1 2017-11-13 Martin Liska <mliska@suse.cz>
3 * gcov.c (struct function_info): Remove typedef for function_t.
4 (struct source_info): Likewise.
5 (source_info::get_functions_at_location): Likewise.
6 (solve_flow_graph): Likewise.
7 (find_exception_blocks): Likewise.
8 (add_line_counts): Likewise.
9 (output_intermediate_file): Likewise.
10 (process_file): Likewise.
11 (generate_results): Likewise.
12 (release_structures): Likewise.
13 (read_graph_file): Likewise.
14 (read_count_file): Likewise.
15 (accumulate_line_counts): Likewise.
16 (output_lines): Likewise.
18 2017-11-13 Martin Liska <mliska@suse.cz>
20 * gcov.c (function_info::function_info): Remove num_counts
21 and add vector<gcov_type>.
22 (function_info::~function_info): Use the vector.
23 (process_file): Likewise.
24 (read_graph_file): Likewise.
25 (read_count_file): Likewise.
26 (solve_flow_graph): Likewise.
28 2017-11-13 Martin Liska <mliska@suse.cz>
30 * gcov.c (function_info::is_artificial): New function.
31 (process_file): Erase all artificial early.
32 (generate_results): Skip as all artificial are already
35 2017-11-13 Martin Liska <mliska@suse.cz>
37 * gcov.c (read_graph_file): Store to global vector of functions.
38 (read_count_file): Iterate the vector.
39 (process_file): Likewise.
40 (generate_results): Likewise.
41 (release_structures): Likewise.
43 2017-11-13 Jakub Jelinek <jakub@redhat.com>
45 PR tree-optimization/82954
46 * gimple-ssa-store-merging.c
47 (imm_store_chain_info::coalesce_immediate_stores): If
48 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
50 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
52 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
53 Upddate call to ENDIAN_LANE_N.
54 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
55 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
56 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
57 and use aarch64_endian_lane_rtx.
58 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
60 2017-11-12 Tom de Vries <tom@codesourcery.com>
62 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
64 2017-11-12 Tom de Vries <tom@codesourcery.com>
66 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
68 2017-11-12 Tom de Vries <tom@codesourcery.com>
70 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
72 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
73 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
74 * defaults.h (ASM_OUTPUT_LABELREF): Same.
76 2017-11-11 Martin Sebor <msebor@redhat.com>
79 * doc/extend.texi (attribute nonstring): Remove spurious argument.
82 * prefic.c (translate_name): Replace strncpy with memcpy to
83 avoid -Wstringop-truncation.
85 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
87 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
89 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
91 * predict.c (maybe_hot_frequency_p): Do not use cfun.
93 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
95 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
98 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
100 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
101 enable generating XXBRH if the value is in a vector register.
102 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
103 value is in a vector register.
104 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
105 register to register bswap64's instead of doing the GPR sequence
106 used on previous machines.
107 (bswapdi2_xxbrd): New insn.
108 (bswapdi2_reg): Disallow on ISA 3.0.
109 (register to register bswap64 splitter): Do not split the insn on
110 ISA 3.0 systems that use XXBRD.
112 2017-11-10 Martin Sebor <msebor@redhat.com>
115 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
117 (find_subframework_file): Same.
119 2017-11-10 Jeff Law <law@redhat.com>
121 * vr-values.h (VR_INITIALIZER): Move #define here.
122 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
123 * Makefile.in (OBJS): Add tree-evrp.o
124 * tree-vrp.h (assert_info): Move structure definition here.
125 (set_value_range_to_varying): Prototype.
126 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
127 (infer_value_range, register_edge_assert_for): Likewise.
128 (stmt_interesting_for_vrp): Likewise.
129 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
130 (set_value_range_to_varying): No longer static.
131 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
132 (infer_value_range, register_edge_assert_for): Likewise.
134 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
136 * auto-profile.c (afdo_indirect_call): Drop frequency.
137 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
138 (cgraph_node::create_edge): Drop frequency argument.
139 (cgraph_node::create_indirect_edge): Drop frequency argument.
140 (cgraph_edge::make_speculative): Drop frequency arguments.
141 (cgraph_edge::resolve_speculation): Do not update frequencies
142 (cgraph_edge::dump_edge_flags): Do not dump frequency.
143 (cgraph_node::dump): Check consistency in IPA mode.
144 (cgraph_edge::maybe_hot_p): Use IPA counter.
145 (cgraph_edge::verify_count_and_frequency): Rename to ...
146 (cgraph_edge::verify_count): ... this one; drop frequency checking.
147 (cgraph_node::verify_node): Update.
148 * cgraph.h (struct cgraph_edge): Drop frequency.
149 (cgraph_edge::frequency): New function.
150 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
152 (cgraph_edge::rebuild_edges): Likewise.
153 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
154 (duplicate_thunk_for_node): Do not pass frequency.
155 (cgraph_node::create_clone): Scale only counts.
156 (cgraph_node::create_virtual_clone): Do not pass frequency.
157 (cgraph_node::create_edge_including_clones): Do not pass frequency.
158 (cgraph_node::create_version_clone): Do not pass frequency.
159 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
160 (cgraph_node::expand_thunk): Do not pass frequency.
161 (cgraph_node::create_wrapper): Do not pass frequency.
162 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
164 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
165 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
166 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
167 (gather_caller_stats): Use frequency function.
168 (ipcp_cloning_candidate_p): Use frequency function.
169 (ipcp_propagate_stage): Use frequency function.
170 (get_info_about_necessary_edges): Use frequency function.
171 (update_profiling_info): Update only IPA profile.
172 (update_specialized_profile): Use frequency functoin.
173 (perhaps_add_new_callers): Update only IPA profile.
174 * ipa-devirt.c (ipa_devirt): Use IPA profile.
175 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
176 (dump_ipa_call_summary): Use frequency function.
177 (estimate_edge_size_and_time): Use frequency function.
178 (ipa_merge_fn_summary_after_inlining): Use frequency function.
179 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
180 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
181 (update_noncloned_counts): ... ths one; scale counts only.
182 (clone_inlined_nodes): Do not scale frequency.
183 (inline_call): Do not pass frequency.
184 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
185 (compute_inlined_call_time): Use IPA profile.
186 (want_inline_small_function_p): Use IPA profile.
187 (want_inline_self_recursive_call_p): Use IPA profile.
188 (edge_badness): Use IPA profile.
189 (lookup_recursive_calls): Use IPA profile.
190 (recursive_inlining): Do not pass frequency.
191 (resolve_noninline_speculation): Do not update frequency.
192 (inline_small_functions): Collect max of IPA profile.
193 (dump_overall_stats): Dump IPA porfile.
194 (dump_inline_stats): Dump IPA porfile.
195 (ipa_inline): Collect IPA stats.
196 * ipa-inline.h (clone_inlined_nodes): Update prototype.
197 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
198 (ipa_propagate_frequency): Use frequency function.
199 (ipa_profile): Cleanup.
200 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
201 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
202 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
203 (input_node): Do not stream frequency.
204 (input_edge): Do not stream frequency.
205 (merge_profile_summaries): Scale only IPA profiles.
206 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
207 * predict.c (drop_profile): Do not recompute frequency.
208 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
209 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
210 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
211 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
212 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
213 * tree-inline.c (copy_bb): Do not scale frequency.
214 (expand_call_inline): Do not scale frequency.
215 (tree_function_versioning): Do not scale frequency.
216 * ubsan.c (ubsan_create_edge): Do not pass frequency.
218 2017-11-10 Julia Koval <julia.koval@intel.com>
220 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
221 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
222 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
223 _mm256_maskz_gf2p8affine_epi64_epi8,
224 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
225 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
226 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
227 __builtin_ia32_vgf2p8affineqb_v32qi,
228 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
229 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
231 2017-11-10 Tamar Christina <tamar.christina@arm.com>
235 (arm_option_override): Refactor.
236 (arm_option_reconfigure_globals): New.
237 (arm_options_perform_arch_sanity_checks): New.
238 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
239 (arm_options_perform_arch_sanity_checks): Likewise
241 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
243 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
245 * rs6000/rs6000.md ("type" attr): Add qmul.
246 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
247 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
248 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
250 2017-11-10 Martin Sebor <msebor@redhat.com>
253 * builtins.c (compute_objsize): Handle arrays that
254 compute_builtin_object_size likes to fail for. Make extern.
255 * builtins.h (compute_objsize): Declare.
256 (check_strncpy_sizes): New function.
257 (expand_builtin_strncpy): Call check_strncpy_sizes.
258 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
259 -Wstringop-truncation.
260 (gimple_fold_builtin_strncat): Same.
261 * gimple.c (gimple_build_call_from_tree): Set call location.
262 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
263 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
264 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
265 (handle_builtin_strlen): Use strlen_to_stridx.
266 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
268 Use strlen_to_stridx.
269 (pass_strlen::execute): Release strlen_to_stridx.
270 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
271 (-Wstringop-truncation): Document new option.
273 2017-11-10 Martin Liska <mliska@suse.cz>
275 PR gcov-profile/82702
276 * gcov.c (main): Handle intermediate files in a different
278 (get_gcov_intermediate_filename): New function.
279 (output_gcov_file): Remove support of intermediate files.
280 (generate_results): Allocate intermediate file.
281 (release_structures): Clean-up properly fn_end.
282 (output_intermediate_file): Start iterating with line 1.
284 2017-11-10 Jakub Jelinek <jakub@redhat.com>
286 PR tree-optimization/82929
287 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
288 ops_swapped_p non-static data member.
289 (store_immediate_info::store_immediate_info): Clear it.
290 (imm_store_chain_info::coalesce_immediate_stores): If swapping
291 ops set ops_swapped_p.
292 (count_multiple_uses): Handle ops_swapped_p.
294 2017-11-10 Martin Liska <mliska@suse.cz>
296 * coverage.c (coverage_init): Stream information about
297 support of has_unexecuted_blocks.
298 * doc/gcov.texi: Document that.
299 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
300 * gcov.c (read_graph_file): Likewise.
301 (output_line_beginning): Fix a small issue with
304 2017-11-10 Bin Cheng <bin.cheng@arm.com>
306 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
307 reference of trivial component.
309 2017-11-10 Jakub Jelinek <jakub@redhat.com>
312 * gimple-ssa-store-merging.c
313 (pass_store_merging::terminate_all_aliasing_chains): For
314 gimple_store_p stmts also call refs_output_dependent_p.
316 PR rtl-optimization/82913
317 * compare-elim.c (try_merge_compare): Punt if def_insn is not
320 2017-11-09 Jeff Law <law@redhat.com>
322 * vr-values.h: New file with vr_values class.
323 * tree-vrp.c: Include vr-values.h
324 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
325 data objects into the vr_values class.
326 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
327 (get_value_range): Make it a member function within vr_values class.
328 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
329 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
330 (op_with_constant_singleton_value_range): Likewise.
331 (extract_range_for_var_from_comparison_expr): Likewise.
332 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
333 (extract_range_from_binary_expr): Likewise.
334 (extract_range_from_unary_expr): Likewise.
335 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
336 (check_for_binary_op_overflow, extract_range_basic): Likewise.
337 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
338 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
339 (compare_name_with_value, compare_names): Likewise.
340 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
341 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
342 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
343 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
344 (extract_range_from_phi_node): Likewise.
345 (simplify_truth_ops_using_ranges): Likewise.
346 (simplify_div_or_mod_using_ranges): Likewise.
347 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
348 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
349 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
350 (simplify_float_conversion_using_ranges): Likewise.
351 (simplify_internal_call_using_ranges): Likewise.
352 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
353 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
354 poitner via x_vr_values for calls into gimple folder.
355 (vrp_initialize_lattice): Make this the vr_values ctor.
356 (vrp_free_lattice): Make this the vr_values dtor.
357 (set_vr_value): New function.
358 (class vrp_prop): Add vr_values data member. Add various member
359 functions as well as member functions that delegate to vr_values.
360 (check_array_ref): Make a member function within vrp_prop class.
361 (search_for_addr_array, vrp_initialize): Likewise.
362 (vrp_finalize): Likewise. Revamp to avoid direct access to
363 vr_value, values_propagated, etc.
364 (check_array_bounds): Extract vrp_prop class instance pointer from
365 walk info structure. Use it to call member functions.
366 (check_all_array_refs): Make a member function within vrp_prop class.
367 Smuggle class instance pointer via walk info structure.
368 (x_vr_values): New local static.
369 (vrp_valueize): Use x_vr_values to get class instance.
370 (vr_valueize_1): Likewise.
371 (class vrp_folder): Add vr_values data member. Add various member
372 functions as well as member functions that delegate to vr_values.
