 2017-02-16 Richard Biener <rguenther@suse.de>
30609 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30611 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30612 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30613 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30614 (isl_val_int_from_wi): New function.
30615 (extract_affine_gmp): Rename to ...
30616 (extract_affine_wi): ... this, take a widest_int.
30617 (extract_affine_int): Just wrap extract_affine_wi.
30618 (add_param_constraints): Use isl_val_int_from_wi.
30619 (add_loop_constraints): Likewise, and extract_affine_wi.
30621 2017-02-15 Jeff Law <law@redhat.com>
30623 PR middle-end/79521
30624 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30625 ira_init_register_move_cost_if_necessary.
30627 2017-02-15 Martin Sebor <msebor@redhat.com>
30629 PR middle-end/32003
30630 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30631 removed in a prior commit.
30633 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30635 PR tree-optimization/79347
30636 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30637 counters during peeling.
30639 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30641 * Makefile.in (site.exp): Remove "set ISLVER".
30643 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30646 * real.c (real_from_integer): Call real_convert even for decimal.
30648 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30651 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30653 2017-02-14 Andrew Pinski <apinski@cavium.com>
30655 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30656 cores and change the partno/implementer to be correct.
30657 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30658 the 'B" as the implementer.
30659 * config/aarch64/aarch64-tune.md: Regenerate.
30661 2017-02-14 Carl Love <cel@us.ibm.com>
30663 * config/rs6000/rs6000.c: Add case statement entry to make the
30664 xvcvuxdsp built-in argument unsigned.
30665 * config/rs6000/vsx.md: Fix the source and return operand types so they
30666 match the instruction definitions from the ISA document. Fix typo
30667 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30670 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30673 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30674 member early_clobber_alts.
30675 * lra-lives.c (reg_early_clobber_p): New.
30676 (process_bb_lives): Use it.
30677 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30678 (debug_operand_data): Initialize early_clobber_alts.
30679 (setup_operand_alternative): Set up early_clobber_alts.
30680 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30681 alternatives to new_insn_reg.
30682 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30684 (lra_update_insn_regno_info): Pass the new arg.
30686 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30688 PR middle-end/79505
30689 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30690 (new_oacc_loop_raw): Don't clear already cleared fields.
30693 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30694 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30695 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30697 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30700 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30702 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30705 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30706 the extra instruction to the right place to store 128-bit constant
30709 2017-02-14 Martin Sebor <msebor@redhat.com>
30711 PR middle-end/79448
30712 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30713 warning for strings of unknown length.
30715 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30717 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30719 2017-02-14 Jeff Law <law@redhat.com>
30722 * ira-costs.c (scan_one_insn): Initialize register move costs
30723 for pseudos seen in USE/CLOBBER insns.
30725 PR tree-optimization/79095
30726 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30727 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30728 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30729 if the operands are known to be not equal, then the resulting range
30731 (intersect_ranges): If the new range is ~[0,0] and the old range is
30732 wide, then prefer ~[0,0].
30733 * tree-vrp.c (overflow_comparison_p_1): New function.
30734 (overflow_comparison_p): New function.
30735 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30736 if NAME is used in an overflow test.
30737 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30738 overflow check that can be expressed as an equality test, then adjust
30739 ops to be that equality test.
30741 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30743 * config/s390/s390-builtin-types.def: Remove flags argument.
30744 * config/s390/s390.c (s390_init_builtins): Likewise.
30746 2017-02-14 Martin Liska <mliska@suse.cz>
30748 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30749 vector. Fix trailing white spaces.
30751 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30753 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30756 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30758 PR rtl-optimization/68664
30759 * config/arm/arm.c (arm_sched_can_speculate_insn):
30760 New function. Declare prototype.
30761 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30763 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30765 PR rtl-optimization/68664
30766 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30768 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30770 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30772 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30773 max skip bytes for function, loop and jump.
30775 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30777 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30778 ABS_EXPR for gimple dump.
30780 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30783 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30785 PR tree-optimization/79408
30786 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30787 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30788 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30789 also if rhs1 is INTEGER_CST.
30791 2017-02-14 Richard Biener <rguenther@suse.de>
30793 PR middle-end/79432
30794 * tree-into-ssa.c (insert_phi_nodes): When the function can
30795 have abnormal edges rewrite SSA names with broken use-def
30796 dominance out of SSA and register them for PHI insertion.
30798 2017-02-13 Martin Sebor <msebor@redhat.com>
30800 PR middle-end/79496
30801 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30802 clearing info.nowrite flag when snprintf size argument is a range.
30804 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30806 * cprop.c (cprop_jump): Add missing space in string literal.
30807 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30808 (get_constraint_for_component_ref): Likewise.
30809 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30810 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30811 * lra-constraints.c (process_alt_operands): Likewise.
30812 * ipa-inline.c (inline_small_functions): Likewise.
30813 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30814 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30815 * trans-mem.c (diagnose_tm_1_op): Likewise.
30816 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30817 (grid_parallel_clauses_gridifiable): Likewise.
30819 * config/nvptx/mkoffload.c (process): Add space in between
30822 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30823 "MOD4_SSE_REGS" and "ALL_REGS".
30825 * spellcheck.c (test_data): Add , in between "foo" and "food".
30827 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30830 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30831 boundary crossing check and subsequent code generation agree.
30833 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30835 * config/aarch64/aarch64.c (has_memory_op): Delete.
30836 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30839 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30841 PR rtl-optimization/79388
30842 PR rtl-optimization/79450
30843 * combine.c (distribute_notes): When removing TEM_INSN for which
30844 corresponding dest has last value recorded, invalidate that last
30847 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30849 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30850 of explicit '@'. Add missing assembly comment marker on branch costs
30853 2017-02-13 Nathan Sidwell <nathan@acm.org>
30855 * gengtype-lex.l (<in_struct>): Add '/'.
30857 2017-02-13 Martin Liska <mliska@suse.cz>
30860 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30862 2017-02-13 Richard Biener <rguenther@suse.de>
30864 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30866 * configure: Re-generate.
30867 * config.in: Likewise.
30868 * graphite-dependences.c: Simplify as if
30869 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30870 * graphite-isl-ast-to-gimple.c: Likewise.
30871 * graphite-optimize-isl.c: Likewise.
30872 * graphite-poly.c: Likewise.
30873 * graphite-sese-to-poly.c: Likewise.
30874 * graphite.h: Likewise.
30875 * toplev.c: Include isl/version.h and use isl_version () for
30876 printing the ISL version.
30877 * doc/install.texi: Update ISL requirement.
30879 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30881 * doc/standards.texi (Standards): Update reference to
30884 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30886 * doc/extend.texi (Named Address Spaces): sourceware.org now
30888 * doc/install.texi (Binaries): Ditto.
30891 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30893 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30894 standard terminology "stringize"/"stringizing" throughout.
30895 * doc/cppinternals.texi: Likewise.
30897 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30899 * doc/extend.texi: Fix some spelling mistakes and typos.
30900 * doc/invoke.texi: Likewise.
30902 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30905 * params.def (inline-min-speedup) Change from 10 to 8.
30907 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30909 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30912 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30915 * ipa-inline-analysis.c (get_minimal_bb): New function.
30916 (record_modified): Use it.
30917 (remap_edge_change_prob): Handle also ancestor functions.
30919 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30921 * doc/contrib.texi (Contributors): Remove broken link into
30922 the Mauve CVS repository.
30924 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30926 PR middle-end/79454
30927 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30928 result computation whenever lhs doesn't have vector mode, not
30929 just when it has BLKmode.
30931 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30933 * doc/makefile.texi (profiledbootstrap): Refer to the
30934 installation instructions only in textual form.
30936 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30939 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30941 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30943 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30944 (Specific): Update mingw-w64 reference.
30946 (Specific): Remove broken link to Renesas RX processor.
30948 2017-02-10 Richard Biener <rguenther@suse.de>
30950 * toplev.c (process_options): Do not mention obsolete graphite
30951 options when printing sorry message about missing graphite support.
30952 Mention -floop-nest-optimize.
30954 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30956 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30957 (vtst_p16): Likewise.
30958 (vtstq_p8): Likewise.
30959 (vtstq_p16): Likewise.
30961 (vtstq_p64): Likewise.
30962 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30963 (vset_lane_p64): New.
30964 (vsetq_lane_p64): New.
30966 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30968 PR tree-optimization/79411
30969 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30970 stmt operands are SSA_NAMEs used in abnormal phis.
30971 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30974 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30977 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30980 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30982 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30984 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30986 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30990 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30991 not arbitrary TREE_CONSTANT.
30994 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30995 "omp declare target link" attribute unless is_global_var.
30996 * omp-offload.c (find_link_var_op): Likewise.
30998 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30999 Chung-Lin Tang <cltang@codesourcery.com>
31001 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31003 (gimplify_adjust_omp_clauses): Don't delete TILE.
31004 (gimplify_omp_for): Deal with TILE.
31005 * internal-fn.c (expand_GOACC_TILE): New function.
31006 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31008 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31009 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31011 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31012 avoid DIV for outermost collapse var.
31013 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31014 Remove out of date comments, fix whitespace.
31015 * omp-general.c (omp_extract_for_data): Deal with tiling.
31016 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31017 adjust OLF_DIM_BASE value.
31018 (struct omp_for_data): Add tiling field.
31019 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31020 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31021 for auto loops. Remove default auto determining, moved to
31022 oacc_loop_fixed_partitions.
31023 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31024 stmts, add e_mask field.
31025 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31026 (oacc_thread_numbers): Use oacc_dim_call.
31027 (oacc_xform_tile): New.
31028 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31029 (finish_oacc_loop): Adjust for ifns vector.
31030 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31031 add case for GOACC_TILE fns.
31032 (oacc_loop_xform_loop): Delete.
31033 (oacc_loop_process): Iterate over call list directly, and add
31034 handling for GOACC_TILE fns.
31035 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31037 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31038 vector partitioning to outer loops. Assign 2 partitions to loops
31039 when available. Add TILE handling.
31040 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31041 (execite_oacc_device_lower): Process GOACC_TILE fns,
31042 ignore unknown specs.
31043 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31044 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31045 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31047 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31049 * configure.ac (ACX_BUGURL): Update.
31050 * configure: Regenerate.
31052 2017-02-09 Richard Biener <rguenther@suse.de>
31054 PR tree-optimization/69823
31055 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31056 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31058 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31060 * config/arc/arc-c.def: Add __NPS400__ definition.
31061 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31062 (TARGET_NPS400): Define.
31064 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31066 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31068 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31069 pointer, arch_info.
31070 (arc_cpu_types): Fill the arch_info field with a pointer into the
31071 arc_arch_types table.
31072 (arc_selected_cpu): Declare.
31073 * config/arc/arc.c (arc_selected_cpu): Make global.
31074 (arc_selected_arch): Delete.
31075 (arc_base_cpu): Delete.
31076 (arc_override_options): Remove references to deleted variables,
31077 update access to arch information.
31078 (ARC_OPT): Update access to arch information.
31079 (ARC_OPTX): Likewise.
31080 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31081 (TARGET_ARC600): Update access to arch information.
31082 (TARGET_ARC601): Likewise.
31083 (TARGET_ARC700): Likewise.
31084 (TARGET_EM): Likewise.
31085 (TARGET_HS): Likewise.
31086 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31089 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31092 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31093 condition/operands for integer GE/LE/GEU/LEU operations.
31095 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31097 PR translation/79397
31098 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31101 2017-02-08 Martin Jambor <mjambor@suse.cz>
31104 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31105 whether allocation happened.
31106 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31107 nothing was allocated.
31109 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31111 PR tree-optimization/79408
31112 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31113 constant, but SSA_NAME with a known integer range, use the minimum
31114 of that range instead of op1 to determine if modulo can be replaced
31115 with its first operand.
31117 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31119 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31121 2017-02-08 Richard Biener <rguenther@suse.de>
31123 PR tree-optimization/71824
31124 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31125 Check all loops contained in the merged region.
31127 2017-02-07 Andrew Pinski <apinski@cavium.com>
31129 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31131 2017-02-07 Andrew Pinski <apinski@cavium.com>
31133 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31134 (thunderxt88): Likewise.
31135 (thunderxt81): Disable LSE and change v8.1 to v8.
31136 (thunderxt83): Likewise.
31138 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31139 Richard Biener <rguenther@suse.de>
31141 PR middle-end/79399
31142 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31143 type from int to size_t.
31144 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31146 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31148 PR rtl-optimization/79386
31149 * cprop.c (bypass_conditional_jumps): Initialize
31150 bypass_last_basic_block already before splitting bbs after
31151 unconditional traps...
31152 (bypass_conditional_jumps): ... rather than here.
31155 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31156 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31157 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31158 fix -masm=intel patterns.
31160 2017-02-07 Richard Biener <rguenther@suse.de>
31162 PR tree-optimization/79256
31163 PR middle-end/79278
31164 * builtins.c (get_object_alignment_2): Use min_align_of_type
31165 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31166 and ADJUST_FIELD_ALIGN.
31168 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31170 * doc/tm.texi: Regenerate.
31171 * stor-layout.c (layout_decl): Adjust.
31172 (update_alignment_for_field): Likewise.
31173 (place_field): Likewise.
31174 (min_align_of_type): Likewise.
31175 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31176 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31177 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31178 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31179 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31180 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31181 * config/i386/i386.c (x86_field_alignment): Likewise.
31182 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31183 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31184 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31185 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31186 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31187 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31191 2017-01-30 Richard Biener <rguenther@suse.de>
31193 PR tree-optimization/79256
31194 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31195 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31198 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31200 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31201 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31202 builtins to SImode and emit a zero-extend, if necessary.
31204 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31206 * docs/invoke.texi (RISC-V Options): Alphabetize.
31208 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31210 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31213 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31215 * config/riscv/riscv.c: New file.
31216 * common/config/riscv/riscv-common.c: Likewise.
31217 * config.gcc: Likewise.
31218 * config/riscv/constraints.md: Likewise.