373 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
374 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
375 class instance from vr_values. Use it to call member functions.
376 (vrp_dom_walker): Add vr_values data member.
377 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
378 instance via x_vr_values.
379 (identify_jump_threads): Accept vr_values as argument. Store
380 it into the walker structure.
381 (evrp_dom_walker): Add vr_values class data member. Add various
383 (evrp_dom_walker::try_find_new_range): Use vr_values data
384 member to access the memory allocator.
385 (evrp_dom_walker::before_dom_children): Store vr_values class
386 instance into the vrp_folder class.
387 (evrp_dom_walker::push_value_range): Rework to avoid direct
388 access to num_vr_values and vr_value.
389 (evrp_dom_walker::pop_value_range): Likewise.
390 (execute_early_vrp): Remove call to vrp_initialize_lattice.
391 Use vr_values to get to dump_all_value_ranges member function.
392 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
393 and simplify_cond_using_ranges_2 via vrp_prop class instance.
394 Pass vr_values class instance down to identify_jump_threads.
395 Remove call to vrp_free_lattice.
396 (debug_all_value_ranges): Remove.
398 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
399 (vrp_folder): Likewise.
401 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
402 Get it from the existing bitmap instead.
403 (vrp_intersect_ranges_1): Likewise.
405 2017-11-09 Jakub Jelinek <jakub@redhat.com>
407 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
409 (store_immediate_info::store_immediate_info): Add bitnotp argument,
411 (imm_store_chain_info::coalesce_immediate_stores): Break group
412 if bit_not_p is different.
413 (count_multiple_uses, split_group,
414 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
415 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
416 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
417 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
418 to store_immediate_info ctor.
420 2017-11-09 Jim Wilson <jimw@sifive.com>
422 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
423 (scan_prog_file): Likewise.
425 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
427 * bb-reorder.c (max_entry_frequency): Remove.
428 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
429 connect_traces, push_to_next_round_p): Remove prototypes.
430 (find_traces_1_round): Use counts only.
431 (push_to_next_round_p): Likewise.
432 (find_traces): Likewise.
433 (rotate_loop): Likewise.
434 (find_traces_1_round): Likewise.
435 (connect_traces): Likewise.
436 (edge_order): Likewise.
438 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
440 * config/arm/arm.c (output_return_instruction): Add comments to
441 indicate requirement for cmse_nonsecure_entry return to account
442 for the size of clearing instruction output here.
443 (thumb_exit): Likewise.
444 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
445 return in hardfloat mode.
447 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
449 * config/rs6000/rs6000.c (machine_function): Add a bool,
450 "toc_is_wrapped_separately".
451 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
452 if it wasn't explicitly set or unset, we are optimizing for speed, and
453 doing separate shrink-wrapping.
454 (rs6000_get_separate_components): Enable the TOC component if
455 saving the TOC register in the prologue.
456 (rs6000_components_for_bb): Handle the TOC component.
457 (rs6000_emit_prologue_components): Store the TOC register where needed.
458 (rs6000_set_handled_components): Mark TOC as handled, if handled.
459 (rs6000_emit_prologue): Don't save the TOC if that is already done.
461 2017-11-09 Martin Jambor <mjambor@suse.cz>
463 * ipa-param-manipulation.c: New file.
464 * ipa-param-manipulation.h: Likewise.
465 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
466 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
467 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
468 (ipa_parm_adjustment): Likewise.
469 (ipa_parm_adjustment_vec): Likewise.
470 (ipa_get_vector_of_formal_parms): Moved declaration to
471 ipa-param-manipulation.h.
472 (ipa_get_vector_of_formal_parm_types): Likewise.
473 (ipa_modify_formal_parameters): Likewise.
474 (ipa_modify_call_arguments): Likewise.
475 (ipa_combine_adjustments): Likewise.
476 (ipa_dump_param_adjustments): Likewise.
477 (ipa_modify_expr): Likewise.
478 (ipa_get_adjustment_candidate): Likewise.
479 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
480 ipa-param-manipulation.c.
481 (ipa_get_vector_of_formal_parm_types): Likewise.
482 (ipa_modify_formal_parameters): Likewise.
483 (ipa_modify_call_arguments): Likewise.
484 (ipa_modify_expr): Likewise.
485 (get_ssa_base_param): Likewise.
486 (ipa_get_adjustment_candidate): Likewise.
487 (index_in_adjustments_multiple_times_p): Likewise.
488 (ipa_combine_adjustments): Likewise.
489 (ipa_dump_param_adjustments): Likewise.
490 * tree-sra.c: Also include ipa-param-manipulation.h
491 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
494 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
495 Alan Hayward <alan.hayward@arm.com>
496 David Sherwood <david.sherwood@arm.com>
498 * doc/sourcebuild.texi (vect_masked_store): Document.
500 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
506 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
507 Alan Hayward <alan.hayward@arm.com>
508 David Sherwood <david.sherwood@arm.com>
510 * doc/sourcebuild.texi (vect_variable_length): Document.
512 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
513 Alan Hayward <alan.hayward@arm.com>
514 David Sherwood <david.sherwood@arm.com>
516 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
518 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
519 Alan Hayward <alan.hayward@arm.com>
520 David Sherwood <david.sherwood@arm.com>
522 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
524 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
525 Alan Hayward <alan.hayward@arm.com>
526 David Sherwood <david.sherwood@arm.com>
528 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
529 previously undocumented selectors.
530 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
532 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
534 * doc/rtl.texi (const_vector): Say that elements can be
536 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
537 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
538 (gen_vec_duplicate): Use it instead of CONSTANT_P.
539 * optabs.c (expand_vector_broadcast): Likewise.
541 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
542 Alan Hayward <alan.hayward@arm.com>
543 David Sherwood <david.sherwood@arm.com>
545 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
546 scaled index even if the unscaled address was invalid.
547 Don't increase the complexity of using a scale in that case.
549 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * doc/rtl.texi: Rewrite the subreg rules so that they partition
554 the inner register into REGMODE_NATURAL_SIZE bytes rather than
555 UNITS_PER_WORD bytes.
556 * emit-rtl.c (validate_subreg): Divide subregs into blocks
557 based on REGMODE_NATURAL_SIZE of the inner mode.
558 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
559 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
560 * expmed.c (lowpart_bit_field_p): Divide the value up into
561 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
562 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
563 whether something is likely to occupy more than one register.
565 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
568 * ipa-inline-transform.c (update_noncloned_frequencies): Use
569 profile_count::adjust_for_ipa_scaling.
570 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
571 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
573 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
575 2017-11-09 Jakub Jelinek <jakub@redhat.com>
577 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
578 (split_group): Add total_orig and total_new arguments, estimate the
579 number of statements related to the store group without store merging
580 and with store merging.
581 (imm_store_chain_info::output_merged_store): Adjust split_group
582 callers, punt if estimated number of statements with store merging
583 is not smaller than estimated number of statements without it.
585 (handled_load): Remove has_single_use checks.
586 (pass_store_merging::process_store): Likewise.
588 2017-11-09 Richard Biener <rguenther@suse.de>
590 PR tree-optimization/82902
591 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
593 2017-11-09 Martin Liska <mliska@suse.cz>
596 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
599 2017-11-09 Martin Liska <mliska@suse.cz>
601 PR tree-optimization/82669
602 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
604 2017-11-09 Martin Liska <mliska@suse.cz>
606 PR gcov-profile/48463
607 * coverage.c (coverage_begin_function): Output also end locus
608 of a function and information whether the function is
610 * gcov-dump.c (tag_function): Parse and print the information.
611 * gcov.c (INCLUDE_MAP): Add include.
612 (INCLUDE_SET): Likewise.
613 (struct line_info): Move earlier in the source file because
614 of vector<line_info> in function_info structure.
615 (line_info::line_info): Likewise.
616 (line_info::has_block): Likewise.
617 (struct source_info): Add new member index.
618 (source_info::get_functions_at_location): New function.
619 (function_info::group_line_p): New function.
620 (output_intermediate_line): New function.
621 (output_intermediate_file): Use the mentioned function.
622 (struct function_start): New.
623 (struct function_start_pair_hash): Likewise.
624 (process_file): Add code that identifies group functions.
625 Assign lines either to global or function scope.
626 (generate_results): Skip artificial functions.
627 (find_source): Assign index for each source file.
628 (read_graph_file): Read new flag artificial and end_line.
629 (add_line_counts): Assign it either to global of function scope.
630 (accumulate_line_counts): Isolate core of the function to
631 accumulate_line_info and call it for both function and global
633 (accumulate_line_info): New function.
634 (output_line_beginning): Fix GNU coding style.
635 (print_source_line): New function.
636 (output_line_details): Likewise.
637 (output_function_details): Likewise.
638 (output_lines): Iterate both source (global) scope and function
640 (struct function_line_start_cmp): New class.
641 * doc/gcov.texi: Reflect changes in documentation.
643 2017-11-09 Jakub Jelinek <jakub@redhat.com>
646 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
647 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
648 and similarly for not instead of neg.
650 2017-11-08 Andi Kleen <ak@linux.intel.com>
652 * config/i386/i386.opt: Add -mforce-indirect-call.
653 * config/i386/predicates.md: Check for flag_force_indirect_call.
654 * doc/invoke.texi: Document -mforce-indirect-call
656 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
658 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
660 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
661 config/riscv/riscv.c (predict.h): New include.
662 (riscv_slow_unaligned_access_p): No longer static.
663 (riscv_block_move_straight): Add require.
664 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
666 2017-11-08 Jakub Jelinek <jakub@redhat.com>
669 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
670 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
671 nonimmediate_operand predicate for operand 1 instead of
674 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
676 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
678 * config/aarch64/constraints.md (Uml): New constraint.
679 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
682 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
684 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
685 of two vec_duplicates into a vec_concat.
687 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
689 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
690 Simplify vec_merge of vec_duplicate and vec_concat.
691 * config/aarch64/constraints.md (Utq): New constraint.
692 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
695 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
697 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
698 Simplify vec_merge of vec_duplicate and const_vector.
699 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
701 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
702 mode iterator. Update predicate on operand 1 to
703 handle non-const_vec constants. Delete constraints.
704 (*aarch64_combinez_be<mode>): Likewise for operand 2.
706 2017-11-08 Jakub Jelinek <jakub@redhat.com>
708 PR tree-optimization/78821
709 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
711 (store_operand_info::store_operand_info): Initialize it to false.
712 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
713 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
714 store in the group, and if chain_info is non-NULL, to ignore altogether
716 (compatible_load_p): Fail if bit_not_p does not match.
717 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
718 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
719 (pass_store_merging::process_store): Adjust
720 terminate_all_aliasing_chains calls to pass NULL in all current spots,
721 call terminate_all_aliasing_chains newly when adding a store into
722 a chain with non-NULL chain_info.
724 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
726 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
728 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
730 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
732 (aarch64_layout_frame): Initialise emit_frame_chain.
733 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
734 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
736 2017-11-08 Martin Liska <mliska@suse.cz>
738 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
739 of gimple_call_internal_p.
741 2017-11-07 Tom de Vries <tom@codesourcery.com>
743 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
745 2017-11-07 Tom de Vries <tom@codesourcery.com>
747 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
750 2017-11-08 Martin Liska <mliska@suse.cz>
753 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
755 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
757 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
759 2017-11-07 Jakub Jelinek <jakub@redhat.com>
762 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
763 (*cmp<mode>_ccz_1): New insn with $k alternative.
766 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
767 RTX_COMM_COMPARE as commutative as well.
768 (ix86_binary_operator_ok): Formatting fix.
769 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
770 *<code><mode>3<mask_name><round_saeonly_name>,
771 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
772 *<s>mul<mode>3_highpart<mask_name>,
773 *vec_widen_umult_even_v16si<mask_name>,
774 *vec_widen_umult_even_v8si<mask_name>,
775 *vec_widen_umult_even_v4si<mask_name>,
776 *vec_widen_smult_even_v16si<mask_name>,
777 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
778 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
779 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
780 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
781 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
782 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
783 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
784 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
785 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
786 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
787 ix86_binary_operator_ok. Formatting fixes.
788 (*<plusminus_insn><mode>3<mask_name><round_name>,
789 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
792 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
794 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
795 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
796 eq and ne if TARGET_ISEL.
797 (cmp): New code_iterator.
798 (UNS, UNSU_, UNSIK): New code_attrs.
799 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
800 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
802 ("eq<mode>3"): ... this.
803 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
805 ("ne<mode>3"): ... this.
807 2017-11-07 Julia Koval <julia.koval@intel.com>
810 * common/config/i386/i386-common.c
811 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
812 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
813 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
814 * config/i386/i386.opt: Ditto.
815 * config/i386/i386.c (ix86_target_string): Ditto.
816 (ix86_option_override_internal): Ditto.
817 (ix86_init_mpx_builtins): Move MPX to args2.
818 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
819 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
820 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
821 __builtin_ia32_vgf2p8affineinvqb_v32qi,
822 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
823 __builtin_ia32_vgf2p8affineinvqb_v16qi,
824 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
826 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
829 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
830 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
831 (zero-extendsidi peephole2): Remove peephole.
833 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
836 * stor-layout.c: Include attribs.h.
837 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
838 explicit lookup of "aligned" attribute.
840 2017-11-07 Andrew Waterman <andrew@sifive.com>
842 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
843 (riscv_expand_block_move): Likewise.
844 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
846 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
847 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
848 * config/riscv/riscv.c (riscv_block_move_straight): New function.
849 (riscv_adjust_block_mem): Likewise.
850 (riscv_block_move_loop): Likewise.
851 (riscv_expand_block_move): Likewise.
852 gcc/config/riscv/riscv.md (movmemsi): New pattern.
854 2017-11-07 Michael Clark <michaeljclark@mac.com>
856 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
857 (MUSL_DYNAMIC_LINKER): Likewise.
859 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
861 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
862 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
865 2017-11-07 Richard Biener <rguenther@suse.de>
867 * match.pd: Fix build.
869 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
870 Jackson Woodruff <jackson.woodruff@arm.com>
872 PR tree-optimization/71026
873 * match.pd: Canonicalize negate in division.
875 2017-11-07 Sudakshina Das <sudi.das@arm.com>
878 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
880 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
882 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
883 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
885 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
887 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
888 non-scalar integral types.
889 * match.pd (negate_expr_p): Handle MINUS_EXPR.
890 (-(A-B), -(~A)): New transformations.
892 2017-11-07 Tom de Vries <tom@codesourcery.com>
894 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
895 semicolon after "do {} while (0)".
896 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
897 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
898 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
899 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
900 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
902 2017-11-07 Tom de Vries <tom@codesourcery.com>
904 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
905 after "do {} while (0)".
906 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
907 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
908 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
909 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
910 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
912 2017-11-07 Tom de Vries <tom@codesourcery.com>
915 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
917 (arm_rtx_costs_internal): Add missing semicolon after
918 HANDLE_NARROW_SHIFT_ARITH call.
920 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
923 disable isel if it was not set explicitly.
925 2017-11-06 James Bowman <james.bowman@ftdichip.com>
927 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
929 (add_type_attribute) likewise.
931 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
933 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
934 of struct ix86_frame.
935 (ix86_initial_elimination_offset): Likewise.
936 (ix86_expand_split_stack_prologue): Likewise.
938 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
940 * tree-vrp.h (enum value_range_type): Update stale comment.
942 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
943 Alan Hayward <alan.hayward@arm.com>
944 David Sherwood <david.sherwood@arm.com>
946 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
947 (aarch64_expand_vec_perm_const): Take the number of units too.
948 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
949 (aarch64_expand_vec_perm_const): Likewise.
950 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
951 (vec_perm<mode>): Update accordingly.
953 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
958 Take the number of units too.
959 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
960 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
961 but check for a vector mode before rather than after the call.
962 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
963 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
964 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
965 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
966 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
967 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
968 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
969 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
970 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
971 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
972 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
973 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
974 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
975 (aarch64_sqdmull2_n<mode>): Update accordingly.
977 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
978 Alan Hayward <alan.hayward@arm.com>
979 David Sherwood <david.sherwood@arm.com>
981 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
982 the number of units too.
983 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
984 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
985 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
986 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
987 (vec_store_lanesxi<mode>): Update accordingly.
989 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
994 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
995 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
996 of units rather than the mode.
997 * config/aarch64/iterators.md (nunits): New mode attribute.
998 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
999 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1000 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1001 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1002 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1003 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1004 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1005 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1006 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1007 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1008 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1009 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1010 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1011 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1012 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1013 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1014 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1015 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1016 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1017 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1018 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1019 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1020 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1021 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1022 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1023 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1024 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1025 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1026 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1027 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1028 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1029 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1030 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1031 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1032 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1033 (aarch64_simd_vec_setv2di): Likewise.
1035 2017-11-06 Carl Love <cel@us.ibm.com>
1037 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1039 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1040 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1042 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1044 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1045 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1046 Add power 8 macro expansions.
1047 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1048 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1049 power 8 instructions. (VSX_XXBR): Add iterator.
1051 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1053 * config/arm/arm.md (predicable_short_it): Change default to "no",
1054 improve documentation, remove uses that are identical to the default.
1055 (enabled_for_depr_it): Rename to enabled_for_short_it.
1056 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1057 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1058 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1059 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1060 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1062 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1065 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1066 float128 helper macros, which are no longer used after deleting
1067 the old 'q' built-in functions, and moving the round to odd
1068 built-in functions to being special built-in functions.
1069 (BU_FLOAT128_2): Likewise.
1070 (BU_FLOAT128_1_HW): Likewise.
1071 (BU_FLOAT128_2_HW): Likewise.
1072 (BU_FLOAT128_3_HW): Likewise.
1073 (FABSQ): Delete old 'q' built-in functions.
1074 (COPYSIGNQ): Likewise.
1075 (SQRTF128_ODD): Move round to odd built-in functions to be
1076 special built-in functions, so that we can handle
1077 -mabi=ieeelongdouble.
1078 (TRUNCF128_ODD): Likewise.
1079 (ADDF128_ODD): Likewise.
1080 (SUBF128_ODD): Likewise.
1081 (MULF128_ODD): Likewise.
1082 (DIVF128_ODD): Likewise.
1083 (FMAF128_ODD): Likewise.
1084 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1085 built-in names to 'f128'.
1086 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1087 old 'q' built-in functions, as the machine independent code for
1088 'f128' built-in functions handles this.
1089 (rs6000_expand_builtin): Add expansion for float128 round to odd
1090 functions, keying off on -mabi=ieeelongdouble of whether to use
1091 the KFmode or TFmode variant.
1092 (rs6000_init_builtins): Initialize the _Float128 round to odd
1094 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1095 _Float128 'q' built-in functions are now mapped into the new
1096 'f128' built-in functions.
1098 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1100 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1101 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1103 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1106 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1107 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1108 * toplev.c: Include "ipa-fnsummary.h".
1109 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1111 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1113 PR tree-optimization/82838
1114 * gimple-ssa-store-merging.c
1115 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1116 on a separate gimple_seq which is then appended to seq.
1118 2017-11-06 Jeff Law <law@redhat.com>
1121 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1122 (get_probe_interval): New functions.
1123 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1124 (ix86_adjust_stack_and_probe): Likewise.
1125 (output_adjust_stack_and_probe): Likewise.
1126 (ix86_emit_probe_stack_range): Likewise.
1127 (ix86_expand_prologue): Likewise.
1129 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1131 PR tree-optimization/82816
1132 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1133 if the modes of the two types are the same.
1134 (convert_plusminus_to_widen): Likewise.
1136 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1138 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1140 (usadv16qi): New define_expand.
1141 (usadv8hi): New define_expand.
1143 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1146 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1148 (inline_transform): Likewise.
1149 * predict.c (counts_to_freqs): Remove useless conditional.
1150 * profile-count.h (profile_count::apply_scale): Move sanity check.
1151 * tree-inline.c (copy_bb): Always scale.
1152 (copy_cfg_body): Likewise.
1154 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1157 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1159 (*cmp_ite0): Add enabled_for_depr_it attribute.
1160 (*cmp_ite1): Likewise.
1162 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1164 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1167 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1169 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1170 return the value by pointer.
1171 (extract_range_from_multiplicative_op_1): Update accordingly.
1172 Return as soon as an operation fails.
1174 2017-11-05 Tom de Vries <tom@codesourcery.com>
1177 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1178 (DEF_SANITIZER_BUILTIN): ... here.
1179 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1180 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1182 2017-11-05 Tom de Vries <tom@codesourcery.com>
1185 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1187 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1188 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1189 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1191 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1192 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1193 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1195 2017-11-05 Tom de Vries <tom@codesourcery.com>
1198 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1201 2017-11-04 Michael Clark <michaeljclark@mac.com>
1203 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1204 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1206 (*addsi3_extended): Likewise.
1207 (*addsi3_extended2): Likewise.
1208 (<optab>si3): Likewise.
1209 (<optab>di3): Likewise.
1210 (<optab><mode>3): Likewise.
1211 (<*optabe>si3_internal): Likewise.
1212 (zero_extendqi<SUPERQI:mode>2): Likewise.
1213 (*add<mode>hi3): Likewise.
1214 (*xor<mode>hi3): Likewise.
1215 (<optab>di3): Likewise.
1216 (*<optab>si3_extend): Likewise.
1217 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1218 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1219 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1221 2017-11-04 Andrew Waterman <andrew@sifive.com>
1223 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1224 TARGET_STRICT_ALIGN based upon -mtune argument.
1226 2017-11-04 Andrew Waterman <andrew@sifive.com>
1228 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1230 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1232 config/i386/i386.c (choose_basereg): Use optional scratch
1233 register and add assertion.
1234 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1235 needed, and don't allocate stack.
1236 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1237 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1238 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1240 2017-11-03 Jeff Law <law@redhat.com>
1242 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1243 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1244 to probe at the start of a noreturn function.
1246 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1248 PR tree-optimization/78821
1249 * gimple-ssa-store-merging.c: Update the file comment.
1250 (MAX_STORE_ALIAS_CHECKS): Define.
1251 (struct store_operand_info): New type.
1252 (store_operand_info::store_operand_info): New constructor.
1253 (struct store_immediate_info): Add rhs_code and ops data members.
1254 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1255 and op1r arguments to the ctor, initialize corresponding data members.
1256 (struct merged_store_group): Add load_align_base and load_align
1258 (merged_store_group::merged_store_group): Initialize them.
1259 (merged_store_group::do_merge): Update them.
1260 (merged_store_group::apply_stores): Pick the constant for
1261 encode_tree_to_bitpos from one of the two operands, or skip
1262 encode_tree_to_bitpos if neither operand is a constant.
1263 (class pass_store_merging): Add process_store method decl. Remove
1264 bool argument from terminate_all_aliasing_chains method decl.
1265 (pass_store_merging::terminate_all_aliasing_chains): Remove
1266 var_offset_p argument and corresponding handling.
1267 (stmts_may_clobber_ref_p): New function.
1268 (compatible_load_p): New function.
1269 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1270 if there is overlap and rhs_code is not INTEGER_CST. For
1271 non-overlapping stores terminate group if rhs is not mergeable.
1272 (get_alias_type_for_stmts): Change first argument from
1273 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1274 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1275 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1277 (get_location_for_stmts): Change first argument from
1278 auto_vec<gimple *> & to vec<gimple *> &.
1279 (struct split_store): Remove orig_stmts data member, add orig_stores.
1280 (split_store::split_store): Create orig_stores rather than orig_stmts.
1281 (find_constituent_stmts): Renamed to ...
1282 (find_constituent_stores): ... this. Change second argument from
1283 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1284 to info structures rather than the statements.
1285 (split_group): Rename ALLOW_UNALIGNED argument to
1286 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1287 it. Adjust find_constituent_stores caller.
1288 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1289 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1290 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1291 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1292 (mem_valid_for_store_merging): New function.
1293 (handled_load): New function.
1294 (pass_store_merging::process_store): New method.
1295 (pass_store_merging::execute): Use process_store method. Adjust
1296 terminate_all_aliasing_chains caller.
1298 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1300 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1301 Return true for more constants, symbols and label references.
1302 (aarch64_valid_floating_const): Remove unused function.
1304 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1307 * config/aarch64/aarch64.c (aarch64_layout_frame):
1308 Undo forcing of LR at bottom of frame.
1310 2017-11-03 Jeff Law <law@redhat.com>
1313 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1314 for int_registers_saved.
1316 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1317 extracted from tree-ssa-dom.c.
1318 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1319 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1320 (record_equivalences_from_incoming_edge): Add additional argument
1321 to single_pred_edge_ignoring_loop_edges call.
1322 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1323 (uncprop_dom_walker::before_dom_children): Add additional argument
1324 to single_pred_edge_ignoring_loop_edges call.
1325 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1326 single_pred_edge_ignoring_loop_edges rather than open coding.
1327 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1329 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1331 * match.pd (-(-A)): Rewrite.
1333 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1335 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1336 (rs6000_emit_int_cmove): New declaration.
1337 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1338 (rs6000_emit_sISEL): Delete.
1339 (rs6000_emit_int_cmove): Make non-static.
1340 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1341 instead of rs6000_emit_sISEL.