31219 * config/riscv/elf.h: Likewise.
31220 * config/riscv/generic.md: Likewise.
31221 * config/riscv/linux.h: Likewise.
31222 * config/riscv/multilib-generator: Likewise.
31223 * config/riscv/peephole.md: Likewise.
31224 * config/riscv/pic.md: Likewise.
31225 * config/riscv/predicates.md: Likewise.
31226 * config/riscv/riscv-builtins.c: Likewise.
31227 * config/riscv/riscv-c.c: Likewise.
31228 * config/riscv/riscv-ftypes.def: Likewise.
31229 * config/riscv/riscv-modes.def: Likewise.
31230 * config/riscv/riscv-opts.h: Likewise.
31231 * config/riscv/riscv-protos.h: Likewise.
31232 * config/riscv/riscv.h: Likewise.
31233 * config/riscv/riscv.md: Likewise.
31234 * config/riscv/riscv.opt: Likewise.
31235 * config/riscv/sync.md: Likewise.
31236 * config/riscv/t-elf-multilib: Likewise.
31237 * config/riscv/t-linux: Likewise.
31238 * config/riscv/t-linux-multilib: Likewise.
31239 * config/riscv/t-riscv: Likewise.
31240 * configure.ac: Likewise.
31241 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31242 Waterman as RISC-V maintainers.
31243 * doc/install.texi: Add RISC-V entries.
31244 * doc/invoke.texi: Add RISC-V options section.
31245 * doc/md.texi: Add RISC-V constraints section.
31246 * configure: Regenerated.
31248 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31251 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31252 false values to be constant vectors with all 0 or all 1 bits set.
31253 (vcondu<mode><mode>): Likewise.
31254 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31256 (fpmask_comparison_operator): Update comment.
31257 (vecint_comparison_operator): New predicate.
31258 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31259 vector conditionals when the true and false values are constant
31260 vectors with all 0 bits or all 1 bits set.
31262 2017-02-06 Martin Sebor <msebor@redhat.com>
31264 PR tree-optimization/79376
31265 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31267 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31269 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31270 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31271 to simplify split condition.
31273 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31275 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31276 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31279 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31281 PR rtl-optimization/68664
31282 * target.def (can_speculate_insn): New hook.
31283 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31284 * doc/tm.texi: Regenerate.
31285 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31286 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31287 (rs6000_sched_can_speculate_insn): New function.
31289 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31291 PR tree-optimization/79284
31292 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31293 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31294 vectorizable_mask_load_store, vectorizable_operation,
31295 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31296 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31297 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31298 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31299 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31300 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31301 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31302 is_gimple_assign (stmt). Replace another such test with
31303 is_gimple_assign (stmt).
31305 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31308 * config/avr/avr.c (rtl-iter.h): Include it.
31309 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31310 (avr_legitimate_combined_insn): ...and implementation.
31312 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31314 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31315 * config/s390/s390.c (s390_const_operand_ok)
31316 (s390_canonicalize_comparison, s390_extract_part)
31317 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31318 (s390_contiguous_bitmask_p, s390_rtx_costs)
31319 (legitimize_pic_address): Likewise.
31320 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31321 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31322 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31323 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31324 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31326 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31328 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31329 REGNO($0) == REGNO($1).
31331 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31333 * config/s390/linux.h(SIZE_TYPE): Add comment.
31335 2017-02-06 Julian Brown <julian@codesourcery.com>
31336 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31337 Virendra Pathak <virendra.pathak@broadcom.com>
31339 * config/aarch64/aarch64-cores.def: Change the scheduler
31341 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31342 * config/aarch64/thunderx2t99.md: New file.
31344 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31346 * doc/standards.texi (Go Language): Update link to language
31349 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31351 * tree-eh.c (lower_resx): Sanitize profile.
31352 (cleanup_empty_eh_move_lp): Likewise.
31354 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31357 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31359 * cfgloopmanip.h (loop_version): Update prototype.
31360 * modulo-sched.c (sms_schedule): Update call of loop_version.
31361 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31362 * tree-parloops.c (gen_parallel_loop): Likewise.
31363 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31364 * tree-ssa-loop-split.c (split_loop): Likewise.
31365 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31366 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31368 2017-02-05 Martin Liska <mliska@suse.cz>
31371 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31373 (print_operand_address): Initialize a struct to zero.
31375 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31377 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31378 garbage collector only in textual form.
31380 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31382 * doc/extend.texi (x86 specific memory model extensions for
31383 transactional memory): Simplify a phrase.
31385 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31388 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31389 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31390 (atomic_storedi_1): Likewise.
31392 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31394 PR tree-optimization/79338
31395 * tree-parloops.c (gather_scalar_reductions): Don't call
31396 vect_analyze_loop_form for loop->inner before destroying loop's
31399 2017-02-03 Martin Sebor <msebor@redhat.com>
31401 PR tree-optimization/79327
31402 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31403 when precision has resulted in leading zeros.
31404 (format_integer): Adjust the likely counter to assume an unknown
31405 argument that may be zero is non-zero.
31407 2017-02-03 Jason Merrill <jason@redhat.com>
31410 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31411 avoid copying non-taken branch.
31413 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31415 PR tree-optimization/79340
31416 * tree-vect-loop.c (vectorizable_reduction): Release
31417 vec_defs elements after safe_splicing them into other vectors.
31420 PR tree-optimization/79327
31421 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31422 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31424 (format_integer): Use wide_int_to_tree instead of build_int_cst
31425 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31426 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31427 of shortest and longest sequence.
31429 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31431 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31432 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31434 2017-02-03 Walter Lee <walt@tilera.com>
31437 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31438 after initial stackframe link reg save.
31439 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31441 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31444 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31445 wu for stxssp alternative.
31447 2017-02-03 Martin Sebor <msebor@redhat.com>
31449 PR tree-optimization/79352
31450 * gimple-fold.c (get_range_strlen): Add argument.
31451 (get_range_strlen): Change return type to bool.
31452 (get_maxval_strlen): Pass in a dummy argument.
31453 * gimple-fold.h (get_range_strlen): Change return type to bool.
31454 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31455 * tree.h (array_at_struct_end_p): Add argument.
31456 * tree.c (array_at_struct_end_p): Handle it.
31458 2017-02-03 Martin Liska <mliska@suse.cz>
31461 * multiple_target.c (create_dispatcher_calls): Redirect edge
31462 from a caller of a dispatcher.
31463 (expand_target_clones): Make the clones local.
31464 (ipa_target_clone): Do both target clones and resolvers.
31465 (ipa_dispatcher_calls): Remove the pass.
31466 (pass_dispatcher_calls::gate): Likewise.
31467 (make_pass_dispatcher_calls): Likewise.
31468 * passes.def (pass_target_clone): Put as very first IPA early
31471 2017-02-03 Martin Liska <mliska@suse.cz>
31473 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31474 in case of a function with ifunc attribute.
31476 2017-02-03 Martin Liska <mliska@suse.cz>
31478 * cgraph.c (cgraph_node::dump): Dump function version info.
31479 * symtab.c (symtab_node::dump_base): Add missing new line.
31481 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31483 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31484 (ifcombine_ifandif): Use it.
31486 2017-02-03 Martin Liska <mliska@suse.cz>
31488 * doc/invoke.texi: Document default value for
31489 use-after-scope-direct-emission-threshold.
31491 2017-02-03 Martin Liska <mliska@suse.cz>
31493 PR tree-optimization/79339
31494 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31495 (format_floating): Likewise.
31497 2017-02-03 Martin Liska <mliska@suse.cz>
31500 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31502 (ipa_node_params_t::remove): Likewise.
31503 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31504 initialization from removed ipa_node_params_t::insert.
31505 (ipa_node_params::~ipa_node_params): Move from removed
31506 ipa_node_params_t::release.
31507 * symbol-summary.h (symbol_summary::m_released): New member.
31508 Do not release a summary twice. Do not allow to call finalizer
31509 for types of a summary that live in GGC memory.
31511 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31513 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31516 2017-02-02 Martin Sebor <msebor@redhat.com>
31518 PR middle-end/79275
31519 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31520 (format_string): Tighten up the range of output for non-constant
31521 strings and correct the expected range for wide non-constant strings.
31523 2017-02-02 Martin Sebor <msebor@redhat.com>
31525 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31527 PR middle-end/32003
31528 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31530 (-fdump-tree-@var): Add to index and document how to come up
31531 with pass-specific option and dump file names.
31532 (-fdump-passes): Clarify where to look for output.
31534 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31536 PR middle-end/77445
31537 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31538 statistics of the analyzed path; allow threading for speed when
31539 any of BBs along the path are optimized for speed.
31541 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31543 PR middle-end/78468
31544 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31545 settings of the virtual registers.
31548 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31550 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31551 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31554 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31556 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31557 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31559 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31561 * config/s390/s390.md: Add missing comments with the expanded
31563 * config/s390/vector.md: Likewise.
31564 * config/s390/vx-builtins.md: Likewise.
31566 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31569 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31570 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31571 conditions on a single line.
31573 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31575 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31576 __S390_VX__ to __VX__.
31578 2017-02-01 Andrew Pinski <apinski@cavium.com>
31580 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31581 stmt_info to record_stmt_cost.
31582 (vect_get_known_peeling_cost): Pass stmt_info if known to
31584 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31585 cpu_vector_cost field into
31586 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31587 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31588 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31589 splitting of scalar_stmt_cost and vec_stmt_cost.
31590 (thunderx_vector_cost): Likewise.
31591 (cortexa57_vector_cost): LIkewise.
31592 (exynosm1_vector_cost): Likewise.
31593 (xgene1_vector_cost): Likewise.
31594 (thunderx2t99_vector_cost): Improve after the splitting of the two
31596 (aarch64_builtin_vectorization_cost): Update for the splitting of
31597 scalar_stmt_cost and vec_stmt_cost.
31599 2017-02-01 Torvald Riegel <triegel@redhat.com>
31600 Richard Henderson <rth@redhat.com>
31602 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31603 conditional on existance of a fast atomic load.
31604 * optabs-query.c (can_atomic_load_p): New function.
31605 * optabs-query.h (can_atomic_load_p): Declare it.
31606 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31607 no fast atomic load is available for the particular size of access.
31608 (expand_atomic_compare_and_swap): Likewise.
31609 (expand_atomic_load): Likewise.
31610 (expand_atomic_store): Likewise.
31611 (expand_atomic_fetch_op): Likewise.
31612 * testsuite/lib/target-supports.exp
31613 (check_effective_target_sync_int_128): Remove x86 because it provides
31614 no fast atomic load.
31615 (check_effective_target_sync_int_128_runtime): Likewise.
31617 2017-02-01 Richard Biener <rguenther@suse.de>
31619 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31620 (graphite_transform_loops): Provide opt-info for optimized nests.
31621 * tree-parloop.c (parallelize_loops): Provide opt-info for
31622 parallelized loops.
31624 2017-02-01 Richard Biener <rguenther@suse.de>
31626 PR middle-end/79315
31627 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31628 was not set before.
31630 2017-02-01 Richard Biener <rguenther@suse.de>
31632 PR tree-optimization/71824
31633 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31634 Verify the loops are valid in the merged SESE region.
31635 (scop_detection::can_represent_loop_1): Check analyzing the
31636 evolution of the number of iterations in the region succeeds.
31638 2017-01-31 Ian Lance Taylor <iant@golang.org>
31640 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31641 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31643 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31645 PR preprocessor/79210
31646 * input.c (get_substring_ranges_for_loc): Replace line_width
31647 assertion with error-handling.
31649 2017-01-31 Richard Biener <rguenther@suse.de>
31651 PR tree-optimization/77318
31652 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31653 (create_pw_aff_from_tree): Take loop parameter.
31654 (add_condition_to_pbb): Pass loop of the condition to
31655 create_pw_aff_from_tree.
31657 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31659 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31660 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31662 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31666 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31668 (convert_int_to_float128): Likewise.
31669 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31670 (convert_int_to_float128): Likewise.
31671 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31672 (UNSPEC_IEEE128_CONVERT): Likewise.
31673 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31674 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31675 Use local variables for IBM extended format.
31676 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31677 (fix_trunc<mode>si2_fprs): Likewise.
31678 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31679 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31680 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31681 to know that we can now have integers of all sizes in vector
31683 (fix<uns>_<mode>di2_hw): Likewise.
31684 (float<uns>_<mode>si2_hw): Likewise.
31685 (fix_<mode>si2_hw): Likewise.
31686 (fixuns_<mode>si2_hw): Likewise.
31687 (float<uns>_<mode>di2_hw): Likewise.
31688 (float_<mode>di2_hw): Likewise.
31689 (float_<mode>si2_hw): Likewise.
31690 (floatuns_<mode>di2_hw): Likewise.
31691 (floatuns_<mode>si2_hw): Likewise.
31692 (xscvqp<su>wz_<mode>): Delete, no longer used.
31693 (xscvqp<su>dz_<mode>): Likewise.
31694 (xscv<su>dqp_<mode>): Likewise.
31695 (ieee128_mfvsrd_64bit): Likewise.
31696 (ieee128_mfvsrd_32bit): Likewise.
31697 (ieee128_mfvsrwz): Likewise.
31698 (ieee128_mtvsrw): Likewise.
31699 (ieee128_mtvsrd_64bit): Likewise.
31700 (ieee128_mtvsrd_32bit): Likewise.
31702 2017-01-31 Martin Liska <mliska@suse.cz>
31705 * ipa-prop.c (ipa_free_all_node_params): Call release method
31706 instead of ~sumbol_summary to not to trigger double times
31709 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31711 PR tree-optimization/71691
31712 * bitmap.h (class auto_bitmap): New.
31713 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31714 is_maybe_undefined instead of ssa_undefined_value_p.
31716 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31718 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31719 __S390_ARCH_LEVEL__ to __ARCH__.
31721 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31723 PR tree-optimization/79267
31724 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31725 if should_remove_lhs_p is true.