1343 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1345 * asan.c (create_cond_insert_point): Maintain profile.
1346 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1348 * basic-block.h (struct basic_block_def): Remove frequency.
1349 (EDGE_FREQUENCY): Use to_frequency
1350 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1352 (find_traces): Update to use to_frequency.
1353 (find_traces_1_round): Likewise; use only IPA counts.
1354 (bb_to_key): Likewise.
1355 (connect_traces): Use IPA counts only.
1356 (copy_bb_p): Update to use to_frequency.
1357 (fix_up_crossing_landing_pad): Likewise.
1358 (sanitize_hot_paths): Likewise.
1359 * bt-load.c (basic_block_freq): Likewise.
1360 * cfg.c (init_flow): Set count_max to uninitialized.
1361 (check_bb_profile): Remove frequencies; check counts.
1362 (dump_bb_info): Do not dump frequencies.
1363 (update_bb_profile_for_threading): Update counts only.
1364 (scale_bbs_frequencies_int): Likewise.
1365 (MAX_SAFE_MULTIPLIER): Remove.
1366 (scale_bbs_frequencies_gcov_type): Update counts only.
1367 (scale_bbs_frequencies_profile_count): Update counts only.
1368 (scale_bbs_frequencies): Update counts only.
1369 * cfg.h (struct control_flow_graph): Add count-max.
1370 (update_bb_profile_for_threading): Update prototype.
1371 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1372 (find_many_sub_basic_blocks): Likewise.
1373 * cfgcleanup.c (try_forward_edges): Likewise.
1374 (try_crossjump_to_edge): Likewise.
1375 * cfgexpand.c (expand_gimple_cond): Likewise.
1376 (expand_gimple_tailcall): Likewise.
1377 (construct_init_block): Likewise.
1378 (construct_exit_block): Likewise.
1379 * cfghooks.c (verify_flow_info): Check consistency of counts.
1380 (dump_bb_for_graph): Do not dump frequencies.
1381 (split_block_1): Do not update frequencies.
1382 (split_edge): Do not update frequencies.
1383 (make_forwarder_block): Do not update frequencies.
1384 (duplicate_block): Do not update frequencies.
1385 (account_profile_record): Do not update frequencies.
1386 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1387 for global heuristics.
1388 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1389 (expected_loop_iterations_unbounded): Use counts only.
1390 * cfgloopmanip.c (scale_loop_profile): Simplify.
1391 (create_empty_loop_on_edge): Simplify
1393 (duplicate_loop_to_header_edge): Simplify
1394 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1395 (update_br_prob_note): Take care of removing note when profile
1397 (relink_block_chain): Do not dump frequency.
1398 (rtl_account_profile_record): Use to_frequency.
1399 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1400 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1401 (cgraph_update_edges_for_call_stmt_node): Likewise.
1402 (cgraph_edge::verify_count_and_frequency): Update.
1403 (cgraph_node::verify_node): Temporarily disable frequency verification.
1404 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1405 to_cgraph_frequency.
1406 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1407 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1409 (cgraph_node::expand_thunk): Update profile.
1410 * except.c (dw2_build_landing_pads): Do not update frequency.
1411 * final.c (compute_alignments): Use to_frequency.
1412 (dump_basic_block_info): Do not dump frequency.
1413 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1414 (dump_gimple_bb_header): Do not dump frequency.
1415 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1417 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1418 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1419 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1420 (init_before_recovery): Do not update frequency.
1421 (sched_create_recovery_edges): Do not update frequency.
1422 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1423 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1424 (ipa_cp_c_finalize): Set max_count to uninitialized.
1425 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1426 (param_change_prob): Use counts.
1427 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1429 * ipa-split.c (consider_split): Use to_frequency.
1430 (split_function): Use to_frequency.
1431 * ira-build.c (loop_compare_func): Likewise.
1432 (mark_loops_for_removal): Likewise.
1433 (mark_all_loops_for_removal): Likewise.
1434 * loop-doloop.c (doloop_modify): Do not update frequency.
1435 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1437 * lto-streamer-in.c (input_function): Update count_max.
1438 * omp-expand.c (expand_omp_taskreg): Update count_max.
1439 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1440 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1441 (maybe_hot_count_p): Use ipa counts only.
1442 (maybe_hot_bb_p): Simplify.
1443 (maybe_hot_edge_p): Simplify.
1444 (probably_never_executed): Do not take frequency argument.
1445 (probably_never_executed_bb_p): Do not pass frequency.
1446 (probably_never_executed_edge_p): Likewise.
1447 (combine_predictions_for_bb): Check that profile is nonzero.
1448 (propagate_freq): Do not set frequency.
1449 (drop_profile): Simplify.
1450 (counts_to_freqs): Simplify.
1451 (expensive_function_p): Use to_frequency.
1452 (propagate_unlikely_bbs_forward): Simplify.
1453 (determine_unlikely_bbs): Simplify.
1454 (estimate_bb_frequencies): Add hack to silence graphite issues.
1455 (compute_function_frequency): Use ipa counts.
1456 (pass_profile::execute): Update.
1457 (rebuild_frequencies): Use counts only.
1458 (force_edge_cold): Use counts only.
1459 * profile-count.c (profile_count::dump): Dump new count types.
1460 (profile_count::differs_from_p): Check compatiblity.
1461 (profile_count::to_frequency): New function.
1462 (profile_count::to_cgraph_frequency): New function.
1463 * profile-count.h (struct function): Declare.
1464 (enum profile_quality): Add profile_guessed_local and
1465 profile_guessed_global0.
1466 (class profile_proability): Decrease number of bits to 29;
1467 update from_reg_br_prob_note and to_reg_br_prob_note.
1468 (class profile_count: Update comment; decrease number of bits
1469 to 61. Check compatibility.
1470 (profile_count::compatible_p): New private member function.
1471 (profile_count::ipa_p): New member function.
1472 (profile_count::operator<): Handle global zero correctly.
1473 (profile_count::operator>): Handle global zero correctly.
1474 (profile_count::operator<=): Handle global zero correctly.
1475 (profile_count::operator>=): Handle global zero correctly.
1476 (profile_count::nonzero_p): New member function.
1477 (profile_count::force_nonzero): New member function.
1478 (profile_count::max): New member function.
1479 (profile_count::apply_scale): Handle IPA scalling.
1480 (profile_count::guessed_local): New member function.
1481 (profile_count::global0): New member function.
1482 (profile_count::ipa): New member function.
1483 (profile_count::to_frequency): Declare.
1484 (profile_count::to_cgraph_frequency): Declare.
1485 * profile.c (OVERLAP_BASE): Delete.
1486 (compute_frequency_overlap): Delete.
1487 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1488 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1489 * sched-ebb.c (rank): Use counts only.
1490 * shrink-wrap.c (handle_simple_exit): Use counts only.
1491 (try_shrink_wrapping): Use counts only.
1492 (place_prologue_for_one_component): Use counts only.
1493 * tracer.c (find_best_predecessor): Use to_frequency.
1494 (find_trace): Use to_frequency.
1495 (tail_duplicate): Use to_frequency.
1496 * trans-mem.c (expand_transaction): Do not update frequency.
1497 * tree-call-cdce.c: Do not update frequency.
1498 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1499 (gimple_merge_blocks): Likewise.
1500 (gimple_split_edge): Likewise.
1501 (gimple_duplicate_sese_region): Likewise.
1502 (gimple_duplicate_sese_tail): Likewise.
1503 (move_sese_region_to_fn): Likewise.
1504 (gimple_account_profile_record): Likewise.
1505 (insert_cond_bb): Likewise.
1506 * tree-complex.c (expand_complex_div_wide): Likewise.
1507 * tree-eh.c (lower_resx): Update profile.
1508 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1510 (initialize_cfun): Do not initialize frequencies
1511 (freqs_to_counts): Delete.
1512 (copy_cfg_body): Ignore count parameter.
1513 (copy_body): Update.
1514 (expand_call_inline): Update count_max.
1515 (optimize_inline_calls): Update count_max.
1516 (tree_function_versioning): Update count_max.
1517 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1518 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1520 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1521 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1522 (try_peel_loop): Likewise.
1523 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1525 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1526 (tree_transform_and_unroll_loop): Do not use frequencies
1527 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1528 Use reliable prediction only.
1529 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1530 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1531 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1532 probability scaling.
1533 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1534 not update frequency
1535 (any_remaining_duplicated_blocks): Likewise.
1536 (update_profile): Likewise.
1537 (estimated_freqs_path): Delete.
1538 (freqs_to_counts_path): Delete.
1539 (clear_counts_path): Delete.
1540 (ssa_fix_duplicate_block_edges): Likewise.
1541 (duplicate_thread_path): Likewise.
1542 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1543 * tree-tailcall.c (decrease_profile): Do not update frequency.
1544 (eliminate_tail_call): Likewise.
1545 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1546 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1547 (optimize_mask_stores): Likewise.
1548 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1549 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1550 (ubsan_expand_ptr_ifn): Update profile.
1551 * value-prof.c (gimple_ic): Simplify.
1552 * value-prof.h (gimple_ic): Update prototype.
1553 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1554 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1556 (want_inline_self_recursive_call_p): Likewise.
1557 (resolve_noninline_speculation): Only cummulate defined counts.
1558 (inline_small_functions): Use nonzero_p.
1559 (ipa_inline): Do not access freed node.
1561 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1563 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1564 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1566 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1568 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1569 non-legitimate address.
1571 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1573 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1574 (*lt0_<mode>di, *lt0_<mode>si): New.
1576 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1578 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1579 TARGET_PAIRED_FLOAT.
1581 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1582 Jim Wilson <jim.wilson@linaro.org>
1584 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1585 * config/aarch64/aarch64-tune.md: Regenerated.
1586 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1587 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1589 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1591 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1592 (arc_expand_prologue): Restore blink for millicode.
1593 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1595 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1598 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1599 gen_vec_duplicate after forcing the scalar into a register.
1601 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1603 * combine (try_combine): Print the insns input to try_combine to the
1606 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1609 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1610 Remove second argument from aarch64_process_target_attr call.
1611 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1613 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1615 (aarch64_handle_attr_arch): Remove second argument.
1616 (aarch64_handle_attr_cpu): Ditto.
1617 (aarch64_handle_attr_tune): Ditto.
1618 (aarch64_handle_attr_isa_flags): Ditto.
1619 (aarch64_process_one_target_attr): Ditto.
1620 (aarch64_process_target_attr): Ditto.
1621 (aarch64_option_valid_attribute_p): Remove second argument.
1622 on aarch64_process_target_attr call.
1624 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1626 * diagnostic.c: Include "selftest-diagnostic.h".
1627 (selftest::assert_location_text): New function.
1628 (selftest::test_diagnostic_get_location_text): New function.
1629 (selftest::diagnostic_c_tests): Call it.
1631 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1633 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1634 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1635 (class selftest::test_diagnostic_context): Move to...
1636 * selftest-diagnostic.c: New file.
1637 * selftest-diagnostic.h: New file.
1639 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1641 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1642 offset range for FT32B.
1643 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1644 * config/ft32/ft32.md: Add TARGET_NOPM.
1645 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1646 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1648 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1650 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1652 2017-11-02 Jeff Law <law@redhat.com>
1654 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1655 virtual keyword on FINAL OVERRIDE members.
1657 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1658 virtuals together. Add virtual destructor.
1659 (substitute_and_fold_engine): Similarly.
1661 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1663 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1665 2017-11-02 Richard Biener <rguenther@suse.de>
1667 PR tree-optimization/82795
1668 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1670 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1672 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1674 (gcc_GAS_CHECK_FEATURE): Remove.
1675 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1676 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1677 * configure: Regenerate.
1679 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1681 * config/arc/arc.c (hwloop_optimize): Account for empty
1684 2017-11-02 Richard Biener <rguenther@suse.de>
1687 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1688 Truncate ARRAY_REF index and element size.
1690 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1692 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1694 2017-11-01 Jeff Law <law@redhat.com>
1696 * tree-ssa-ccp.c (ccp_folder): New class derived from
1697 substitute_and_fold_engine.
1698 (ccp_folder::get_value): New member function.
1699 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1700 (ccp_fold_stmt): Remove prototype.
1701 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1702 * tree-ssa-copy.c (copy_folder): New class derived from
1703 substitute_and_fold_engine.
1704 (copy_folder::get_value): Renamed from get_value.
1705 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1706 * tree-vrp.c (vrp_folder): New class derived from
1707 substitute_and_fold_engine.
1708 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1709 (vrp_folder::get_value): New member function.
1710 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1711 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1712 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1713 provide a class interface to folder/substitute routines.
1714 (ssa_prop_fold_stmt_fn): Remove typedef.
1715 (ssa_prop_get_value_fn): Likewise.
1716 (subsitute_and_fold): Remove prototype.
1717 (replace_uses_in): Likewise.
1718 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1719 Renamed from replace_uses_in. Call the virtual member function
1720 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1721 (substitute_and_fold_dom_walker): Remove initialization of
1722 data member entries for calbacks. Add substitute_and_fold_engine
1723 member and initialize it.
1724 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1725 member functions for get_value, replace_phi_args_in c
1726 replace_uses_in, and fold_stmt calls.
1727 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1728 substitute_and_fold. Remove assert. Update ctor call.
1730 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1731 (ssa_prop_visit_phi_fn): Likewise.
1732 (class ssa_propagation_engine): New class to provide an interface
1734 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1736 (ssa_prop_visit_phi): Likewise.
1737 (ssa_propagation_engine::simulate_stmt): Moved into class.
1738 Call visit_phi/visit_stmt from the class rather than via
1739 file scoped static variables.
1740 (ssa_propagation_engine::simulate_block): Moved into class.
1741 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1742 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1743 set file scoped statics for the visit_stmt/visit_phi callbacks.
1744 * tree-complex.c (complex_propagate): New class derived from
1745 ssa_propagation_engine.
1746 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1747 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1748 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1750 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1751 ssa_propagation_engine.
1752 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1753 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1754 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1755 * tree-ssa-copy.c (copy_prop): New class derived from
1756 ssa_propagation_engine.
1757 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1758 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1759 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1760 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1761 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1762 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1763 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1765 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1767 PR rtl-optimization/82778
1768 PR rtl-optimization/82597
1769 * compare-elim.c (struct comparison): Add in_a_setter field.
1770 (find_comparison_dom_walker::before_dom_children): Remove killed
1771 bitmap and df_simulate_find_defs call, instead walk the defs.
1772 Compute last_setter and initialize in_a_setter. Merge definitions
1773 with first initialization for a few variables.
1774 (try_validate_parallel): Use insn_invalid_p instead of
1775 recog_memoized. Return insn rather than just the pattern.
1776 (try_merge_compare): Fix up comment. Don't uselessly test if
1777 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1779 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1782 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1787 aarch64_hard_regno_nregs to get the number of registers
1790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * config/aarch64/constraints.md (Upl): Rename to...
1796 * config/aarch64/aarch64.md
1797 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1800 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1801 Alan Hayward <alan.hayward@arm.com>
1802 David Sherwood <david.sherwood@arm.com>
1804 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1805 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1808 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1813 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1814 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1816 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1817 of a single element.
1818 * config/aarch64/iterators.md: Add a comment above the permute
1819 unspecs to say that they are generated directly by
1820 aarch64_expand_vec_perm_const.
1821 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1823 2017-11-01 Nathan Sidwell <nathan@acm.org>
1825 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1827 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1829 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1830 and medany code models, and describe what they do.
1832 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1834 Revert accidental duplicate:
1836 * combine.c (can_change_dest_mode): Reject changes in
1837 REGMODE_NATURAL_SIZE.
1839 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1841 PR rtl-optimization/64682
1842 PR rtl-optimization/69567
1843 PR rtl-optimization/69737
1844 PR rtl-optimization/82683
1845 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1846 register mentioned in the note, drop the note, unless it came from I3,
1847 in which case it should go to I3 again.
1849 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1851 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1852 and return false if not.
1853 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1855 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1857 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1858 range is known to be empty.
1860 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1865 and CONST_VECTOR_NUNITS instead of computing the number of units from
1866 the byte sizes of the vector and element.
1867 (simplify_binary_operation_1): Likewise.
1868 (simplify_const_binary_operation): Likewise.
1869 (simplify_ternary_operation): Likewise.
1871 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1872 Alan Hayward <alan.hayward@arm.com>
1873 David Sherwood <david.sherwood@arm.com>
1875 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1876 (int_mem_offset): ...this new function.
1877 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1878 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1879 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1882 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * lower-subreg.c (interesting_mode_p): New function.
1887 (compute_costs, find_decomposable_subregs, decompose_register)
1888 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1889 (resolve_clobber, dump_choices): Use it.
1891 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1892 Alan Hayward <alan.hayward@arm.com>
1893 David Sherwood <david.sherwood@arm.com>
1895 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1897 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1898 Alan Hayward <alan.hayward@arm.com>
1899 David Sherwood <david.sherwood@arm.com>
1901 * alias.c (find_base_value, find_base_term): Only process integer
1902 truncations. Check the precision rather than the size.
1904 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1905 Alan Hayward <alan.hayward@arm.com>
1906 David Sherwood <david.sherwood@arm.com>
1908 * machmode.h (is_narrower_int_mode): New function
1909 * optabs.c (expand_float, expand_fix): Use it.
1910 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1912 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1913 Alan Hayward <alan.hayward@arm.com>
1914 David Sherwood <david.sherwood@arm.com>
1916 * rtl.h (narrower_subreg_mode): New function.
1917 * ira-color.c (update_costs_from_allocno): Use it.
1919 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1920 Alan Hayward <alan.hayward@arm.com>
1921 David Sherwood <david.sherwood@arm.com>
1923 * optabs-query.h (convert_optab_p): New function, split out from...
1924 (convert_optab_handler): ...here.
1925 (widening_optab_handler): Delete.
1926 (find_widening_optab_handler): Remove permit_non_widening parameter.
1927 (find_widening_optab_handler_and_mode): Likewise. Provide an
1928 override that operates on mode class wrappers.
1929 * optabs-query.c (widening_optab_handler): Delete.
1930 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1931 parameter. Assert that the two modes are the same class and that
1932 the "from" mode is narrower than the "to" mode. Use
1933 convert_optab_handler instead of widening_optab_handler.
1934 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1935 instead of widening_optab_handler.
1936 * expr.c (expand_expr_real_2): Update calls to
1937 find_widening_optab_handler.
1938 * optabs.c (expand_widen_pattern_expr): Likewise.
1939 (expand_binop_directly): Take the insn_code as a parameter.
1940 (expand_binop): Only call find_widening_optab_handler for
1941 conversion optabs; use optab_handler otherwise. Update calls
1942 to find_widening_optab_handler and expand_binop_directly.
1943 Use convert_optab_handler instead of widening_optab_handler.
1944 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1945 find_widening_optab_handler and use scalar_mode rather than
1947 (convert_plusminus_to_widen): Likewise.
1949 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1950 Alan Hayward <alan.hayward@arm.com>
1951 David Sherwood <david.sherwood@arm.com>
1953 * machmode.h (fixed_size_mode): New class.
1954 * rtl.h (get_pool_mode): Return fixed_size_mode.
1955 * gengtype.c (main): Add fixed_size_mode.
1956 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1957 (get_raw_arg_mode): Likewise.
1958 * doc/tm.texi: Regenerate.
1959 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1960 * targhooks.c (default_get_reg_raw_mode): Likewise.
1961 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1962 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1963 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1964 (msp430_get_raw_result_mode): Likewise.
1965 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1966 * dbxout.c (dbxout_parms): Require fixed-size modes.
1967 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1968 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1969 * omp-low.c (lower_oacc_reductions): Likewise.
1970 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1971 (simplify_subreg): Update accordingly.
1972 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1973 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1974 that aren't fixed-size.
1975 (get_pool_mode): Return a fixed_size_mode.
1976 (output_constant_pool_2): Take a fixed_size_mode.
1978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1979 Alan Hayward <alan.hayward@arm.com>
1980 David Sherwood <david.sherwood@arm.com>
1982 * doc/rtl.texi (vec_series): Document.
1983 (const): Say that the operand can be a vec_series.
1984 * rtl.def (VEC_SERIES): New rtx code.
1985 * rtl.h (const_vec_series_p_1): Declare.
1986 (const_vec_series_p): New function.
1987 * emit-rtl.h (gen_const_vec_series): Declare.
1988 (gen_vec_series): Likewise.
1989 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1990 (gen_vec_series): Likewise.
1991 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1992 * simplify-rtx.c (simplify_unary_operation): Handle negations
1994 (simplify_binary_operation_series): New function.
1995 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1996 (test_vector_ops_series): New function.
1997 (test_vector_ops): Call it.
1998 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1999 gen_const_vec_series.
2000 (altivec_lvsr): Likewise.
2001 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2003 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2004 Alan Hayward <alan.hayward@arm.com>
2005 David Sherwood <david.sherwood@arm.com>
2007 * doc/rtl.texi (const): Update description of address constants.
2008 Say that vector constants are allowed too.
2009 * common.md (E, F): Use CONSTANT_P instead of checking for
2011 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2012 checking for CONST_VECTOR.
2013 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2015 * expr.c (expand_expr_real_2): Check for vector modes instead
2016 of checking for CONST_VECTOR.
2017 * rtl.h (const_vec_p): New function.
2018 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2019 (unwrap_const_vec_duplicate): Handle them here too.
2021 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2022 David Malcolm <dmalcolm@redhat.com>
2023 Alan Hayward <alan.hayward@arm.com>
2024 David Sherwood <david.sherwood@arm.com>
2026 * rtl.h (vec_duplicate_p): New function.
2027 * selftest-rtl.c (assert_rtx_eq_at): New function.
2028 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2029 (assert_rtx_eq_at): Declare.
2030 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2031 * selftest-run-tests.c (selftest::run_tests): Call it.
2032 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2033 (simplify_unary_operation_1): Recursively handle vector duplicates.
2034 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2036 (simplify_subreg): Handle subregs of vector duplicates.
2037 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2038 (selftest::simplify_rtx_c_tests): New functions.
2040 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2041 Alan Hayward <alan.hayward@arm.com>
2042 David Sherwood <david.sherwood@arm.com>
2044 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2045 (gen_vec_duplicate): Likewise.
2046 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2048 (gen_const_vector): ...here.
2049 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2050 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2051 whose elements are all equal.
2052 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2053 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2054 (simplify_relational_operation): Likewise.
2055 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2057 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2058 (aarch64_expand_vector_init): Likewise.
2059 * config/arm/arm.c (neon_vdup_constant): Likewise.
2060 (neon_expand_vector_init): Likewise.
2061 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2062 (arm_block_set_unaligned_vect): Likewise.
2063 (arm_block_set_aligned_vect): Likewise.
2064 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2065 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2066 (expand_vec_perm_even_odd_pack): Likewise.
2067 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2068 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2069 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2070 gen_const_vec_duplicate.
2071 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2072 * config/mips/mips.c (mips_gen_const_int_vector): Use
2073 gen_const_vec_duplicate.
2074 (mips_expand_vector_init): Use CONST0_RTX.
2075 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2076 (define_split): Use gen_const_vec_duplicate.
2077 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2078 (define_split): Use gen_const_vec_duplicate.
2079 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2080 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2081 * config/spu/spu.c (spu_const): Likewise.
2083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2084 Alan Hayward <alan.hayward@arm.com>
2085 David Sherwood <david.sherwood@arm.com>
2087 * combine.c (can_change_dest_mode): Reject changes in
2088 REGMODE_NATURAL_SIZE.
2090 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2092 * configure.ac (--enable-libssp): New.
2093 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2094 trying to determine target-specific default. Adjust indentation.
2095 * configure: Regenerated.
2096 * doc/install.texi (Configuration): Expand --disable-libssp
2099 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2101 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2104 2017-10-31 Martin Jambor <mjambor@suse.cz>
2107 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2109 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2111 * auto-profile.c (autofdo_source_profile::read): Use
2112 UNKNOWN_LOCATION rather than 0.
2113 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2114 (warning_at): ...this overload.
2115 (warning_at_rich_loc_n): Rename to...
2116 (warning_n): ...this overload.
2117 (error_at_rich_loc): Rename to...
2118 (error_at): ...this overload.
2119 (pedwarn_at_rich_loc): Rename to...
2120 (pedwarn): ...this overload.
2121 (permerror_at_rich_loc): Rename to...
2122 (permerror): ...this overload.
2123 (inform_at_rich_loc): Rename to...
2124 (inform): ...this overload.
2125 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2126 (diagnostic_n_impl_richloc): Rename to...
2127 (diagnostic_n_impl): ...this rich_location *-based decl.
2128 (inform_at_rich_loc): Rename to...
2129 (inform): ...this, and add an assertion.
2130 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2131 (warning_at_rich_loc): Rename to...
2132 (warning_at): ...this, and add an assertion.
2133 (warning_at_rich_loc_n): Rename to...
2134 (warning_n): ...this, and add an assertion.
2135 (warning_n): Update location_t-based implementation for removal of
2136 location_t-based diagnostic_n_impl.
2137 (pedwarn_at_rich_loc): Rename to...