31727 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31730 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31731 (add_alignment_attribute): New.
31732 (base_type_die): Add alignment attribute.
31733 (subrange_type_die): Likewise.
31734 (modified_type_die): Likewise.
31735 (gen_array_type_die): Likewise.
31736 (gen_descr_array_type_die: Likewise.
31737 (gen_enumeration_type_die): Likewise.
31738 (gen_subprogram_die): Likewise.
31739 (gen_variable_die): Likewise.
31740 (gen_field_die): Likewise.
31741 (gen_ptr_to_mbr_type_die): Likewise.
31742 (gen_struct_or_union_type_die): Likewise.
31743 (gen_subroutine_type_die): Likewise.
31744 (gen_typedef_die): Likewise.
31745 (base_type_cmp): Compare alignment attribute.
31747 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31750 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31751 (setb_unsigned) New pattern for setb with CCUNS.
31752 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31753 subfc./subfe sequence to avoid overflow problems. Generate a
31754 shorter sequence with cmpld/setb for power9.
31755 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31756 for generating subfc. instruction.
31757 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31758 now uses this instruction.
31760 2017-01-30 Ian Lance Taylor <iant@google.com>
31763 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31764 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31766 2017-01-30 Martin Sebor <msebor@redhat.com>
31768 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31769 Move constant to the right of a relational operator.
31770 (get_mpfr_format_length, format_character, format_string): Ditto.
31771 (should_warn_p, maybe_warn): Same.
31773 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31775 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31778 * asan.c (get_translation_unit_decl): Remove function.
31779 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31781 2017-01-30 Martin Liska <mliska@suse.cz>
31783 PR gcov-profile/79259
31784 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31785 -fprofile-generate.
31787 2017-01-30 Martin Liska <mliska@suse.cz>
31790 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31791 Initialize variables with NULL value.
31793 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31796 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31798 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31800 2017-01-30 Richard Biener <rguenther@suse.de>
31802 PR tree-optimization/79276
31803 * tree-vrp.c (process_assert_insertions): Properly adjust common
31804 when removing a duplicate.
31806 2017-01-30 Richard Biener <rguenther@suse.de>
31808 PR tree-optimization/79256
31809 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31810 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31812 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31814 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31817 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31818 ("*r<noxa>sbg_<mode>_sll_bitmask")
31819 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31820 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31821 Use contiguous_bitmask_nowrap_operand.
31823 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31826 * config/rs6000/altivec.h (vec_xl): Revise #define.
31827 (vec_xst): Likewise.
31829 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31831 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31833 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31835 PR rtl-optimization/79194
31836 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31837 traps before call to bypass_conditional_jumps.
31839 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31841 PR tree-optimization/71374
31842 * lra-constraints.c (check_conflict_input_operands): New.
31843 (match_reload): Use it.
31845 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31848 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31849 account to calculate conflict_set.
31851 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31853 PR rtl-optimization/78559
31854 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31855 other_insn in combine.
31857 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31859 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31860 uint16_type_node for BT_UINT16.
31862 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31864 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31865 "RTL Tests" to menu.
31866 (GIMPLE Tests): New node.
31867 (RTL Tests): New node.
31869 2017-01-27 Richard Biener <rguenther@suse.de>
31871 PR tree-optimization/79245
31872 * tree-loop-distribution.c (distribute_loop): Apply cost
31873 modeling also to detected patterns.
31875 2017-01-27 Richard Biener <rguenther@suse.de>
31877 PR tree-optimization/71433
31878 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31879 (compare_assert_loc): New function.
31880 (process_assert_insertions): Sort and optimize assert locations
31881 to remove duplicates and push down identical assertions on
31882 edges to their destination block.
31884 2017-01-27 Richard Biener <rguenther@suse.de>
31886 PR tree-optimization/79244
31887 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31888 out SSA names even if abnormal.
31890 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31892 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31893 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31894 instead of MPFR_RNDN.
31896 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31899 * arm.c (arm_option_override): Don't call build_target_option_node
31900 until after doing all option overrides.
31901 (arm_valid_target_attribute_tree): Likewise.
31903 2017-01-27 Martin Liska <mliska@suse.cz>
31905 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31906 for {cd}tors and C++ {cd}tors.
31908 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31910 * config/s390/s390.md ("*setmem_long_and")
31911 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31913 2017-01-26 Martin Sebor <msebor@redhat.com>
31915 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31918 2017-01-26 Martin Sebor <msebor@redhat.com>
31920 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31921 HAVE_DFmode before using XFmode or DFmode.
31922 (parse_directive): Avoid using the z length modifier to avoid
31923 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31925 PR middle-end/78703
31926 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31927 to accept adjustment as an array.
31928 (get_int_range): New function.
31929 (struct directive): Make width and prec arrays.
31930 (directive::set_width, directive::set_precision): Call get_int_range.
31931 (format_integer, format_floating): Handle width and precision ranges.
31932 (format_string, parse_directive): Same.
31934 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31937 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31938 comdat_type_p set, just clone them, but keep the children in the
31942 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31943 which have direct callers with -fvar-tracking-assignments enabled
31945 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31946 inside of type units.
31948 2017-01-26 Martin Sebor <msebor@redhat.com>
31950 PR middle-end/78703
31951 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31953 (struct format_result): Replace number_chars, number_chars_min,
31954 and number_chars_max with a single member of struct result_range.
31956 (format_result::operator+=): Adjust.
31957 (struct fmtresult): Remove bounded. Handle likely and unlikely
31959 (fmtresult::adjust_for_width_or_precision): New function.
31960 (fmtresult:type_max_digits): New function.
31961 (bytes_remaining): Handle likely and unlikely counters.
31962 (min_bytes_remaining): Remove.
31963 (format_percent): Simplify.
31964 (format_integer, format_floating): Set likely and unlikely counters.
31965 (get_string_length, format_character, format_string): Same.
31966 (format_plain, should_warn_p): New function.
31967 (maybe_warn): Call should_warn_p. Update diagnostic messages
31968 and handle those for all directives, including plain strings.
31969 (format_directive): Handle likely and unlikely counters.
31970 Remove unnecessary quoting from diagnostics. Add an informational
31972 (add_bytes): Remove.
31973 (pass_sprintf_length::compute_format_length): Simplify.
31974 (try_substitute_return_value): Handle likely and unlikely counters.
31976 2017-01-26 Carl Love <cel@us.ibm.com>
31978 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31979 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31981 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31984 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31985 endianess for subregs into account.
31986 * lra-constraints.c (lra_constraints): Do risky transformations
31987 always on the first iteration.
31988 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31989 last_call_used_reg_set.
31990 (process_bb_lives): Define and use last_call_used_reg_set.
31991 * lra.c (lra): Always continue after lra_constraints on the first
31994 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31996 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31998 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32000 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32002 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32003 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32004 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32005 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32006 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32007 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32008 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32009 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32010 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32012 2017-01-26 Marek Polacek <polacek@redhat.com>
32015 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32016 for the third operand.
32018 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32020 PR middle-end/79236
32021 * omp-low.c (struct omp_context): Add simt_stmt field.
32022 (scan_omp_for): Return omp_context *.
32023 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32024 context to the _simt_ SIMD stmt.
32025 (lower_omp_for): For combined SIMD with sibling _simt_
32026 SIMD, make sure to use the same decls in _looptemp_
32027 clauses as in the sibling.
32029 2017-01-26 David Sherwood <david.sherwood@arm.com>
32031 PR middle-end/79212
32032 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32035 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32038 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32039 emit fld b; fld a; if possible.
32041 * brig-builtins.def: Update copyright years.
32042 * config/arm/arm_acle_builtins.def: Update copyright years.
32044 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32047 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32048 constraint instead of o for the stxsd instruction.
32050 2017-01-25 Carl Love <cel@us.ibm.com>
32052 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32053 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32055 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32057 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32059 2017-01-25 Richard Biener <rguenther@suse.de>
32061 PR tree-optimization/69264
32062 * target.def (vector_alignment_reachable): Improve documentation.
32063 * doc/tm.texi: Regenerate.
32064 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32066 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32067 earlier changes with respect to TYPE_USER_ALIGN.
32068 (vector_alignment_reachable_p): Likewise. Improve dumping.
32070 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32073 * config/arm/arm.md (xordi3): Force constant operand into a register
32076 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32078 * doc/invoke.texi (-fstore-merging): Correct default optimization
32079 levels at which it is enabled.
32080 (-O): Move -fstore-merging from list to...
32083 2017-01-25 Richard Biener <rguenther@suse.de>
32086 * omp-expand.c: Include debug.h.
32087 (expand_omp_taskreg): Make sure to generate early debug before
32088 outlining anything from a function.
32089 (expand_omp_target): Likewise.
32090 (grid_expand_target_grid_body): Likewise.
32092 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32095 * asan.c (get_translation_unit_decl): New function.
32096 (asan_add_global): Extract modules file name from globals
32097 TRANSLATION_UNIT_DECL name.
32099 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32102 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32103 for long calls with APCS frame and VFP.
32105 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32107 * cfg.c (original_copy_tables_initialized_p): New function.
32108 * cfg.h (original_copy_tables_initialized_p): New decl.
32109 * cfgrtl.c (relink_block_chain): Guard the call to
32110 free_original_copy_tables with a call to
32111 original_copy_tables_initialized_p.
32112 * cgraph.h (symtab_node::native_rtl_p): New decl.
32113 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32114 (symtab_node::needed_p): Don't assert for early assembly output
32115 for __RTL functions.
32116 (cgraph_node::finalize_function): Set "force_output" for __RTL
32118 (cgraph_node::analyze): Bail out early for __RTL functions.
32119 (analyze_functions): Update assertion to support __RTL functions.
32120 (cgraph_node::expand): Bail out early for __RTL functions.
32121 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32123 * function.h (struct function): Update comment for field
32125 * gimple-expr.c: Include "tree-pass.h".
32126 (gimple_has_body_p): Return false for __RTL functions.
32127 * Makefile.in (OBJS): Add run-rtl-passes.o.
32128 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32130 (gcc::pass_manager::get_clean_slate): New accessor.
32131 * passes.c: Include "insn-addr.h".
32132 (should_skip_pass_p): Add logging. Update logic for running
32133 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32134 property-provider override so it is only done for gimple passes.
32136 (skip_pass): New function.
32137 (execute_one_pass): Call skip_pass when skipping passes.
32138 * read-md.c (md_reader::read_char): Support filtering
32139 the input to a subset of line numbers.
32140 (md_reader::md_reader): Initialize fields
32141 m_first_line and m_last_line.
32142 (md_reader::read_file_fragment): New function.
32143 * read-md.h (md_reader::read_file_fragment): New decl.
32144 (md_reader::m_first_line): New field.
32145 (md_reader::m_last_line): New field.
32146 * read-rtl-function.c (function_reader::create_function): Only
32147 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32148 curr_properties. Set DECL_INITIAL to a dummy block.
32149 (read_rtl_function_body_from_file_range): New function.
32150 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32152 * run-rtl-passes.c: New file.
32153 * run-rtl-passes.h: New file.
32155 2017-01-24 Jeff Law <law@redhat.com>
32157 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32160 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32162 PR tree-optimization/79159
32163 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32164 (record_nonwrapping_iv): Improve boundary using above function if no
32165 value range information.
32167 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32168 Martin Jambor <mjambor@suse.cz>
32170 * brig-builtins.def: New file.
32171 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32172 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32173 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32174 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32175 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32176 * builtin-types.def (BT_INT8): New.
32177 (BT_INT16): Likewise.
32178 (BT_UINT8): Likewise.
32179 (BT_UINT16): Likewise.
32180 (BT_FN_ULONG): Likewise.
32181 (BT_FN_UINT_INT): Likewise.
32182 (BT_FN_UINT_ULONG): Likewise.
32183 (BT_FN_UINT_LONG): Likewise.
32184 (BT_FN_UINT_PTR): Likewise.
32185 (BT_FN_ULONG_PTR): Likewise.
32186 (BT_FN_INT8_FLOAT): Likewise.
32187 (BT_FN_INT16_FLOAT): Likewise.
32188 (BT_FN_UINT32_FLOAT): Likewise.
32189 (BT_FN_UINT16_FLOAT): Likewise.
32190 (BT_FN_UINT8_FLOAT): Likewise.
32191 (BT_FN_UINT64_FLOAT): Likewise.
32192 (BT_FN_UINT16_UINT32): Likewise.
32193 (BT_FN_UINT32_UINT16): Likewise.
32194 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32195 (BT_FN_INT_PTR_INT): Likewise.
32196 (BT_FN_UINT_PTR_UINT): Likewise.
32197 (BT_FN_LONG_PTR_LONG): Likewise.
32198 (BT_FN_ULONG_PTR_ULONG): Likewise.
32199 (BT_FN_VOID_UINT64_UINT64): Likewise.
32200 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32201 (BT_FN_INT8_INT8_INT8): Likewise.
32202 (BT_FN_INT16_INT16_INT16): Likewise.
32203 (BT_FN_INT_INT_INT): Likewise.
32204 (BT_FN_UINT_FLOAT_UINT): Likewise.
32205 (BT_FN_FLOAT_UINT_UINT): Likewise.
32206 (BT_FN_ULONG_UINT_UINT): Likewise.
32207 (BT_FN_ULONG_UINT_PTR): Likewise.
32208 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32209 (BT_FN_UINT_UINT_UINT): Likewise.
32210 (BT_FN_VOID_UINT_PTR): Likewise.
32211 (BT_FN_UINT_UINT_PTR: Likewise.
32212 (BT_FN_UINT32_UINT64_PTR): Likewise.
32213 (BT_FN_INT_INT_UINT_UINT): Likewise.
32214 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32215 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32216 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32217 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32218 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32219 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32220 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32221 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32222 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32223 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32224 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32225 * doc/frontends.texi: List BRIG FE.
32226 * doc/install.texi (Testing): Add BRIG tesring requirements.
32227 * doc/invoke.texi (Overall Options): Mention BRIG.
32228 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32230 2017-01-24 Richard Biener <rguenther@suse.de>
32232 PR translation/79208
32233 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32235 2017-01-24 Martin Jambor <mjambor@suse.cz>
32238 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32239 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32240 and known_contexts.
32242 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32244 PR middle-end/79123
32245 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32246 casts from signed to unsigned really don't have a range.
32248 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32250 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32251 GMP_RNDx for compatiblity.
32253 2017-01-24 Martin Liska <mliska@suse.cz>
32256 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32257 that would prevent us to call alloca with -1 as argument.
32259 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32261 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32262 Avoid -Wformat-security warning.
32264 2017-01-23 Andrew Pinski <apinski@cavium.com>
32266 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32269 2017-01-23 Martin Sebor <msebor@redhat.com>
32271 PR middle-end/78703
32272 * gimple-ssa-sprintf.c (warn_level): New global.
32273 (format_integer): Use it here and throughout the rest of the file.
32274 Use the same switch to compute sign as base.
32275 (maybe_warn): New function.
32276 (format_directive): Factor out warnings into maybe_warn.
32277 Add debugging output. Use warn_level.
32278 (add_bytes): Use warn_level.
32279 (pass_sprintf_length::compute_format_length): Add debugging output.
32280 (try_substitute_return_value): Same.
32281 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32283 PR middle-end/78703
32284 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32285 (struct fmtresult, format_integer, format_floating): Adjust.
32286 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32287 (get_string_length, format_string,format_directive): Same.
32288 (pass_sprintf_length::compute_format_length): Same.
32289 (try_substitute_return_value): Simplify slightly.
32291 PR middle-end/78703
32292 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32293 (fmtresult::operator+=): Outlined.
32294 (struct fmtresult): Add ctors.
32295 (struct conversion_spec): Rename...
32296 (struct directive): ...to this. Add and remove data members.
32297 (directive::set_width, directive::set_precision): New functions.
32298 (format_percent): Use fmtresult ctor.
32299 (get_width_and_precision): Remove.
32300 (format_integer): Make naming changes. Avoid computing width and
32302 (format_floating): Same. Adjust indentation.
32303 (format_character, format_none): New functions.
32304 (format_string): Moved character handling to format_character.
32305 (format_directive): Remove arguments, change return type.
32306 (parse_directive): New function.
32307 (pass_sprintf_length::compute_format_length): Move directive
32308 parsing to parse_directive.
32310 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32312 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32313 (assign_assembler_name_if_needed): ... this.
32314 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32315 (assign_assembler_name_if_needed): ... this.
32316 (free_lang_data_in_cgraph): Adjust callers.
32317 * cgraphunit.c (cgraph_node::analyze): Likewise.
32318 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32321 2017-01-23 Richard Biener <rguenther@suse.de>
32323 PR tree-optimization/79088
32324 PR tree-optimization/79188
32325 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32326 resetting loop bounds after last path deletion. Reset loop
32327 bounds of the target loop, make code match the comments.
32328 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32329 Make sure loops need no fixups.
32331 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32333 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32334 exponent support with double type for first argument.
32335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32336 type returned by __builtin_vec_extract_sig,
32337 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32338 functions from "vector int" to "vector unsigned int" or from
32339 "vector long long int" to "vector unsigned long long int".
32340 Changed type returned by __builtin_vec_extract_exp,
32341 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32342 functions from "vector int" to "vector unsigned int" or from
32343 "vector long long int" to "vector unsigned long long int".
32344 Changed return type of __builtin_vec_test_data_class,
32345 __builtin_vec_test_data_class_sp, and
32346 __builtin_vec_test_data_class_dp from "vector int" to
32347 "vector bool int" or from "vector long long int" to "vector bool
32348 long long int" and changed second argument type from "unsigned
32349 int" to "int". Added new overloaded function forms "vector float
32350 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32351 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32352 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32353 double, vector unsigned long long int)" and "vector double
32354 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32355 long int)". Changed return type of
32356 __builtin_scalar_test_data_class and
32357 __builtin_scalar_test_data_class_sp and
32358 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32359 int" and changed second argument from "unsigned int" to "int".
32360 Changed type returned by __builtin_scalar_test_neg,
32361 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32362 from "int" to "bool int". Added new overloaded function form
32363 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32364 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32365 exponent double-precision with floating point first argument.
32366 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32367 documentation of scalar_test_data_class, scalar_test_neg,
32368 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32369 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32370 vec_test_data_class built-in functions to reflect refinements in
32371 their type signatures.
32373 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32375 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32377 (aarch64_elf_asm_destructor): Likewise.
32379 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32381 PR rtl-optimization/78634
32382 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32383 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32384 * ifcvt.c (noce_try_cmove): Add missing cost check.
32386 PR rtl-optimization/71724
32387 * combine.c (if_then_else_cond): Look for situations where it is
32388 beneficial to undo the work of one of the recursive calls.
32390 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32392 PR tree-optimization/70754
32393 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32394 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32395 combined stmt before it if not NULL.
32396 (combine_chains): Process refs reversely and compute dominance point
32399 2017-01-23 Martin Liska <mliska@suse.cz>
32401 PR tree-optimization/79196
32402 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32403 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32405 (strlen_optimize_stmt): Call the renamed function.
32407 2017-01-23 Michael Matz <matz@suse.de>
32409 PR tree-optimization/78384
32410 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32412 2017-01-23 Richard Biener <rguenther@suse.de>
32414 PR tree-optimization/79186
32415 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32416 both incoming edges before moving an assert.
32418 2017-01-23 Martin Jambor <mjambor@suse.cz>
32420 * ipa-prop.c (load_from_param_1): Removed.
32421 (load_from_unmodified_param): Bits from load_from_param_1 put back
32423 (load_from_param): Removed.
32424 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32425 with stmt. Reverted back to use of load_from_unmodified_param.
32427 2017-01-23 Martin Jambor <mjambor@suse.cz>
32430 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32431 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32432 field a pointer to garbage collected vector, mark lattices and
32433 ipcp_orig_node with GTY((skip)).
32434 (ipa_get_param_count): Adjust to descriptors being a pointer.
32435 (ipa_get_param): Likewise.
32436 (ipa_get_type): Likewise.
32437 (ipa_get_param_move_cost): Likewise.
32438 (ipa_set_param_used): Likewise.
32439 (ipa_get_controlled_uses): Likewise.
32440 (ipa_set_controlled_uses): Likewise.
32441 (ipa_is_param_used): Likewise.
32442 (ipa_node_params_t): Move into garbage collector. New methods insert
32444 (ipa_node_params_sum): Annotate wth GTY(()).
32445 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32447 (ipa_load_from_parm_agg): Adjust declaration.
32448 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32449 * ipa-profile.c (ipa_profile): Likewise.
32450 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32451 (ipa_populate_param_decls): Make descriptors parameter garbage
32453 (ipa_dump_param): Adjust to descriptors being a pointer.
32454 (ipa_alloc_node_params): Likewise.
32455 (ipa_initialize_node_params): Likewise.
32456 (load_from_param_1): Make descriptors parameter garbage collected.
32457 (load_from_unmodified_param): Likewise.
32458 (load_from_param): Likewise.
32459 (ipa_load_from_parm_agg): Likewise.
32460 (ipa_node_params::~ipa_node_params): Removed.
32461 (ipa_free_all_node_params): Remove call to delete operator.
32462 (ipa_node_params_t::insert): New.
32463 (ipa_node_params_t::remove): Likewise.
32464 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32465 copy known_csts and known_contexts vectors.
32466 (ipa_read_node_info): Adjust to descriptors being a pointer.
32467 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32469 (ipcp_transform_function): Make descriptors variable garbage
32472 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32474 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32475 * config/i386/avx512dqintrin.h: Ditto.
32476 * config/i386/avx512fintrin.h: Ditto.
32477 * config/i386/i386.c: Handle new builtins.
32478 * config/i386/i386-builtin.def: Add new builtins.
32479 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32480 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32482 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32483 Martin Liska <mliska@suse.cz>
32485 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32486 * asan.c (asan_expand_poison_ifn): Support stores and use
32487 appropriate ASAN report function.
32488 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32489 * internal-fn.def (ASAN_POISON_USE): Declare.
32490 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32491 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32492 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32493 ASAN_POISON calls w/o LHS.
32494 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32495 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32496 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32497 * gimplify.c (asan_poison_variables): Add attribute
32498 use_after_scope_memory to variables that really needs to live
32500 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32501 having the attribute.
32503 2017-01-23 Martin Liska <mliska@suse.cz>
32505 * asan.c (create_asan_shadow_var): New function.
32506 (asan_expand_poison_ifn): Likewise.
32507 * asan.h (asan_expand_poison_ifn): New declaration.
32508 * internal-fn.c (expand_ASAN_POISON): Likewise.
32509 * internal-fn.def (ASAN_POISON): New builtin.
32510 * sanopt.c (pass_sanopt::execute): Expand
32511 asan_expand_poison_ifn.
32512 * tree-inline.c (copy_decl_for_dup_finish): Make function
32514 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32515 * tree-ssa.c (is_asan_mark_p): New function.
32516 (execute_update_addresses_taken): Rewrite local variables
32517 (identified just by use-after-scope as addressable) into SSA.
32519 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32521 * doc/install.texi (Specific): opensource.apple.com uses https
32522 now. Remove trailing slash.
32524 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32526 * README.Portability: Remove note on an Irix compatibility issue.
32528 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32530 * gcov.c (INCLUDE_ALGORITHM): Define.
32531 (INCLUDE_VECTOR): Define.
32532 No longer include <vector> and <algorithm> directly.
32534 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32536 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32538 * doc/invoke.texi (Code Gen Options): Ditto.
32540 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32543 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32545 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32547 rtl-optimization/79125
32548 * cprop.c (local_cprop_pass): Handle cases where we make an
32549 unconditional trap.
32551 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32555 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32556 read from, for big endian.
32558 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32560 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32561 register pauth builtins for LP64 only.
32563 2017-01-20 Marek Polacek <polacek@redhat.com>
32566 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32569 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32571 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32573 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32574 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32575 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32577 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32580 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32581 in big-endian mode when they are not a single duplicated value.
32583 2017-01-20 Richard Biener <rguenther@suse.de>
32585 * BASE-VER: Bump to 7.0.1.
32587 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32589 * omp-low.c (omplow_simd_context): New struct. Use it...
32590 (lower_rec_simd_input_clauses): ...here and...
32591 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32592 references to idx, lane, max_vf, is_simt.
32594 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32596 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32599 2017-01-20 Martin Jambor <mjambor@suse.cz>
32601 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32602 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32604 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32605 (GTFILES): Rename hsa.c to hsa-common.c.
32606 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32607 * hsa-dump.c: Likewise.
32608 * hsa-gen.c: Likewise.
32609 * hsa-regalloc.c: Likewise.
32610 * ipa-hsa.c: Likewise.
32611 * omp-expand.c: Likewise.
32612 * omp-low.c: Likewise.
32613 * toplev.c: Likewise.
32615 2017-01-20 Marek Polacek <polacek@redhat.com>
32618 * doc/invoke.texi: Document -Wduplicated-branches.
32619 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32620 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32621 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32622 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32623 return 0 only when not OEP_LEXICOGRAPHIC.
32624 (fold_build_cleanup_point_expr): Use the expression
32625 location when building CLEANUP_POINT_EXPR.
32626 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32627 * tree.c (add_expr): Handle error_mark_node.
32629 2017-01-20 Martin Liska <mliska@suse.cz>
32632 * tree-profile.c (init_ic_make_global_vars): Do not call
32634 (gimple_init_gcov_profiler): Likewise.
32636 2017-01-20 Martin Liska <mliska@suse.cz>
32639 * cgraph.h (maybe_create_reference): Remove argument and
32641 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32643 * ipa-cp.c (create_specialized_node): Likewise.
32644 * symtab.c (symtab_node::maybe_create_reference): Handle
32645 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32647 2017-01-20 Martin Liska <mliska@suse.cz>
32649 * read-rtl-function.c (function_reader::create_function): Use
32650 build_decl instread of build_decl_stat.
32652 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32654 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32655 * config/i386/avx512dqintrin.h: Ditto.
32656 * config/i386/avx512fintrin.h: Ditto.
32657 * config/i386/i386-builtin-types.def: Add new types.
32658 * config/i386/i386.c: Handle new types.
32659 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32660 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32661 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32662 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32663 (__builtin_ia32_kshiftridi): New.
32664 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32666 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32670 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32671 define to rs6000_init_stack_protect_guard.
32672 (rs6000_init_stack_protect_guard): New function.
32674 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32675 Yunqiang Su <yunqiang.su@imgtec.com>
32677 * config.gcc (supported_defaults): Add madd4.
32678 (with_madd4): Add validation.
32679 (all_defaults): Add madd4.
32680 * config/mips/mips.opt (mmadd4): New option.
32681 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32683 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32685 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32686 (ISA_HAS_FUSED_MADD4): Likewise.
32687 * doc/invoke.texi (-mmadd4): Document the new option.
32688 * doc/install.texi (--with-madd4): Document the new option.
32690 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32692 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32693 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32694 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32695 (aarch64_init_pauth_hint_builtins): New.
32696 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32697 (aarch64_expand_builtin): Expand new builtins.
32699 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32701 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32702 * combine-stack-adj.c (no_unhandled_cfa): Handle
32703 REG_CFA_TOGGLE_RA_MANGLE.
32704 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32705 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32706 info for return address signing.
32707 (aarch64_expand_epilogue): Likewise.
32709 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32711 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32712 * config/aarch64/aarch64-protos.h
32713 (aarch64_return_address_signing_enabled): New declaration.
32714 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32716 (aarch64_expand_prologue): Sign return address before it's pushed onto
32718 (aarch64_expand_epilogue): Authenticate return address fetched from
32720 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32721 (aarch64_attributes): New function attributes for "sign-return-address".