2138 (pedwarn): ...this, and add an assertion.
2139 (permerror_at_rich_loc): Rename to...
2140 (permerror): ...this, and add an assertion.
2141 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2142 (error_at_rich_loc): Rename to...
2143 (error_at): ...this, and add an assertion.
2144 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2145 (driver::do_spec_on_infiles): Likewise.
2146 * substring-locations.c (format_warning_va): Update for renaming
2147 of inform_at_rich_loc.
2149 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2151 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2152 _Float<N>X built-in functions so that the variant without the
2153 "__builtin_" prefix is only enabled for the GNU C and Objective C
2154 languages when they are in non-strict ANSI/ISO mode.
2155 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2156 * target.def (floatn_builtin_p): Add a target hook to control
2157 whether _Float<N> and _Float<N>X built-in functions without the
2158 "__builtin_" prefix are enabled, and return true for C and
2159 Objective C in the default hook. Include langhooks.h in
2161 * targhooks.h (default_floatn_builtin_p): Likewise.
2162 * targhooks.c (default_floatn_builtin_p): Likewise.
2163 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2164 floatn_builtin_p target hook.
2165 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2167 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2168 Eric Botcazou <ebotcazou@adacore.com>
2170 PR rtl-optimization/81803
2171 * lra-constraints.c (curr_insn_transform): Also reload the whole
2172 register for a strict subreg no wider than a word if this is for
2173 a WORD_REGISTER_OPERATIONS target.
2175 2017-10-31 Jason Merrill <jason@redhat.com>
2177 * gdbinit.in: Skip over inlines from timevar.h.
2179 2017-10-31 Martin Liska <mliska@suse.cz>
2181 * doc/gcov.texi: Document new option.
2182 * gcov.c (print_usage): Likewise print it.
2183 (process_args): Support the argument.
2184 (format_count): New function.
2185 (format_gcov): Use the function.
2187 2017-10-31 Martin Liska <mliska@suse.cz>
2189 * gcov.c (struct name_map): do not use typedef.
2190 Define operator== and operator<.
2191 (name_search): Remove.
2192 (name_sort): Remove.
2193 (main): Do not allocate names.
2194 (process_file): Add vertical space.
2195 (generate_results): Use std::find.
2196 (release_structures): Do not release memory.
2197 (find_source): Use std::find.
2199 2017-10-31 Martin Liska <mliska@suse.cz>
2201 * gcov.c (struct line_info): Remove it's typedef.
2202 (line_info::line_info): Add proper ctor.
2203 (line_info::has_block): Do not use a typedef.
2204 (struct source_info): Do not use typedef.
2205 (circuit): Likewise.
2206 (get_cycles_count): Likewise.
2207 (output_intermediate_file): Iterate via vector iterator.
2208 (add_line_counts): Use std::vector methods.
2209 (accumulate_line_counts): Likewise.
2210 (output_lines): Likewise.
2212 2017-10-31 Martin Liska <mliska@suse.cz>
2214 * gcov.c (struct source_info): Remove typedef.
2215 (source_info::source_info): Add proper ctor.
2216 (accumulate_line_counts): Use struct, not it's typedef.
2217 (output_gcov_file): Likewise.
2218 (output_lines): Likewise.
2219 (main): Do not allocate an array.
2220 (output_intermediate_file): Use size of vector container.
2221 (process_file): Resize the vector.
2222 (generate_results): Do not preallocate, use newly added vector
2224 (release_structures): Do not release sources.
2225 (find_source): Use vector methods.
2226 (add_line_counts): Do not use typedef.
2228 2017-10-31 Martin Liska <mliska@suse.cz>
2230 * doc/gcov.texi: Document that.
2231 * gcov.c (add_line_counts): Mark lines with a non-executed
2233 (output_line_beginning): Handle such lines.
2234 (output_lines): Pass new argument.
2235 (output_intermediate_file): Print it in intermediate format.
2237 2017-10-31 Martin Liska <mliska@suse.cz>
2239 * color-macros.h: New file.
2240 * diagnostic-color.c: Factor out color related to macros to
2242 * doc/gcov.texi: Document -k option.
2243 * gcov.c (INCLUDE_STRING): Include string.h.
2244 (print_usage): Add -k option.
2245 (process_args): Parse it.
2246 (pad_count_string): New function.
2247 (output_line_beginning): Likewise.
2248 (DEFAULT_LINE_START): New macro.
2249 (output_lines): Support color output.
2251 2017-10-31 Martin Liska <mliska@suse.cz>
2253 PR gcov-profile/82633
2254 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2255 their interaction with GCOV infrastructure.
2256 * configure.ac: Add -fkeep-{inline,static}-functions to
2258 * configure: Regenerate.
2260 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2263 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2265 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2268 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2269 into a register if it does not fit into an immediate offset field.
2271 2017-10-31 Olivier Hainque <hainque@adacore.com>
2273 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2275 2017-10-31 Julia Koval <julia.koval@intel.com>
2277 * config.gcc: Add gfniintrin.h.
2278 * config/i386/gfniintrin.h: New.
2279 * config/i386/i386-builtin-types.def
2280 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2281 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2282 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2283 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2284 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2285 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2286 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2287 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2288 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2289 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2290 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2291 * config/i386/immintrin.h: Include gfniintrin.h.
2292 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2294 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2296 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2298 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2300 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2301 (arm_ashldi3_1bit): Remove pattern.
2302 (ashrdi3): Remove shift by 1 expansion.
2303 (arm_ashrdi3_1bit): Remove pattern.
2304 (lshrdi3): Remove shift by 1 expansion.
2305 (arm_lshrdi3_1bit): Remove pattern.
2306 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2307 cost of ashldi3 by 1.
2308 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2309 (<shift>di3_neon): Likewise.
2311 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2313 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2314 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2315 and (*aarch64_simd_mov<VQ:mode>).
2316 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2317 pattern alternative.
2318 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2319 attributes to match pattern alternative.
2321 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2323 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2324 * config/rs6000/emmintrin.h: New file.
2325 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2327 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2329 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2330 * (movdi_vfp_cortexa8): Remove pattern.
2332 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2334 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2336 (Specific, alpha*-dec-osf5.1): Remove.
2337 (Specific, mips-sgi-irix5): Remove.
2338 (Specific, mips-sgi-irix6): Remove.
2340 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2343 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2344 arguments to clear_bit_region_be.
2346 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2348 * gimplify.c: Include memmodel.h.
2350 2017-10-30 Martin Jambor <mjambor@suse.cz>
2352 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2353 condition whether loop should be executed at all.
2355 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2357 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2358 gimple folding of vec_madd() intrinsics.
2359 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2360 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2361 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2363 2017-10-30 Richard Biener <rguenther@suse.de>
2365 PR tree-optimization/82762
2367 2017-10-23 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/82129
2371 2017-08-01 Richard Biener <rguenther@suse.de>
2373 PR tree-optimization/81181
2374 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2375 (compute_antic): ... end of iteration here.
2377 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2379 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2380 -std=iso9899:2017 and -std=gnu17.
2381 * doc/standards.texi (C Language): Document C17 support.
2382 * doc/cpp.texi (Overview): Mention -std=c17.
2383 (Standard Predefined Macros): Document C11 and C17 values of
2384 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2385 * doc/extend.texi (Inline): Do not list individual options for
2386 standards newer than C99.
2387 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2389 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2392 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2394 * asan.c (asan_finish_file): Align asan globals array by shadow
2397 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2400 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2401 (struct store_immediate_info): Add bitregion_start and bitregion_end
2403 (store_immediate_info::store_immediate_info): Add brs and bre
2404 arguments and initialize bitregion_{start,end} from those.
2405 (struct merged_store_group): Add bitregion_start, bitregion_end,
2406 align_base and mask fields. Drop unnecessary struct keyword from
2407 struct store_immediate_info. Add do_merge method.
2408 (clear_bit_region_be): Use memset instead of loop storing zeros.
2409 (merged_store_group::do_merge): New method.
2410 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2411 stores as long as the surrounding bitregions have no gaps.
2412 (merged_store_group::merge_overlapping): Use do_merge.
2413 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2414 is byte aligned, rather than requiring that start and width are
2415 byte aligned. Drop unnecessary struct keyword from
2416 struct store_immediate_info. Allocate and populate also mask array.
2417 Make start of the arrays relative to bitregion_start rather than
2418 start and size them according to bitregion_{end,start} difference.
2419 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2420 struct store_immediate_info.
2421 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2422 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2423 struct keyword from struct store_immediate_info.
2424 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2425 between stores as long as the surrounding bitregions have no gaps.
2427 (struct split_store): Add orig non-static data member.
2428 (split_store::split_store): Initialize orig to false.
2429 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2430 if there is exactly a single original stmt. Change stmts argument
2431 to pointer from reference, if NULL, don't push anything to it. Add
2432 first argument, use it to optimize skipping over orig stmts that
2433 are known to be before bitpos already. Simplify.
2434 (split_group): Return unsigned int count how many stores are or
2435 would be needed rather than a bool. Add allow_unaligned argument.
2436 Change split_stores argument from reference to pointer, if NULL,
2437 only do a dry run computing how many stores would be produced.
2438 Rewritten algorithm to use both alignment and misalign if
2439 !allow_unaligned and handle bitfield stores with gaps.
2440 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2441 from bitregion_start instead of start. Compute allow_unaligned
2442 here, if true, do 2 split_group dry runs to compute which one
2443 produces fewer stores and prefer aligned if equal. Punt if
2444 new count is bigger or equal than original before emitting any
2445 statements, rather than during that. Remove no longer needed
2446 new_ssa_names tracking. Replace num_stmts with
2447 split_stores.length (). Use 32-bit stack allocated entries
2448 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2449 if possible. Handle bitfields with gaps.
2450 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2451 Compute bitregion_{start,end} for the stores and construct
2452 store_immediate_info with that. Formatting fixes.
2454 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2457 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2458 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2460 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2462 * gimplify.c: Include tm_p.h.
2464 * common.opt (gcoff): Re-add as ignored option.
2465 (gcoff1, gcoff2, gcoff3): Likewise.
2467 * Makefile.in (OBJS): Delete sdbout.o.
2468 (GTFILES): Delete $(srcdir)/sdbout.c.
2469 * debug.h: Delete sdb_debug_hooks.
2470 * final.c: Delete sdbout.h include.
2471 (final_scan_insn): Delete SDB_DEBUG check.
2472 (rest_of_clean_state): Likewise.
2473 * output.h: Delete sdb_begin_function_line.
2476 * toplev.c: Delete sdbout.h include.
2477 (process_options): Delete SDB_DEBUG check.
2478 * tree-core.h (tree_type_common): Delete pointer field of
2480 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2481 TYPE_SYMTAB_POINTER.
2482 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2483 (TYPE_SYMTAB_IS_POINTER): Delete.
2484 (TYPE_SYMTAB_IS_DIE): Renumber.
2485 * xcoffout.c: Refer to former sdbout.c file.
2486 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2488 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2489 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2490 (Debugging Options): Delete -gcoff.
2491 (-gstabs): Delete SDB reference.
2493 (-gcoff@var{level}): Delete.
2494 * doc/passes.texi (Debugging information output): Delete SDB and
2495 sdbout.c references.
2496 * doc/tm.texi: Regenerate.
2497 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2498 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2500 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2501 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2502 (SDB_DEBUGGING_INFO): Delete.
2503 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2504 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2505 * target.def (output_source_filename): Delete COFF reference.
2507 * common.opt (gcoff): Delete.
2508 (gxcoff+): Update Negative chain.
2509 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2511 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2512 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2513 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2515 (expand_function_start): Change sdb reference to past tense.
2516 (expand_function_end): Change sdb reference to past tense.
2517 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2518 * opts.c (debug_type_names): Delete coff entry.
2519 (common_handle_option): Delete OPT_gcoff case.
2520 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2522 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2523 * config/cris/cris.h: Delete SDB reference in comment.
2524 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2525 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2526 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2527 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2529 (ix86_expand_prologue): Likewise.
2530 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2531 * config/ia64/ia64.h: Likewise.
2532 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2533 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2535 * config/mmix/mmix.h: Likewise.
2536 * config/nds32/nds32.c: Likewise.
2537 * config/stormy/storym16.h: Likewise.
2538 * config/visium/visium.h: Likewise.
2539 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2541 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2543 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2544 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2545 FRAME_POINTER_REGNUM point at high end of local var area.
2547 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2549 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2550 Move comment around. Do not reset best_edge for a copiable
2551 destination if the copy would cause a partition change.
2552 (better_edge_p): Remove redundant check.
2554 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2556 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2558 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2560 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2561 for math functions that have _Float<N> and _Float<N>X variants.