32722 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32723 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32724 ("*do_return"): Generate combined instructions according to key index.
32725 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32726 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32728 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32729 * config/aarch64/aarch64.opt (msign-return-address=): New.
32730 * doc/extend.texi (AArch64 Function Attributes): Documents
32731 "sign-return-address=".
32732 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32734 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32736 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32737 overall option summary.
32739 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32741 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32742 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32743 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32744 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32746 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32748 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32749 -mpower9-minmax by default for -mcpu=power9.
32750 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32751 128-bit floating point.
32753 2017-01-20 Alan Modra <amodra@gmail.com>
32755 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32756 optimizing for size.
32758 2017-01-20 Alan Modra <amodra@gmail.com>
32761 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32762 for strcmp and strncmp from corresponding builtin decl.
32764 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32766 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32767 instead of i386/rtems-64.h.
32768 * config/i386/rtems-64.h: Remove.
32770 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32774 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32776 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32778 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32780 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32781 Change int to HOST_WIDE_INT.
32782 * config/aarch64/aarch64-protos.h
32783 (aarch64_simd_gen_const_vector_dup): Likewise.
32784 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32786 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32788 * langhooks-def.h (lhd_type_for_size): New decl.
32789 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32790 * langhooks.c (lhd_type_for_size): New function, taken from
32793 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32795 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32796 define_bypass for CR latency.
32797 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32798 (power9-alu2): Add define_bypass for CR latency.
32800 (power9-mul): Update insn latency.
32801 (power9-mul-compare): Update insn latency, bypass latency and remove
32804 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32806 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32808 * config/aarch64/aarch64.md
32809 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32810 aarch64_nopcrelative_literal_loads.
32811 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32813 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32815 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32816 TARGET_LOONGSON_3A.
32817 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32819 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32822 * config.gcc (supported_defaults): Add lxc1-sxc1.
32823 (with_lxc1_sxc1): Add validation.
32824 (all_defaults): Add lxc1-sxc1.
32825 * config/mips/mips.opt (mlxc1-sxc1): New option.
32826 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32828 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32829 __mips_no_lxc1_sxc1.
32830 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32831 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32832 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32834 2017-01-19 Richard Biener <rguenther@suse.de>
32836 PR tree-optimization/72488
32837 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32838 sure to restore SSA info.
32839 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32841 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32843 PR rtl-optimization/79121
32844 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32845 of the inner type when shifting an extended value.
32847 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32850 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32851 interposable aliases.
32853 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32856 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32857 Use the evmergelohi instruction.
32858 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32859 (mov_sitf_e500_subreg8_2_be): Likewise.
32860 (mov_sitf_e500_subreg12_2_le): Likewise.
32861 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32862 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32863 (mov_sitf_e500_subreg8_2_le): Likewise.
32864 (mov_sitf_e500_subreg12_2_be): Likewise.
32866 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32868 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32869 attribute from vecsimple to vecperm.
32870 (altivec_vbpermq2): Likewise.
32872 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32875 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32877 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32878 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32879 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32880 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32881 case where N arg is SIZE_MAX.
32882 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32883 (cmpstrsi): Add pattern.
32885 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32888 __builtin_vec_revb builtins.
32889 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32890 built-in functions to support generation of the ISA 3.0 XXBR<x>
32891 vector byte reverse instructions.
32892 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32893 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32894 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32895 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32896 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32897 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32898 (P9V_BUILTIN_VEC_REVB): Likewise.
32899 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32900 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32901 (p9_xxbrq_v16qi): Likewise.
32902 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32903 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32904 (p9_xxbrh_v8hi): Likewise.
32905 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32906 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32907 vec_revb built-in functions.
32909 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32911 PR rtl-optimization/78952
32912 * config/i386/i386.md (any_extract): New code iterator.
32913 (*insvqi_2): Use any_extract for source operand.
32914 (*insvqi_3): Use any_shiftrt for source operand.
32916 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32918 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32920 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32922 2017-01-18 Matthias Klose <doko@ubuntu.com>
32924 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32926 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32928 * config/rs6000/altivec.h (vec_bperm): Change #define.
32929 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32930 (altivec_vbpermq2): New define_insn.
32931 (altivec_vbpermd): Likewise.
32932 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32933 function interface.
32934 (VBPERMD): Likewise.
32935 (VBPERM): New polymorphic function interface.
32936 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32937 Add entries for P9V_BUILTIN_VEC_VBPERM.
32938 * doc/extend.texi: Add interfaces for vec_bperm.
32940 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32942 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32943 first letter of error messages.
32944 (s390_resolve_overloaded_builtin): Likewise.
32945 * config/s390/s390.c (s390_expand_builtin): Likewise.
32946 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32947 (s390_valid_target_attribute_inner_p): Likewise.
32948 * config/s390/s390.md ("tabort"): Likewise.
32950 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32952 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32953 (ISA_AVOID_DIV_HILO): New macro.
32954 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32955 (ISA_HAS_DDIV): Likewise.
32957 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32959 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32961 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32963 * doc/invoke.texi (fabi-version): Spelling fix.
32965 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32968 * doc/invoke.texi (fabi-version): Mention mangling fix for
32971 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32974 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32976 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32979 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32982 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32984 (stack_protect_test): Ditto.
32985 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32986 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32988 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32989 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32990 -mstack-protector-guard-offset=.
32991 (RS/6000 and PowerPC Options): Ditto.
32993 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32995 * config/i386/i386.h (MASK_CLASS_P): New define.
32996 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32997 there are no registers from different register sets also when
32998 mask registers are used. Update function comment.
32999 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33000 to (*k/*r) and (*k/*km) alternatives.
33002 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33004 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33005 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33006 (EH_RETURN_HANDLER_RTX): New define.
33007 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33008 Force frame pointer in EH return functions.
33009 (aarch64_expand_epilogue): Add barrier for eh_return.
33010 (aarch64_final_eh_return_addr): Remove.
33011 (aarch64_eh_return_handler_rtx): New function.
33012 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33014 (aarch64_eh_return_handler_rtx): New prototype.
33016 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33018 * config/rs6000/altivec.h (vec_rlmi): New #define.
33019 (vec_vrlnm): Likewise.
33020 (vec_rlnm): Likewise.
33021 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33022 (UNSPEC_VRLNM): Likewise.
33023 (VIlong): New mode iterator.
33024 (altivec_vrl<VI_char>mi): New define_insn.
33025 (altivec_vrl<VI_char>nm): Likewise.
33026 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33028 (VRLDNM): Likewise.
33029 (RLNM): New polymorphic function entry.
33030 (VRLWMI): New monomorphic function entry.
33031 (VRLDMI): Likewise.
33032 (RLMI): New polymorphic function entry.
33033 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33034 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33035 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33038 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33041 * dwarf2out.c (field_byte_offset): Restore the
33042 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33043 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33044 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33047 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33050 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33052 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33055 * dwarf2out.c (add_data_member_location_attribute): For constant
33056 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33057 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33058 DW_AT_byte_size attributes.
33060 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33062 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33063 after forcing to constant memory when the code model is medium.
33065 2017-01-17 Julia Koval <julia.koval@intel.com>
33068 * config/i386/avx512fintrin.h
33069 (_mm512_i32gather_ps): Change __addr type to void const*.
33070 (_mm512_mask_i32gather_ps): Ditto.
33071 (_mm512_i32gather_pd): Ditto.
33072 (_mm512_mask_i32gather_pd): Ditto.
33073 (_mm512_i64gather_ps): Ditto.
33074 (_mm512_mask_i64gather_ps): Ditto.
33075 (_mm512_i64gather_pd): Ditto.
33076 (_mm512_mask_i64gather_pd): Ditto.
33077 (_mm512_i32gather_epi32): Ditto.
33078 (_mm512_mask_i32gather_epi32): Ditto.
33079 (_mm512_i32gather_epi64): Ditto.
33080 (_mm512_mask_i32gather_epi64): Ditto.
33081 (_mm512_i64gather_epi32): Ditto.
33082 (_mm512_mask_i64gather_epi32): Ditto.
33083 (_mm512_i64gather_epi64): Ditto.
33084 (_mm512_mask_i64gather_epi64): Ditto.
33085 (_mm512_i32scatter_ps): Change __addr type to void*.
33086 (_mm512_mask_i32scatter_ps): Ditto.
33087 (_mm512_i32scatter_pd): Ditto.
33088 (_mm512_mask_i32scatter_pd): Ditto.
33089 (_mm512_i64scatter_ps): Ditto.
33090 (_mm512_mask_i64scatter_ps): Ditto.
33091 (_mm512_i64scatter_pd): Ditto.
33092 (_mm512_mask_i64scatter_pd): Ditto.
33093 (_mm512_i32scatter_epi32): Ditto.
33094 (_mm512_mask_i32scatter_epi32): Ditto.
33095 (_mm512_i32scatter_epi64): Ditto.
33096 (_mm512_mask_i32scatter_epi64): Ditto.
33097 (_mm512_i64scatter_epi32): Ditto.
33098 (_mm512_mask_i64scatter_epi32): Ditto.
33099 (_mm512_i64scatter_epi64): Ditto.
33100 (_mm512_mask_i64scatter_epi64): Ditto.
33101 * config/i386/avx512pfintrin.h
33102 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33103 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33104 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33105 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33106 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33107 (_mm512_prefetch_i32scatter_ps): Ditto.
33108 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33109 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33110 (_mm512_prefetch_i64scatter_pd): Ditto.
33111 (_mm512_prefetch_i64scatter_ps): Ditto.
33112 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33113 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33114 * config/i386/avx512vlintrin.h
33115 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33116 (_mm_mmask_i32gather_ps): Ditto.
33117 (_mm256_mmask_i32gather_pd): Ditto.
33118 (_mm_mmask_i32gather_pd): Ditto.
33119 (_mm256_mmask_i64gather_ps): Ditto.
33120 (_mm_mmask_i64gather_ps): Ditto.
33121 (_mm256_mmask_i64gather_pd): Ditto.
33122 (_mm_mmask_i64gather_pd): Ditto.
33123 (_mm256_mmask_i32gather_epi32): Ditto.
33124 (_mm_mmask_i32gather_epi32): Ditto.
33125 (_mm256_mmask_i32gather_epi64): Ditto.
33126 (_mm_mmask_i32gather_epi64): Ditto.
33127 (_mm256_mmask_i64gather_epi32): Ditto.
33128 (_mm_mmask_i64gather_epi32): Ditto.
33129 (_mm256_mmask_i64gather_epi64): Ditto.
33130 (_mm_mmask_i64gather_epi64): Ditto.
33131 (_mm256_i32scatter_ps): Change __addr type to void*.
33132 (_mm256_mask_i32scatter_ps): Ditto.
33133 (_mm_i32scatter_ps): Ditto.
33134 (_mm_mask_i32scatter_ps): Ditto.
33135 (_mm256_i32scatter_pd): Ditto.
33136 (_mm256_mask_i32scatter_pd): Ditto.
33137 (_mm_i32scatter_pd): Ditto.
33138 (_mm_mask_i32scatter_pd): Ditto.
33139 (_mm256_i64scatter_ps): Ditto.
33140 (_mm256_mask_i64scatter_ps): Ditto.
33141 (_mm_i64scatter_ps): Ditto.
33142 (_mm_mask_i64scatter_ps): Ditto.
33143 (_mm256_i64scatter_pd): Ditto.
33144 (_mm256_mask_i64scatter_pd): Ditto.
33145 (_mm_i64scatter_pd): Ditto.
33146 (_mm_mask_i64scatter_pd): Ditto.
33147 (_mm256_i32scatter_epi32): Ditto.
33148 (_mm256_mask_i32scatter_epi32): Ditto.
33149 (_mm_i32scatter_epi32): Ditto.
33150 (_mm_mask_i32scatter_epi32): Ditto.
33151 (_mm256_i32scatter_epi64): Ditto.
33152 (_mm256_mask_i32scatter_epi64): Ditto.
33153 (_mm_i32scatter_epi64): Ditto.
33154 (_mm_mask_i32scatter_epi64): Ditto.
33155 (_mm256_i64scatter_epi32): Ditto.
33156 (_mm256_mask_i64scatter_epi32): Ditto.
33157 (_mm_i64scatter_epi32): Ditto.
33158 (_mm_mask_i64scatter_epi32): Ditto.
33159 (_mm256_i64scatter_epi64): Ditto.
33160 (_mm256_mask_i64scatter_epi64): Ditto.
33161 (_mm_i64scatter_epi64): Ditto.
33162 (_mm_mask_i64scatter_epi64): Ditto.
33163 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33164 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33165 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33166 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33167 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33168 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33169 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33170 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33171 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33172 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33173 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33174 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33175 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33176 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33177 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33178 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33179 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33180 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33181 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33182 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33183 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33184 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33185 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33186 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33187 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33188 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33189 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33190 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33191 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33192 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33193 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33194 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33195 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33196 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33197 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33198 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33199 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33200 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33201 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33202 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33203 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33204 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33205 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33206 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33207 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33208 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33209 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33210 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33211 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33212 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33213 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33214 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33215 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33216 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33217 definitions accordingly.
33219 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33220 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33223 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33226 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33229 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33230 hard regs for inner regno.
33232 2017-01-17 Martin Liska <mliska@suse.cz>
33235 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33236 assumption and add comment.
33238 2017-01-17 Nathan Sidwell <nathan@acm.org>
33240 * ipa-visibility.c (localize_node): New function, broken out of ...
33241 (function_and_variable_visibility): ... here. Call it.
33243 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33245 PR middle-end/77445
33246 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33247 correctly set frequency of oudgoing edge.
33248 (duplicate_thread_path): Fix profile updating.
33250 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33253 * configure.ac: Add GCC_BASE_VER.
33254 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33255 version from BASE-VER file.
33256 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33257 (gcc.o): Depend on $(BASEVER).
33258 * common.opt (dumpfullversion): New option.
33259 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33260 * doc/invoke.texi: Document -dumpfullversion.
33261 * doc/install.texi: Document --with-gcc-major-version-only.
33262 * configure: Regenerated.
33264 2017-01-17 Richard Biener <rguenther@suse.de>
33266 PR tree-optimization/71433
33267 * tree-vrp.c (register_new_assert_for): Merge same asserts
33268 on all incoming edges.
33269 (process_assert_insertions_for): Handle insertions at the
33272 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33274 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33275 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33277 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33280 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33283 2017-01-17 Alan Modra <amodra@gmail.com>
33286 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33287 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33288 symbolic stack limit when pic.
33290 2017-01-16 Martin Sebor <msebor@redhat.com>
33292 PR tree-optimization/78608
33293 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33295 2017-01-16 Jeff Law <law@redhat.com>
33298 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33299 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33300 for several include directories that may be relative to sysroot.
33301 * config/i386/x-mingw32 (gplus_includedir): Define.
33302 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33303 (native_system_includedir): Likewise.
33304 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33305 override if TARGET_SYSTEM_ROOT is defined.
33306 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33308 PR tree-optimization/79090
33309 PR tree-optimization/33562
33310 PR tree-optimization/61912
33311 PR tree-optimization/77485
33312 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33313 and computed trims into the dump file.
33315 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33317 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33319 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33322 * gimplify.c (gimplify_init_constructor): If want_value and
33323 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33327 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33328 sequence. Formatting fixes.
33329 (doloop_optimize): Formatting fixes.
33332 * gcc.c (debug_level_greater_than_spec_func): New function.
33333 (static_spec_functions): Add debug-level-gt spec function.
33334 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33336 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33337 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33338 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33339 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33340 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33341 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33343 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33345 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33346 QImode fixups to general and mask registers only.
33348 2017-01-16 Carl Love <cel@us.ibm.com>
33350 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33351 for built-in functions
33352 vector signed char vec_nabs (vector signed char)
33353 vector signed short vec_nabs (vector signed short)
33354 vector signed int vec_nabs (vector signed int)
33355 vector signed long long vec_nabs (vector signed long long)
33356 vector float vec_nabs (vector float)
33357 vector double vec_nabs (vector double)
33358 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33360 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33361 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33362 * doc/extend.texi: Update the documentation file for the new built-in
33365 2017-01-16 Martin Sebor <msebor@redhat.com>
33367 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33370 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33372 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33373 UNSPEC_VSX__XXSPLTD to require special splat handling.
33375 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33378 * system.h: Poison strndup.
33380 2017-01-16 Alan Modra <amodra@gmail.com>
33383 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33386 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33388 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33390 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33392 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33393 call recog here. Assert that INSN_CODE (insn) is non-negative.
33395 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33398 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33400 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33401 in the currently scheduled RTL fragment.
33403 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33405 PR rtl-optimization/78751
33406 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33409 2017-01-14 Jeff Law <law@redhat.com>
33411 PR tree-optimization/79090
33412 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33413 variable length stores.
33414 (compute_trims): Delete dead assignment to *trim_tail.
33415 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33418 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33420 PR rtl-optimization/78626
33421 PR rtl-optimization/78727
33422 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33423 of a block, and split such blocks after everything else is finished.
33425 2017-01-14 Alan Modra <amodra@gmail.com>
33428 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33429 target legitimate_combined_insn.
33430 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33431 (rs6000_legitimate_combined_insn): New function.
33432 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33434 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33435 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33436 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33438 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33440 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33442 2017-01-13 Jeff Law <law@redhat.com>
33444 PR tree-optimization/33562
33445 PR tree-optimization/61912
33446 PR tree-optimization/77485
33447 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33449 (delete_dead_assignment): Likewise.
33450 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33451 statement to delete_dead_call and delete_dead_assignment.
33453 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33456 * substring-locations.c (format_warning_va): Strengthen case 1 so
33457 that both endpoints of the substring must be within the format
33458 range for just the substring to be printed.
33460 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33462 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33463 * config/i386/i386.c (ix86_target_string): Add missing options
33464 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33465 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33466 flags_other and ix86_target_other to flags2_other. Display unknown
33468 (ix86_valid_target_attribute_inner_p): Add missing options and
33469 reorder options by implied ISAs, as in ix86_target_string.
33471 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33473 * hash-table.h (hash_table::too_empty_p): New function.
33474 (hash_table::expand): Use it.
33475 (hash_table::traverse): Likewise.
33476 (hash_table::empty_slot): Use sizeof (value_type) instead of
33477 sizeof (PTR) to convert bytes to elements. Shrink the table
33478 if the current size is excessive for the current number of
33481 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33483 * ira-costs.c (record_reg_classes): Break from the inner loop
33484 early once alt_fail is known to be true. Update outer loop
33485 handling accordingly.
33487 2017-01-13 Jeff Law <law@redhat.com>
33489 * tree-ssa-dse.c (decrement_count): New function.
33490 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33491 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33492 when we know the partially dead statement is a mem* function.
33494 PR tree-optimization/61912
33495 PR tree-optimization/77485
33496 * tree-ssa-dse.c: Include expr.h.
33497 (maybe_trim_constructor_store): New function.
33498 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33500 PR tree-optimization/33562
33501 PR tree-optimization/61912
33502 PR tree-optimization/77485
33503 * doc/invoke.texi: Document new dse-max-object-size param.
33504 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33505 * tree-ssa-dse.c: Include params.h.
33506 (dse_store_status): New enum.
33507 (initialize_ao_ref_for_dse): New, partially extracted from
33509 (valid_ao_ref_for_dse, normalize_ref): New.
33510 (setup_live_bytes_from_ref, compute_trims): Likewise.
33511 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33512 (maybe_trim_partially_dead_store): Likewise.
33513 (maybe_trim_complex_store): Likewise.
33514 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33515 Track what bytes live from the original store. Return tri-state
33516 for dead, partially dead or live.
33517 (dse_dom_walker): Add constructor, destructor and new private members.
33518 (delete_dead_call, delete_dead_assignment): New extracted from
33520 (dse_optimize_stmt): Make a member of dse_dom_walker.
33521 Use initialize_ao_ref_for_dse.
33523 PR tree-optimization/33562
33524 PR tree-optimization/61912
33525 PR tree-optimization/77485
33526 * sbitmap.h (bitmap_count_bits): Prototype.
33527 (bitmap_clear_range, bitmap_set_range): Likewise.
33528 * sbitmap.c (bitmap_clear_range): New function.
33529 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33531 2017-01-13 Martin Liska <mliska@suse.cz>
33534 * function.c (set_cfun): Add new argument force.
33535 * function.h (set_cfun): Likewise.
33536 * ipa-inline-transform.c (inline_call): Use the function when
33537 strict alising from is dropped for function we inline to.
33539 2017-01-13 Richard Biener <rguenther@suse.de>
33541 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33542 for dumping GIMPLE INTEGER_CSTs.
33544 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33546 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33547 to 201112L since C++17.
33549 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33552 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33553 if -fsanitize=kernel-address is present.
33555 2017-01-13 Richard Biener <rguenther@suse.de>
33557 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33558 as _Literal ( type ) number in case usual suffixes do not
33559 preserve all information.
33561 2017-01-13 Richard Biener <rguenther@suse.de>
33563 PR tree-optimization/77283
33564 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33565 and ssa-iterators.h.
33566 (is_feasible_trace): Implement a cost model based on joiner
33569 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33572 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33573 char or short to __float128/_Float128 directly.
33575 2017-01-12 Martin Sebor <msebor@redhat.com>
33577 to -Wformat-overflow.
33578 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33579 (min_bytes_remaining): Same.
33580 (get_string_length): Same.
33581 (format_string): Same.
33582 (format_directive): Same.
33584 (pass_sprintf_length::handle_gimple_call): Same.
33586 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33588 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33589 info.nowrite calls with no lhs that can't throw. Return bool
33590 whether gsi_remove has been called or not.
33591 (pass_sprintf_length::handle_gimple_call): Return bool whether
33592 try_substitute_return_value called gsi_remove. Formatting fix.
33593 (pass_sprintf_length::execute): Don't use gsi_remove if
33594 handle_gimple_call returned true.
33597 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33598 be removed due to side-effects, don't remove following barrier nor
33599 turn the successor edge into fallthru edge.
33601 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33604 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33605 element-reversing loads and stores as not swappable.
33607 2017-01-12 Nathan Sidwell <nathan@acm.org>
33608 Nicolai Stange <nicstange@gmail.com>
33610 * combine.c (try_combine): Don't ignore result of overlap checking
33611 loop. Combine overlap & asm check into single loop.
33613 2017-01-12 Richard Biener <rguenther@suse.de>
33615 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33616 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33618 2017-01-12 Richard Biener <rguenther@suse.de>
33620 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33621 and TS_TARGET_OPTION directly derive from TS_BASE.
33622 * tree-core.h (tree_optimization_option): Derive from tree_base.
33623 (tree_target_option): Likewise.
33625 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33627 * config/i386/i386.c (memory_address_length): Increase len
33628 only when rip_relative_addr_p returns false.
33630 2017-01-11 Julia Koval <julia.koval@intel.com>
33632 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33633 (OPTION_MASK_ISA_SGX_SET): New.
33634 (ix86_handle_option): Handle OPT_msgx.
33635 * config.gcc: Added sgxintrin.h.
33636 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33637 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33638 * config/i386/i386.c (ix86_target_string): Add -msgx.
33640 (ix86_option_override_internal): Handle new options.
33641 (ix86_valid_target_attribute_inner_p): Add sgx.
33642 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33643 * config/i386/i386.opt: Add msgx.
33644 * config/i386/sgxintrin.h: New file.
33645 * config/i386/x86intrin.h: Add sgxintrin.h.
33647 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33650 * fold-const.c (maybe_nonzero_address): Return 1 for function
33652 (tree_single_nonzero_warnv_p): Don't handle function local objects
33656 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33659 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33662 * opts.c: Include "spellcheck.h"
33663 (struct string_fragment): New struct.
33664 (struct edit_distance_traits<const string_fragment &>): New
33666 (get_closest_sanitizer_option): New function.
33667 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33669 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33671 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33673 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33674 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33675 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33676 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33677 for initial die_offset if dwarf_split_debug_info.
33678 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33679 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33681 (output_skeleton_debug_sections): Formatting fix. Use
33682 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33683 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33685 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33687 * config/arm/cortex-a53.md: Add bypasses for
33688 cortex_a53_r2f_cvt.
33689 (cortex_a53_r2f): Only use for transfers.
33690 (cortex_a53_f2r): Likewise.
33691 (cortex_a53_r2f_cvt): Add reservation for conversions.
33692 (cortex_a53_f2r_cvt): Likewise.
33694 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33696 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33697 to all inlined functions, change static to extern.
33699 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33702 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33704 (arm_assemble_integer): Likewise.
33706 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33708 * config.gcc: Use new awk script to check CPU, FPU and architecture
33709 parameters for --with-... options.
33710 * config/arm/parsecpu.awk: New file
33711 * config/arm/arm-cpus.in: New file.
33712 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33714 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33716 * config/arm/t-arm: Update dependency rules.
33717 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33718 of processing .def files.
33719 * config/arm/genopt.sh: Deleted.
33720 * config/arm/gentune.sh: Deleted.
33721 * config/arm/arm-cores.def: Deleted.
33722 * config/arm/arm-arches.def: Deleted.
33723 * config/arm/arm-fpus.def: Deleted.
33724 * config/arm/arm-tune.md: Regenerated.
33725 * config/arm/arm-tables.opt: Regenerated.
33726 * config/arm/arm-cpu.h: New generated file.
33727 * config/arm/arm-cpu-data.h: New generated file.
33728 * config/arm/arm-cpu-cdata.h: New generated file.
33730 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33733 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33735 (input_varpool_node): Unpack dynamically_initialized bit.
33737 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33739 PR rtl-optimization/79032
33740 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33741 the alignment of the adjusted memory reference against that of MODE,
33742 instead of the alignment of the original memory reference.
33744 2017-01-11 Martin Jambor <mjambor@suse.cz>
33746 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33748 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33749 decorated functions.
33751 2017-01-11 Richard Biener <rguenther@suse.de>
33753 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33754 set range/nonnull info for PHI results. Do not set it on
33755 stmts marked for removal.
33757 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33759 * expr.c (store_field): In the bitfield case, fetch the return value
33760 from the registers before applying a single big-endian adjustment.
33761 Always do a final load for a BLKmode value not larger than a word.
33763 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33766 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33767 that we correctly handle column numbers greater than
33768 LINE_MAP_MAX_COLUMN_NUMBER.
33770 2017-01-10 Martin Sebor <msebor@redhat.com>
33772 PR middle-end/78245
33773 * gimple-ssa-sprintf.c (get_destination_size): Call
33774 {init,fini}object_sizes.
33775 * tree-object-size.c (addr_object_size): Adjust.
33776 (pass_through_call): Adjust.
33777 (pass_object_sizes::execute): Adjust.
33778 * tree-object-size.h (fini_object_sizes): Declare.
33780 2017-01-10 Martin Sebor <msebor@redhat.com>
33782 PR tree-optimization/78775
33783 * builtins.c (get_size_range): Move...
33784 * calls.c: ...to here.
33785 (alloc_max_size): Accept zero argument.
33786 (operand_signed_p): Remove.
33787 (maybe_warn_alloc_args_overflow): Call get_size_range.
33788 * calls.h (get_size_range): Declare.
33790 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33792 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33793 from TI's devices.csv file as of September 2016.
33794 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33796 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33798 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33799 * doc/invoke.texi: Likewise.
33800 * doc/md.texi: Likewise.
33801 * doc/objc.texi: Likewise.
33803 2017-01-10 Joshua Conner <joshconner@google.com>
33805 * config/arm/fuchsia-elf.h: New file.
33806 * config/fuchsia.h: New file.
33807 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33808 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33810 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33812 2016-01-10 Richard Biener <rguenther@suse.de>
33814 PR tree-optimization/79034
33815 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33816 Propagate out degenerate PHIs in the joiner.