2562 (mathfn_built_in_2): Add support for math functions that have
2563 _Float<N> and _Float<N>X variants.
2564 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2565 (expand_builtin_mathfn_ternary): Add support for fma with
2566 _Float<N> and _Float<N>X variants.
2567 (expand_builtin): Likewise.
2568 (fold_builtin_3): Likewise.
2569 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2570 create math function _Float<N> and _Float<N>X variants as external
2572 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2573 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2574 the __builtin_ prefix and if not strict ansi, without the prefix.
2575 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2576 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2577 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2578 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2579 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2580 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2581 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2582 function signatures for fma _Float<N> and _Float<N>X variants.
2583 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2584 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2585 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2586 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2587 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2588 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2589 * gencfn-macros.c (print_case_cfn): Add support for math functions
2590 that have _Float<N> and _Float<N>X variants.
2591 (print_define_operator_list): Likewise.
2592 (fltfn_suffixes): Likewise.
2594 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2595 for math functions that have _Float<N> and _Float<N>X variants.
2596 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2597 and _Float<N>X variants.
2598 (COPYSIGN): Likewise.
2601 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2602 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2604 (integer_valued_read_call_p): Likewise.
2605 * fold-const-call.c (fold_const_call_ss): Likewise.
2606 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2607 _Float<N> and _Float<N>X variants.
2608 (fold_const_call_ssss): Add support for fma _Float<N> and
2609 _Float<N>X variants.
2610 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2611 support for copysign and fma _Float<N> and _Float<N>X variants.
2612 (backprop::process_builtin_call_use): Likewise.
2613 * tree-call-cdce.c (can_test_argument_range); Add support for
2614 sqrt _Float<N> and _Float<N>X variants.
2615 (edom_only_function): Likewise.
2616 (get_no_error_domain): Likewise.
2617 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2618 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2619 copysign _Float<N> and _Float<N>X variants.
2620 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2621 handled by machine independent code.
2622 (FMAF128): Likewise.
2623 * doc/cpp.texi (Common Predefined Macros): Document defining
2624 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2625 fma _Float<N> and _Float<N>X variants.
2627 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2630 * config/i386/i386-modes.def (CCFPU): Remove definition.
2631 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2632 (ix86_cc_modes_compatible): Ditto.
2633 (ix86_expand_carry_flag_compare): Ditto.
2634 (ix86_expand_int_movcc): Ditto.
2635 (ix86_expand_int_addcc): Ditto.
2636 (ix86_reverse_condition): Ditto.
2637 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2638 Return true/false for unordered/ordered fp comparisons.
2639 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2640 (ix86_prepare_fp_compare_args): Update for rename.
2641 (ix86_expand_fp_compare): Update for rename. Generate unordered
2642 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2643 (ix86_expand_sse_compare_and_jump): Ditto.
2644 * config/i386/predicates.md (fcmov_comparison_operator):
2645 Remove CCFPU mode handling.
2646 (ix86_comparison_operator): Ditto.
2647 (ix86_carry_flag_operator): Ditto.
2648 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2649 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2650 (*cmpu<mode>_cc_i387): Ditto.
2651 (FPCMP): Remove mode iterator.
2652 (unord): Remove mode attribute.
2653 (unord_subst): New define_subst transformation
2654 (unord): New define_subst attribute.
2656 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2657 (*cmpi<unord>xf_i387): Ditto.
2658 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2659 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2660 using unord_subst transformation.
2661 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2662 (round_saeonly): Also handle CCFP mode.
2663 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2664 Remove UNSPEC_SAHF unspec handling.
2666 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2668 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2670 2017-10-27 Jeff Law <law@redhat.com>
2672 * gimple-ssa-sprintf.c: Include domwalk.h.
2673 (class sprintf_dom_walker): New class, derived from dom_walker.
2674 (sprintf_dom_walker::before_dom_children): New function.
2675 (struct call_info): Moved into sprintf_dom_walker class
2676 (compute_formath_length, handle_gimple_call): Likewise.
2677 (sprintf_length::execute): Call the dominator walker rather
2678 than walking the statements.
2680 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2681 gimple statement locations.
2682 (check_array_bounds): Corresponding changes. Get the statement's
2683 location directly from wi->stmt.
2685 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2688 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2690 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2692 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2693 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2695 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2698 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2699 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2700 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2701 maybe_get_pool_constant.
2702 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2705 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2707 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2708 2.26 caveat. Update gas and gld versions.
2709 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2712 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2714 * cgraph.h (set_malloc_flag): Declare.
2715 * cgraph.c (set_malloc_flag_1): New function.
2716 (set_malloc_flag): Likewise.
2717 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2718 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2720 (read_ipa_call_summary): Add support for reading is_return_callee.
2721 (write_ipa_call_summary): Stream is_return_callee.
2722 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2723 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2724 ipa-prop.h, ipa-fnsummary.h.
2725 (pure_const_names): Change to static.
2726 (malloc_state_e): Define.
2727 (malloc_state_names): Define.
2728 (funct_state_d): Add field malloc_state.
2729 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2730 (check_retval_uses): New function.
2731 (malloc_candidate_p): Likewise.
2732 (analyze_function): Add support for malloc attribute.
2733 (pure_const_write_summary): Stream malloc_state.
2734 (pure_const_read_summary): Add support for reading malloc_state.
2735 (dump_malloc_lattice): New function.
2736 (propagate_malloc): New function.
2737 (warn_function_malloc): New function.
2738 (ipa_pure_const::execute): Call propagate_malloc and
2739 ipa_free_fn_summary.
2740 (pass_local_pure_const::execute): Add support for malloc attribute.
2741 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2742 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2744 2017-10-27 Martin Liska <mliska@suse.cz>
2746 PR gcov-profile/82457
2747 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2748 for fork-like functions to be properly instrumented.
2750 2017-10-27 Richard Biener <rguenther@suse.de>
2753 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2754 info when we redirected EH.
2756 2017-10-26 Michael Collison <michael.collison@arm.com>
2758 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2760 (<optab>_trunchf<GPI:mode>2: New pattern.
2761 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2762 * config/aarch64/iterators.md (wv): New mode attribute.
2763 (vf, VF): New mode attributes.
2764 (vgp, VGP): New mode attributes.
2765 (s): Update attribute with SImode and DImode prefixes.
2767 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2769 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2770 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2771 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2772 (nios2_option_overide): Initialize it. Don't allow R0-relative
2773 addressing with PIC.
2774 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2775 (nios2_symbolic_constant_p): Likewise.
2776 (nios2_legitimate_address_p): Likewise.
2777 (nios2_r0rel_section_name_p): New.
2778 (nios2_symbol_ref_in_r0rel_data_p): New.
2779 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2780 (r0rel_constant_p): New.
2781 (nios2_print_operand_address): Handle r0rel_constant_p.
2782 (nios2_cdx_narrow_form_p): Likewise.
2783 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2784 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2785 (Nios II Options): Document -mr0rel-sec.
2787 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2789 * config/nios2/nios2.c: Include xregex.h.
2790 (nios2_gprel_sec_regex): New.
2791 (nios2_option_overide): Initialize it. Don't allow GP-relative
2792 addressing with PIC.
2793 (nios2_small_section_name_p): Check for regex match.
2794 * config/nios2/nios2.opt (mgprel-sec=): New option.
2795 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2796 (Nios II Options): Document -mgprel-sec.
2798 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2800 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2802 2017-10-26 Tom de Vries <tom@codesourcery.com>
2804 PR tree-optimization/82707
2805 * gimple.c (gimple_copy): Fix unsharing of
2806 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2808 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2810 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2811 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2812 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2813 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2814 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2815 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2816 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2817 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2818 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2819 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2820 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2821 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2822 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2823 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2824 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2825 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2826 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2828 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2830 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2832 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2833 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2834 warning to rs6000.c. Remove the Undocumented flag, since it has
2836 (-mabi=ibmlongdouble): Likewise.
2837 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2838 already set, set the default format for long double.
2839 (rs6000_debug_reg_global): Print whether long double is IBM or
2841 (rs6000_option_override_internal): Rework setting long double
2842 format. Only warn if the user is changing the long double default
2843 and they did not use -Wno-psabi.
2844 * doc/invoke.texi (PowerPC options): Update the documentation for
2845 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2847 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2848 Alan Hayward <alan.hayward@arm.com>
2849 David Sherwood <david.sherwood@arm.com>
2851 * rtl.h (wider_subreg_mode): New function.
2852 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2853 rather than an unsigned int *.
2854 * ira-color.c (regno_max_ref_width): Replace with...
2855 (regno_max_ref_mode): ...this new variable.
2856 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2857 Use wider_subreg_mode.
2858 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2859 rather than an unsigned int *.
2860 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2861 (process_alt_operands): Likewise.
2862 (invariant_p): Likewise.
2863 * lra-spills.c (assign_mem_slot): Likewise.
2864 (add_pseudo_to_slot): Likewise.
2865 * lra.c (collect_non_operand_hard_regs): Likewise.
2866 (add_regs_to_insn_regno_info): Likewise.
2867 * reload1.c (regno_max_ref_width): Replace with...
2868 (regno_max_ref_mode): ...this new variable.
2869 (reload): Update accordingly. Update call to
2870 ira_sort_regnos_for_alter_reg.
2871 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2872 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2873 (scan_paradoxical_subregs): Likewise.
2875 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2877 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2878 (aarch64_frame): Add emit_frame_chain boolean.
2879 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2880 Move eh_return case to aarch64_layout_frame.
2881 (aarch64_layout_frame): Initialize emit_frame_chain.
2882 (aarch64_expand_prologue): Use emit_frame_chain.
2884 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2886 * config/aarch64/aarch64.c (aarch64_layout_frame):
2887 Ensure LR is always stored at the bottom of the callee-saves.
2888 Remove rarely used frame layout which saves callee-saves at top of
2889 frame, so the store of LR can be used as a valid probe in all cases.
2891 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2893 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2894 Improve unaligned TImode/TFmode base/offset split.
2896 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2897 Alan Hayward <alan.hayward@arm.com>
2898 David Sherwood <david.sherwood@arm.com>
2900 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2901 * combine.c (find_single_use_1): Likewise.
2902 (expand_field_assignment): Likewise.
2903 (move_deaths): Likewise.
2904 * lra-constraints.c (simplify_operand_subreg): Likewise.
2905 (curr_insn_transform): Likewise.
2906 * lra.c (collect_non_operand_hard_regs): Likewise.
2907 (add_regs_to_insn_regno_info): Likewise.
2908 * rtlanal.c (reg_referenced_p): Likewise.
2909 (covers_regno_no_parallel_p): Likewise.
2911 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2913 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2914 Don't print any bits outside the precision of the value.
2915 * wide-int.cc (test_printing): Add some new tests.
2917 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2919 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2920 supports -xbrace_comment option.
2921 * configure: Regenerate.
2922 * config.in: Regenerate.
2923 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2924 (ASM_CPU_SPEC): Use it.
2926 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2928 * target.def (static_rtx_alignment): New hook.
2929 * targhooks.h (default_static_rtx_alignment): Declare.
2930 * targhooks.c (default_static_rtx_alignment): New function.
2931 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2932 * doc/tm.texi: Regenerate.
2933 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2934 instead of targetm.constant_alignment. Remove call to
2936 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2937 (cris_preferred_mininum_alignment): New function, split out from...
2938 (cris_constant_alignment): ...here.
2939 (cris_static_rtx_alignment): New function.
2940 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2942 (ix86_constant_alignment): ...here.
2943 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2944 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2945 (mmix_static_rtx_alignment): New function.
2946 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2947 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2949 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2952 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2953 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2955 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2957 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2958 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2959 mark as a sign-extending load.
2960 (local_pic_load_u): Define.
2962 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2965 * fold-const.c (operand_equal_for_comparison_p): Also return true
2966 if ARG0 is a simple variant of ARG1 with narrower precision.
2967 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2969 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2971 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2973 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2974 scatter_static, scatter_per_elt.
2975 * x86-tune-costs.h: Add new cost entries.
2977 2017-10-25 Richard Biener <rguenther@suse.de>
2979 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2980 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2981 class pass_fre): Move to ...
2982 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2983 class pass_fre): ... here and adjust for statistics.
2985 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2988 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2989 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2990 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2991 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2994 2017-10-25 Richard Biener <rguenther@suse.de>
2996 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2997 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2998 eliminate_push_avail, eliminate_insert): Move inside...
2999 (class eliminate_dom_walker): ... this class in preparation
3001 (fini_eliminate): Remove by merging with ...
3002 (eliminate): ... this function. Adjust for class changes.
3003 (pass_pre::execute): Remove fini_eliminate call.