33818 2017-01-10 Martin Liska <mliska@suse.cz>
33820 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33821 (sort_congruence_classes_by_decl_uid): Likewise.
33822 (sort_congruence_class_groups_by_decl_uid): Likewise.
33823 (sem_item_optimizer::merge_classes): Sort class, groups in these
33824 classes and members in the groups by DECL_UID of declarations.
33825 This would make merge operations stable.
33827 2017-01-10 Martin Liska <mliska@suse.cz>
33829 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33830 usage of m_classes_vec.
33831 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33832 (sem_item_optimizer::get_group_by_hash): Likewise.
33833 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33834 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33835 (sem_item_optimizer::verify_classes): Likewise.
33836 (sem_item_optimizer::process_cong_reduction): Likewise.
33837 (sem_item_optimizer::dump_cong_classes): Likewise.
33838 (sem_item_optimizer::merge_classes): Likewise.
33839 * ipa-icf.h (congruence_class_hash): Rename from
33840 congruence_class_group_hash. Remove declaration of m_classes_vec.
33842 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33844 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33845 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33846 * config.gcc: Add avx512vpopcntdqintrin.h.
33847 * config/i386/avx512vpopcntdqintrin.h: New.
33848 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33849 * config/i386/i386-builtin-types.def: Add new types.
33850 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33851 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33852 __builtin_ia32_vpopcountq_v8di_mask): New.
33853 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33854 __AVX512VPOPCNTDQ__.
33855 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33856 (PTA_AVX512VPOPCNTDQ): Define.
33857 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33858 TARGET_AVX512VPOPCNTDQ_P): Define.
33859 * config/i386/i386.opt: Add mavx512vpopcntdq.
33860 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33861 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33863 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33865 PR middle-end/77484
33866 * predict.def (PRED_CALL): Set to 67.
33868 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33870 * expr.c (store_field): In the bitfield case, if the value comes from
33871 a function call and is of an aggregate type returned in registers, do
33872 not modify the field mode; extract the value in all cases if the mode
33873 is BLKmode and the size is not larger than a word.
33875 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33878 * config/i386/cpuid.h: Fix undefined behavior.
33880 2017-01-04 Jeff Law <law@redhat.com>
33882 PR tree-optimization/79007
33883 PR tree-optimization/67955
33884 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33885 conservative for pt.null when flag_non_call_exceptions is on.
33887 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33889 PR translation/79019
33890 PR translation/79020
33891 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33892 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33893 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33895 * config/avr/avr.opt (maccumulate-args): Likewise.
33896 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33897 * common.opt (freport-bug): Likewise.
33898 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33899 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33900 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33901 translatable string.
33902 * config/i386/i386.c (function_value_32): Likewise.
33903 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33904 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33906 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33907 * common/config/msp430/msp430-common.c (msp430_handle_option):
33909 * symtab.c (symtab_node::verify_base): Likewise.
33910 * opts.c (set_debug_level): Likewise.
33911 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33912 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33913 missing whitespace to translatable strings.
33914 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33915 * config/sh/superh.h: Likewise.
33916 * config/i386/xopintrin.h: Likewise.
33917 * config/i386/znver1.md: Likewise.
33918 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33919 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33920 * double-int.h (struct double_int): Likewise.
33921 * double-int.c (div_and_round_double): Likewise.
33922 * wide-int.cc: Likewise.
33923 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33924 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33925 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33926 (crossjumps_occurred): ... this.
33927 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33930 PR tree-optimization/78899
33931 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33932 returning bool return struct loop *, NULL for failure and the new
33934 (versionable_outer_loop_p): Don't version outer loop if it has
33935 dont_vectorized bit set.
33936 (tree_if_conversion): When versioning outer loop, ensure
33937 tree_if_conversion is performed also on the inner loop of the
33938 non-vectorizable outer loop copy.
33939 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33940 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33941 prevent vectorization of it.
33942 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33943 the outer loop vectorization of the non-scalar version is attempted
33944 before vectorization of the inner loop in scalar version. If
33945 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33946 vectorization of its inner loop.
33947 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33948 has 2 inner loops, rename also on edges from bb whose single pred
33949 is outer_loop->header. Fix typo in function comment.
33951 2017-01-09 Martin Sebor <msebor@redhat.com>
33954 * asan.c (asan_emit_stack_protection): Increase local buffer size
33955 to avoid snprintf truncation warning.
33957 2017-01-09 Andrew Pinski <apinski@cavium.com>
33959 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33960 to reference thunderx2t99 for the tuning structure
33961 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33963 (thunderx2t99_extra_costs): This.
33964 * config/aarch64/aarch64-tune.md: Regenerate.
33965 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33966 (vulcan_addrcost_table): This.
33967 (vulcan_regmove_cost): Rename to ...
33968 (thunderx2t99_regmove_cost): This.
33969 (vulcan_vector_cost): Rename to ...
33970 (thunderx2t99_vector_cost): this.
33971 (vulcan_branch_cost): Rename to ...
33972 (thunderx2t99_branch_cost): This.
33973 (vulcan_tunings): Rename to ...
33974 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33975 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33977 2017-01-09 Martin Jambor <mjambor@suse.cz>
33981 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33982 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33983 (propagate_vr_accross_jump_function): Use the above function for all
33984 value range computations for pass-through jump functions and type
33985 converasion from explicit value range values.
33986 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33987 parameters from TYPE_ARG_TYPES.
33988 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33989 (ipa_write_node_info): Stream type of the actual argument.
33990 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33992 2017-01-09 Martin Liska <mliska@suse.cz>
33995 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33996 (lookup_compiler): Do not show error message with have_E.
33998 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34000 PR tree-optimization/78938
34001 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34002 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34003 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34004 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34007 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34009 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34012 2017-01-09 Richard Biener <rguenther@suse.de>
34014 PR tree-optimization/78997
34015 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34016 name condition properly.
34018 2017-01-09 Richard Biener <rguenther@suse.de>
34021 * dwarf2out.c (is_cxx): New overload with context.
34022 (is_naming_typedef_decl): Use it.
34024 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34026 * invoke.texi (Option Summary): Correct spacing in option lists
34027 and add line breaks to fix over-long lines.
34029 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34031 PR middle-end/17660
34033 * extend.texi (Common Variable Attributes): Add xref to GCC
34034 Internals manual to explain mode attribute keywords.
34036 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34039 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34040 and Preprocessor Options.
34041 (Options for Linking): Document -pthread here....
34042 (RS/6000 and PowerPC Options): ...not here.
34043 (Solaris 2 Options): ...or here.
34044 * doc/cppopts.texi: Document -pthread.
34046 2017-01-08 Martin Sebor <msebor@redhat.com>
34048 PR middle-end/77708
34049 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34050 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34051 New member functions.
34052 (format_directive): Used them.
34054 (pass_sprintf_length::handle_gimple_call): Same.
34055 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34056 to avoid truncation for any argument.
34057 (extract_affine_mul): Same.
34058 * tree.c (get_file_function_name): Same.
34060 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34062 PR middle-end/77484
34063 * predict.def (PRED_INDIR_CALL): Set to 86.
34065 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34067 PR preprocessor/54124
34068 * doc/cppopts.texi: Reformat -d subtable to list the full name
34069 of the options. Add cross-reference to the docs for the general
34070 compiler -d options.
34071 * doc/invoke.texi (Developer Options): Add cross-reference to the
34072 preprocessor-specific -d option documentation.
34074 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34076 PR preprocessor/13498
34077 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34078 redudant material, and reflect new command-line options.
34079 (System Headers): Likewise.
34081 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34083 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34084 -isystem, and -idirafter. Copy-edit.
34085 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34086 default for -ftrack-macro-expansion. Delete obsolete and
34087 badly-formatted implementation details about -fdebug-cpp output.
34088 * doc/cppwarnopts.texi: Copy-edit.
34090 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34093 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34094 that the transition from a max line width >= 1<<10 to narrower
34095 lines works correctly.
34097 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34099 * doc/options.texi (PerFunction): New.
34100 * opt-functions.awk (switch_flags): Map both Optimization and
34101 PerFunction to CL_OPTIMIZATION.
34102 * opth-gen.awk: Test for PerFunction flag along with
34104 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34105 it only when the latter is present. Skip those that don't in
34106 the hash function generator.
34107 * common.opt (fvar-tracking): Mark as PerFunction instead of
34109 (fvar-tracking-assignments): Likewise.
34110 (fvar-tracking-assignments-toggle): Likewise.
34111 (fvar-tracking-uninit): Likewise.
34113 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34115 PR translation/79018
34116 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34119 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34122 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34123 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34124 TARGET_LONG_JUMP_TABLE_OFFSETS.
34125 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34126 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34127 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34128 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34129 * config/m68k/m68k.md (tablejump expander): Likewise.
34130 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34131 TARGET_LONG_JUMP_TABLE_OFFSETS.
34132 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34133 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34135 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34136 David Holsgrove <david.holsgrove@xilinx.com>
34138 * common/config/microblaze/microblaze-common.c
34139 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34140 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34142 * config/microblaze/microblaze.c (microblaze_must_save_register)
34143 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34145 (microblaze_eh_return): New function.
34146 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34147 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34148 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34149 * config/microblaze/microblaze.md (eh_return): New pattern.
34151 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34153 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34154 GCC_DIAGNOSTIC_STRINGIFY): Define.
34156 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34158 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34160 * config/arm/arm.md (<mcrr>): New.
34162 * config/arm/arm.c (arm_arch5te): New.
34163 (arm_option_override): Set arm_arch5te.
34164 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34166 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34167 (arm_mcrr_qualifiers): ... this. New.
34168 (MRRC_QUALIFIERS): Define to...
34169 (arm_mrrc_qualifiers): ... this. New.
34170 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34172 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34173 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34174 (MRRCI, mrrc, MRRC): New.
34175 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34176 VUNSPEC_MRRC2): New.
34178 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34180 * config/arm/arm.md (<mcr>): New.
34182 * config/arm/arm.c (arm_coproc_builtin_available): Add
34183 support for mcr, mrc, mcr2 and mrc2.
34184 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34185 (arm_mcr_qualifiers): ... this. New.
34186 (MRC_QUALIFIERS): Define to ...
34187 (arm_mrc_qualifiers): ... this. New.
34188 (MCR_QUALIFIERS): Define to ...
34189 (arm_mcr_qualifiers): ... this. New.
34190 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34192 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34193 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34194 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34195 VUNSPEC_MRC2): New.
34197 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34199 * config/arm/arm.md (*ldc): New.
34203 * config/arm/arm.c (arm_coproc_builtin_available): Add
34204 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34205 (arm_coproc_ldc_stc_legitimate_address): New.
34206 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34207 'qualifier_const_pointer'.
34208 (LDC_QUALIFIERS): Define to...
34209 (arm_ldc_qualifiers): ... this. New.
34210 (STC_QUALIFIERS): Define to...
34211 (arm_stc_qualifiers): ... this. New.
34212 * config/arm/arm-protos.h
34213 (arm_coproc_ldc_stc_legitimate_address): New.
34214 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34215 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34216 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34217 stc2, stcl, stc2l): New.
34218 * config/arm/constraints.md (Uz): New.
34219 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34220 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34221 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34222 VUNSPEC_STC2L): New.
34224 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34226 * config/arm/arm.md (<cdp>): New.
34227 * config/arm/arm.c (neon_const_bounds): Rename this ...
34228 (arm_const_bounds): ... this.
34229 (arm_coproc_builtin_available): New.
34230 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34231 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34232 (CDP_QUALIFIERS): Define to...
34233 (arm_cdp_qualifiers): ... this. New.
34235 (arm_expand_builtin_args): Add case for 6 arguments.
34236 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34237 (arm_const_bounds): ... this.
34238 (arm_coproc_builtin_available): New.
34239 * config/arm/arm_acle.h (__arm_cdp): New.
34241 * config/arm/arm_acle_builtins.def (cdp): New.
34243 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34244 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34245 'arm_const_bounds'.
34246 * config/arm/types.md (coproc): New.
34247 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34248 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34249 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34250 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34252 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34254 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34255 (UBINOP_QUALIFIERS): New.
34257 (acle_builtin_data): New. Change comment.
34258 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34259 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34260 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34261 arm_acle_builtins.def.
34262 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34263 (arm_init_acle_builtins): New.
34264 (CRC32_BUILTIN): Remove.
34265 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34266 crc32cb, crc32ch and crc32cw.
34267 (arm_init_crc32_builtins): Remove.
34268 (arm_init_builtins): Use arm_init_acle_builtins rather
34269 than arm_init_crc32_builtins.
34270 (arm_expand_acle_builtin): New.
34271 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34272 * config/arm/arm_acle_builtins.def: New.
34274 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34276 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34277 (arm_builtin_datum): ... this.
34278 (arm_init_neon_builtin): Rename to ...
34279 (arm_init_builtin): ... this. Add a new parameters PREFIX
34280 and USE_SIG_IN_NAME.
34281 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34282 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34283 'arm_builtin_datum'.
34284 (arm_init_vfp_builtins): Likewise.
34285 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34286 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34287 (arm_expand_neon_args): Rename to ...
34288 (arm_expand_builtin_args): ... this. Rename builtin_arg
34289 enum values and differentiate between ARG_BUILTIN_MEMORY
34290 and ARG_BUILTIN_NEON_MEMORY.
34291 (arm_expand_neon_builtin_1): Rename to ...
34292 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34293 values, arm_expand_builtin_args and add bool parameter NEON.
34294 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34295 (arm_expand_vfp_builtin): Likewise.
34296 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34298 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34300 PR middle-end/77484
34301 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34302 * predict.c (tree_estimate_probability_bb): Reverse direction of
34303 polymorphic call predictor.
34305 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34307 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34308 (determine_pass_name_match): ...this new function and...
34309 (should_skip_pass_p): ...this new function.
34311 2017-01-06 Nathan Sidwell <nathan@acm.org>
34313 * ipa-visibility.c (function_and_variable_visibility): Reformat
34314 comments and long lines. Remove extrneous if.
34315 * symtab.c (symtab_node::make_decl_local): Fix code format.
34316 (symtab_node::set_section_for_node): Fix comment typo.
34318 2017-01-06 Martin Liska <mliska@suse.cz>
34321 * lra-constraints.c: Rename invariant to lra_invariant.
34322 * predict.c (set_even_probabilities): Initialize e to NULL.
34324 2017-01-05 Martin Sebor <msebor@redhat.com>
34326 PR tree-optimization/78910
34327 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34328 (format_integer): Correct off-by-one error in the handling
34329 of precision with negative numbers in signed conversions..
34331 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34333 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34335 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34337 PR tree-optimization/71016
34338 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34339 factor_out_conditional_conversion. Formatting fix.
34340 (factor_out_conditional_conversion): Add cond_stmt argument.
34341 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34342 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34345 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34347 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34348 read-rtl-function.o, and selftest-rtl.o.
34349 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34350 (selftest::aarch64_test_loading_full_dump): New function.
34351 (selftest::aarch64_run_selftests): New function.
34352 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34353 selftest::aarch64_run_selftests.
34354 * config/i386/i386.c
34355 (selftest::ix86_test_loading_dump_fragment_1): New function.
34356 (selftest::ix86_test_loading_call_insn): New function.
34357 (selftest::ix86_test_loading_full_dump): New function.
34358 (selftest::ix86_test_loading_unspec): New function.
34359 (selftest::ix86_run_selftests): Call the new functions.
34360 * emit-rtl.c (maybe_set_max_label_num): New function.
34361 * emit-rtl.h (maybe_set_max_label_num): New decl.
34362 * function.c (instantiate_decls): Guard call to
34363 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34364 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34366 * gensupport.c (gen_reader::gen_reader): Pass "false"
34367 for new "compact" param of rtx_reader.
34368 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34369 rather than an empty string for NULL strings.
34370 * read-md.c: Potentially include config.h rather than bconfig.h.
34371 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34372 (have_error): New global, copied from errors.c.
34373 (md_reader::read_name): Rename to...
34374 (md_reader::read_name_1): ...this, adding "out_loc" param,
34375 and converting "missing name or number" to returning false, rather
34377 (md_reader::read_name): Reimplement in terms of read_name_1.
34378 (md_reader::read_name_or_nil): New function.
34379 (md_reader::read_string): Handle "(nil)" by returning NULL.
34380 (md_reader::md_reader): Add new param "compact".
34381 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34382 (md_reader::read_file): New method.
34383 * read-md.h (md_reader::md_reader): Add new param "compact".
34384 (md_reader::read_file): New method.
34385 (md_reader::is_compact): New accessor.
34386 (md_reader::read_name): Convert return type from void to file_location.
34387 (md_reader::read_name_or_nil): New decl.
34388 (md_reader::read_name_1): New decl.
34389 (md_reader::m_compact): New field.
34390 (noop_reader::noop_reader): Pass "false" for new "compact" param
34392 (rtx_reader::rtx_reader): Add new "compact" param.
34393 (rtx_reader::read_rtx_operand): Make virtual and convert return
34394 type from void to rtx.
34395 (rtx_reader::read_until): New decl.
34396 (rtx_reader::handle_any_trailing_information): New virtual function.
34397 (rtx_reader::postprocess): New virtual function.
34398 (rtx_reader::finalize_string): New virtual function.
34399 (rtx_reader::m_in_call_function_usage): New field.
34400 (rtx_reader::m_reuse_rtx_by_id): New field.
34401 * read-rtl-function.c: New file.
34402 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34403 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34404 (selftest::verify_three_block_rtl_cfg): New decl.
34405 * read-rtl-function.h: New file.
34406 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34407 For host, include function.h, memmodel.h, and emit-rtl.h.
34408 (one_time_initialization): New function.
34409 (struct compact_insn_name): New struct.
34410 (compact_insn_names): New array.
34411 (find_code): Handle insn codes in compact dumps.
34412 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34413 (bind_subst_iter_and_attr): Likewise.
34414 (add_condition_to_string): Likewise.
34415 (add_condition_to_rtx): Likewise.
34416 (apply_attribute_uses): Likewise.
34417 (add_current_iterators): Likewise.
34418 (apply_iterators): Likewise.
34419 (initialize_iterators): Guard usage of apply_subst_iterator with
34420 #ifdef GENERATOR_FILE.
34421 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34422 (md_reader::read_mapping): Likewise.
34423 (add_define_attr_for_define_subst): Likewise.
34424 (add_define_subst_attr): Likewise.
34425 (read_subst_mapping): Likewise.
34426 (check_code_iterator): Likewise.
34427 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34429 (one_time_initialization): New function.
34430 (rtx_reader::read_until): New method.
34431 (read_flags): New function.
34432 (parse_reg_note_name): New function.
34433 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34434 Handle reuse_rtx ids.
34435 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34436 Add parsing support for RTL dumps, mirroring the special-cases in
34437 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34438 values, and calling handle_any_trailing_information.
34439 (rtx_reader::read_rtx_operand): Convert return type from void
34440 to rtx, returning return_rtx. Handle case 'e'. Call
34441 finalize_string on XSTR and XTMPL fields.
34442 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34443 "(nil)" values were omitted. Call the postprocess vfunc on the
34445 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34446 class ctor. Initialize m_in_call_function_usage. Call
34447 one_time_initialization.
34448 * rtl-tests.c (selftest::test_uncond_jump): Call
34449 set_new_first_and_last_insn.
34450 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34451 * selftest-rtl.c: New file.
34452 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34453 (selftest::get_insn_by_uid): New decl.
34454 * selftest-run-tests.c (selftest::run_tests): Call
34455 read_rtl_function_c_tests.
34456 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34457 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34460 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34462 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34463 operands in a special way. Assert that pos+len <= mode precision.
34465 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34467 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34468 3 argument Alias with unlimited for the negative form.
34469 (fno-vect-cost-model): Removed.
34471 2017-01-05 Martin Liska <mliska@suse.cz>
34473 * hsa-gen.c (gen_hsa_divmod): New function.
34474 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34476 2017-01-05 Martin Liska <mliska@suse.cz>
34479 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34482 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34484 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34485 small constant length operands.
34487 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34489 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34490 between loop iterations.
34492 2017-01-05 Martin Liska <mliska@suse.cz>
34495 * gimplify.c (gimplify_decl_expr): Compare to
34496 asan_poisoned_variables instread of checking flags.
34497 (gimplify_target_expr): Likewise.
34498 (gimplify_expr): Likewise.
34499 (gimplify_function_tree): Conditionally initialize
34500 asan_poisoned_variables.
34502 2017-01-04 Jeff Law <law@redhat.com>
34504 PR tree-optimizatin/78812
34505 * rtl.h (contains_mem_rtx_p): Prototype.
34506 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34507 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34508 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34509 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34510 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34512 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34514 * input.c (assert_char_at_range): Default-initialize actual_range.
34516 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34518 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34519 to match the caller.
34521 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34523 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34524 insns after final jump in test to emit dummy move.
34526 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34528 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34529 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34531 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34533 * multiple_target.c (create_dispatcher_calls): Init e_next.
34534 * tree-ssa-loop-split.c (split_loop): Init border.
34535 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34538 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34543 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34544 (altivec_register_operand): Do not return true if the operand
34545 contains a SUBREG mixing SImode and SFmode.
34546 (vsx_register_operand): Likewise.
34547 (vsx_reg_sfsubreg_ok): New predicate.
34548 (vfloat_operand): Do not return true if the operand contains a
34549 SUBREG mixing SImode and SFmode.
34550 (vint_operand): Likewise.
34551 (vlogical_operand): Likewise.
34552 (gpc_reg_operand): Likewise.
34553 (int_reg_operand): Likewise.
34554 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34555 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34556 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34558 (rs6000_emit_move_si_sf_subreg): New helper function.
34559 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34560 fixup SUBREGs involving SImode and SFmode.
34561 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34562 numbers for the new peephole2 optimization.
34563 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34564 the GLIBC math library that do AND/IOR/XOR operations on single
34565 precision floating point.
34566 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34567 target macros to say whether we need to avoid SUBREGs mixing
34569 (TARGET_ALLOW_SF_SUBREG): Likewise.
34570 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34571 (UNSPEC_SI_FROM_SF): Likewise.
34572 (iorxor): Change spacing.
34573 (and_ior_xor): New iterator for AND, IOR, and XOR.
34574 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34575 (movdi_from_sf_zero_ext): Likewise.
34576 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34577 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34578 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34579 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34580 (fms<mode>4): Likewise.
34581 (fnma<mode>4): Likewise.
34582 (fnms<mode>4): Likewise.
34583 (nfma<mode>4): Likewise.
34584 (nfms<mode>4): Likewise.
34586 2017-01-04 Marek Polacek <polacek@redhat.com>
34589 * doc/invoke.texi: Document -Wpointer-compare.
34591 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34593 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34596 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34597 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34600 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34603 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34604 documentation of the powerpc_popcntb_ok attribute.
34605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34606 code to issue warning messages if a requested CPU configuration is
34607 not supported by the binary (assembler and loader) toolchain.
34608 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34609 made to define a built-in function that has been disabled.
34610 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34611 made to define a built-in function that has been disabled.
34612 (altivec_init_builtins): Add comment explaining why definition
34613 of the DST built-in functions is not preceded by an assertion
34614 check. Add assertions to prevent ICE if attempts are made to
34615 define an altivec predicate or an abs* built-in function that has
34617 (htm_init_builtins): Add comment explaining why definition of the
34618 htm built-in functions is not preceded by an assertion check.
34620 2017-01-04 Jeff Law <law@redhat.com>
34622 PR tree-optimizatin/67955
34623 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34624 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34625 the points-to solution does not include pt_null. Use DECL_PT_UID
34628 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34630 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34631 Use gen_int_mode instead of gen_lopwart for const_int operands.
34633 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34635 PR tree-optimization/71563
34636 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34637 out of range value - has low bits known to be zero.
34639 2017-01-04 Alan Modra <amodra@gmail.com>
34641 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34642 * configure: Regenerate.
34643 * config.in: Regenerate.
34645 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34648 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34649 a substring of the message, but strcmp with the whole message. Ifdef
34650 ENABLE_NLS, translate the message first using dgettext.
34652 2017-01-03 Jeff Law <law@redhat.com>
34654 PR tree-optimizatin/78856
34655 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34656 (mark_threaded_blocks): Remove code to truncate thread paths that
34657 cross multiple loop headers. Instead invalidate the cached loop
34658 iteration information and handle case of a thread path walking
34659 into an irreducible region.
34661 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34664 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34665 assertions. Add support for doing the signbit if the IEEE 128-bit
34666 floating point value is in a GPR.
34667 * config/rs6000/rs6000.md (Fsignbit): Delete.
34668 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34669 Update the length attribute if the value is in a GPR.
34670 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34671 the sign or zero extension instruction, since the value is always 0/1.
34672 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34675 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34676 extracting SImode to a GPR register so that we can generate a
34677 store, limit the vector to be in a traditional Altivec register
34678 for the vextuwrx instruction.
34680 2017-01-03 Ian Lance Taylor <iant@google.com>
34682 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34684 2017-01-03 Martin Sebor <msebor@redhat.com>
34686 PR tree-optimization/78696
34687 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34688 precision. Use MPFR for %f for greater fidelity. Correct handling
34690 (pass_sprintf_length::compute_format_length): Set width and precision
34691 specified by asrerisk to void_node for vararg functions.
34692 (try_substitute_return_value): Adjust dump output.
34694 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34696 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34698 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34700 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34701 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34702 -mlra/-mno-lra was passed to the compiler.
34704 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34706 PR rtl-optimization/65618
34707 * emit-rtl.c (try_split): Move initialization of "before" and
34708 "after" to just before the call to emit_insn_after_setloc.
34710 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34712 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34714 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34716 * dwarf2out.c (gen_enumeration_type_die): When
34717 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34719 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34721 PR tree-optimization/78965
34722 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34723 Change first argument from const call_info & to call_info &. For %n
34724 set info.nowrite to false.
34726 PR middle-end/78901
34727 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34728 possibly throwing calls.
34730 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34731 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34732 and fns handling, rather than in a separate case SSA_NAME.
34734 2017-01-02 Jeff Law <law@redhat.com>
34736 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34737 fixes for first_period and second_period variables.
34739 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34742 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34743 (*insvqi_1): New insn pattern.
34744 (*insvqi_1_mem_rex64): Ditto.
34745 (*insvqi_2): Ditto.
34746 (*insvqi_3): Rename from *insvqi.
34748 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34750 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34752 * doc/cfg.texi (Edges): Remove reference to Java.
34753 (Maintaining the CFG): Ditto.
34755 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34757 PR middle-end/77674
34758 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34759 transparent aliases.
34761 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34763 PR middle-end/77484
34764 * predict.def (PRED_CALL): Update hitrate.
34765 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34766 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34767 into direct/indirect/polymorphic variants.
34769 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34771 Update copyright years.
34773 * gcc.c (process_command): Update copyright notice dates.
34774 * gcov-dump.c (print_version): Ditto.
34775 * gcov.c (print_version): Ditto.
34776 * gcov-tool.c (print_version): Ditto.
34777 * gengtype.c (create_file): Ditto.
34778 * doc/cpp.texi: Bump @copying's copyright year.
34779 * doc/cppinternals.texi: Ditto.
34780 * doc/gcc.texi: Ditto.
34781 * doc/gccint.texi: Ditto.
34782 * doc/gcov.texi: Ditto.
34783 * doc/install.texi: Ditto.
34784 * doc/invoke.texi: Ditto.
34786 Copyright (C) 2017 Free Software Foundation, Inc.
34788 Copying and distribution of this file, with or without modification,
34789 are permitted in any medium without royalty provided the copyright
34790 notice and this notice are preserved.