3004 (pass_fre::execute): Likewise.
3006 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3009 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3010 (VPERMI2, VPERMI2I): New mode iterators.
3011 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3012 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3014 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3015 mode iterator. Remove 3 old define_insn patterns.
3016 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3017 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3018 VPERMI2 mode iterator, remove the other two expanders.
3019 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3020 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3021 instructions, remove the other two patterns.
3022 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3023 mode iterator, remove the other two patterns.
3024 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3025 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3026 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3027 and adjust argument order accordingly.
3028 (ix86_expand_vec_perm): Adjust caller.
3029 (expand_vec_perm_1): Likewise.
3030 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3031 (expand_vec_perm_vpermt2_vpshub2): ... this.
3032 (ix86_expand_vec_perm_const_1): Adjust caller.
3033 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3036 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3037 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3038 (vec_shl_<mode>): Remove unused expander.
3039 (avx512bw_<shift_insn><mode>3): New define_insn.
3040 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3041 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3043 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3046 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3049 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3051 PR rtl-optimization/82396
3052 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3053 (autopref_multipass_init): Simplify initialization.
3054 (autopref_rank_data): Simplify sort order.
3055 * gcc/sched-int.h (autopref_multipass_data_): Remove
3056 multi_mem_insn_p, min_offset and max_offset.
3058 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3061 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3062 Check special value of flag_omit_frame_pointer.
3063 (aarch64_can_eliminate): Likewise.
3064 (aarch64_override_options_after_change_1): Simplify handling of
3065 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3067 2017-10-24 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/82697
3070 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3071 zero for conditional load and unconditional store.
3073 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3075 * doc/install.texi: Document bootstrap-cet.
3077 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3080 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3081 ENDBR instruction at function entrance if function is only
3084 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3087 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3088 patterns to better describe from which operation the CF is computed.
3089 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3090 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3091 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3092 is 0, use _0 suffixed expanders instead of emitting a comparison
3095 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3097 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3098 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3100 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3103 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3104 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3105 * loop-iv.c (iv_get_reaching_def): Likewise.
3106 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3107 variable is promoted and the partition contains undefined values.
3109 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3111 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3113 (nios2_address_cost): Define.
3114 (nios2_legitimize_address): Recognize (exp + constant) directly.
3115 (TARGET_ADDRESS_COST): Define.
3117 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3119 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3120 (nios2_symbolic_memory_operand_p): Declare.
3121 (nios2_split_large_constant): Declare.
3122 (nios2_split_symbolic_memory_operand): Declare.
3123 * config/nios2/nios2.c: Adjust includes.
3124 (nios2_symbolic_constant_allowed): New.
3125 (nios2_symbolic_constant_p): New.
3126 (nios2_plus_symbolic_constant_p): New.
3127 (nios2_valid_addr_expr_p): Recognize addresses involving
3129 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3130 (nios2_symbolic_memory_operand_p): New.
3131 (nios2_large_constant_p): New.
3132 (nios2_split_large_constant): New.
3133 (nios2_split_plus_large_constant): New.
3134 (nios2_split_symbolic_memory_operand): New.
3135 (nios2_legitimize_address): Code refactoring. Handle addresses
3136 involving symbolic constants.
3137 (nios2_emit_move_sequence): Likewise.
3138 (nios2_print_operand): Improve error output.
3139 (nios2_print_operand_address): Handle LO_SUM.
3140 (nios2_cdx_narrow_form_p): Likewise.
3141 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3142 operands involving symbolic constants.
3143 (movhi_internal, movsi_internal): Likewise.
3144 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3145 (extendhisi2, extendqi<mode>2): Likewise.
3147 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3149 * tree-pass.h (PROP_rtl_split_insns): Define.
3150 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3152 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3154 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3156 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3159 * target.def (const_not_ok_for_debug_p): Default to
3160 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3161 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3162 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3163 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3164 which targetm.const_not_ok_for_debug_p returned true.
3165 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3167 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3169 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3170 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3171 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3173 (ix86_const_not_ok_for_debug_p): New function.
3174 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3175 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3177 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3180 * system.h: Conditionally include "unique-ptr.h" if
3181 INCLUDE_UNIQUE_PTR is defined.
3182 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3183 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3185 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3187 * config/rl78/rl78.md: New define_expand "subdi3".
3189 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3192 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3193 DF_REF_INSN if DF_REF_INSN_INFO is false.
3195 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3197 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3198 xmm_move instead of sse_move.
3199 (sse_store_index): New function.
3200 (ix86_register_move_cost): Be more sensible about mismatch stall;
3201 model AVX moves correctly; make difference between sse->integer and
3203 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3204 moves; make difference between SSE and AVX.
3205 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3206 and zmm_move. Increase size of sse load and store tables;
3207 add unaligned load and store tables; add ssemmx_to_integer.
3208 * x86-tune-costs.h: Update all entries according to real
3209 move latencies from Agner Fog's manual and chip documentation.
3211 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3214 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3215 * config/i386/constraints.md (Wf): New constraint.
3216 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3217 (cmp<dwi>_doubleword): Removed.
3218 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3219 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3220 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3221 expand with cmp<dwi>_doubleword. For LTU and GEU use
3222 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3224 * common.opt (gcolumn-info): Enable by default.
3225 * doc/invoke.texi (gcolumn-info): Document new default.
3227 2017-10-23 Richard Biener <rguenther@suse.de>
3229 PR tree-optimization/82672
3230 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3231 Fold the stmt if we propagated into it.
3233 2017-10-23 Richard Biener <rguenther@suse.de>
3235 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3236 (bitmap_remove_expr_from_set): ... this. All callers call this
3237 for non-constant values.
3238 (bitmap_set_subtract): Rename to...
3239 (bitmap_set_subtract_expressions): ... this. Adjust and
3241 (bitmap_set_contains_value): Remove superfluous check.
3242 (bitmap_set_replace_value): Inline into single caller ...
3243 (bitmap_value_replace_in_set): ... here and simplify.
3244 (dependent_clean): Merge into ...
3245 (clean): ... this using an overload. Adjust.
3246 (prune_clobbered_mems): Adjust.
3247 (compute_antic_aux): Likewise.
3248 (compute_partial_antic_aux): Likewise.
3250 2017-10-23 Richard Biener <rguenther@suse.de>
3252 PR tree-optimization/82129
3254 2017-08-01 Richard Biener <rguenther@suse.de>
3256 PR tree-optimization/81181
3257 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3258 (compute_antic): ... end of iteration here.
3260 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3262 * target.def (starting_frame_offset): New hook.
3263 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3264 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3265 * doc/tm.texi.in: Regenerate.
3266 * hooks.h (hook_hwi_void_0): Declare.
3267 * hooks.c (hook_hwi_void_0): New function.
3268 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3269 STARTING_FRAME_OFFSET.
3270 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3271 * reload1.c (reload): Likewise.
3272 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3273 instead of STARTING_FRAME_OFFSET.
3274 * function.c (try_fit_stack_local): Likewise.
3275 (assign_stack_local_1): Likewise
3276 (instantiate_virtual_regs): Likewise.
3277 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3278 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3279 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3280 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3281 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3282 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3283 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3284 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3285 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3286 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3287 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3288 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3289 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3290 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3291 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3292 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3293 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3294 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3295 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3296 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3297 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3298 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3299 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3300 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3301 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3302 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3303 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3304 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3305 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3306 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3307 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3308 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3309 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3310 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3311 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3312 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3313 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3314 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3315 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3316 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3317 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3318 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3319 return a HOST_WIDE_INT.
3320 (avr_builtin_setjmp_frame_value): Use it instead of
3321 STARTING_FRAME_OFFSET.
3322 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3323 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3324 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3326 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3327 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3328 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3329 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3330 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3331 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3332 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3333 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3334 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3335 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3336 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3337 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3339 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3340 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3341 * config/mips/mips.c (mips_compute_frame_info): Refer to
3342 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3343 (mips_starting_frame_offset): New function.
3344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3345 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3346 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3347 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3348 and return a HOST_WIDE_INT.
3349 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3350 (mmix_initial_elimination_offset): Refer to
3351 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3352 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3353 * config/pa/pa.c (pa_starting_frame_offset): New function.
3354 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3355 (pa_expand_prologue): Likewise.
3356 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3357 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3358 !FRAME_GROWS_DOWNWARD handling to...
3359 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3360 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3361 !FRAME_GROWS_DOWNWARD handling to...
3362 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3363 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3364 !FRAME_GROWS_DOWNWARD handling to...
3365 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3366 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3368 (rs6000_starting_frame_offset): New function.
3369 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3370 !FRAME_GROWS_DOWNWARD handling to...
3371 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3372 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3373 !FRAME_GROWS_DOWNWARD handling to...
3374 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3375 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3376 !FRAME_GROWS_DOWNWARD handling to...
3377 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3378 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3379 (rs6000_starting_frame_offset): New function.
3380 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3381 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3382 * config/vax/vax.c (vax_starting_frame_offset): New function.
3383 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3384 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3385 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3386 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3387 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3388 * system.h (STARTING_FRAME_OFFSET): Poison.
3390 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3392 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3393 SCALAR_TYPE_MODE instead of TYPE_MODE.
3395 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3396 Alan Hayward <alan.hayward@arm.com>
3397 David Sherwood <david.sherwood@arm.com>
3399 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3401 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3402 Alan Hayward <alan.hayward@arm.com>
3403 David Sherwood <david.sherwood@arm.com>
3405 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3407 2017-10-23 Richard Biener <rguenther@suse.de>
3409 PR tree-optimization/82129
3410 * tree-ssa-pre.c (bitmap_set_and): Remove.
3411 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3412 canonicalizing expressions in the set to those with lowest
3413 ID rather than taking that from the first edge.
3415 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3417 * combine.c (rtx_equal_for_field_assignment_p): Use
3418 byte_lowpart_offset.
3420 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3421 Alan Hayward <alan.hayward@arm.com>
3422 David Sherwood <david.sherwood@arm.com>
3424 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3425 to a SUBREG_PROMOTED_VAR.
3427 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3428 Alan Hayward <alan.hayward@arm.com>
3429 David Sherwood <david.sherwood@arm.com>
3431 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3432 (expand_debug_source_expr): Likewise.
3433 * combine.c (combine_simplify_rtx): Likewise.
3434 * cse.c (fold_rtx): Likewise.
3435 * optabs.c (expand_float): Likewise.
3436 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3437 (simplify_binary_operation_1): Likewise.
3439 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3440 Alan Hayward <alan.hayward@arm.com>
3441 David Sherwood <david.sherwood@arm.com>
3443 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3444 (record_promoted_value): Likewise.
3445 * expr.c (expand_expr_real_2): Likewise.
3446 * ree.c (update_reg_equal_equiv_notes): Likewise.
3447 (combine_set_extension): Likewise.
3448 * rtlanal.c (low_bitmask_len): Likewise.
3449 * simplify-rtx.c (neg_const_int): Likewise.
3450 (simplify_binary_operation_1): Likewise.
3452 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3453 Alan Hayward <alan.hayward@arm.com>
3454 David Sherwood <david.sherwood@arm.com>
3456 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3457 * regcprop.c (maybe_mode_change): Likewise.
3458 * reload1.c (alter_reg): Likewise.
3460 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3462 * inchash.h (inchash::hash::add_wide_int): New function.
3463 * lto-streamer-out.c (hash_tree): Use it.
3465 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3467 * inchash.h (inchash::hash::add_wide_int): Rename to...
3468 (inchash::hash::add_hwi): ...this.
3469 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3470 (polymorphic_call_target_hasher::hash): Likewise.
3471 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3472 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3473 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3474 * lto-streamer-out.c (hash_tree): Likewise.
3475 * optc-save-gen.awk: Likewise.
3476 * tree.c (add_expr): Likewise.
3478 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3481 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3482 for ordered inequality comparisons even with TARGET_IEEE_FP.
3484 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3487 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3488 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3489 Expand with cmp<dwi>_doubleword.
3491 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3493 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3494 List CET intrinsics.
3495 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3496 specific to -fcf-protection option.
3498 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3500 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3501 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3502 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3503 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3504 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3505 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3506 (extra_objs): Add cet.o for Linux/x86 targets.
3507 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3508 * config/i386/cet.c: New file.
3509 * config/i386/cetintrin.h: Likewise.
3510 * config/i386/t-cet: Likewise.
3511 * config/i386/cpuid.h (bit_SHSTK): New.
3512 (bit_IBT): Likewise.
3513 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3514 pass IBT and SHSTK bits.
3515 * config/i386/i386-builtin-types.def
3516 (VOID_FTYPE_UNSIGNED_PVOID): New.
3517 (VOID_FTYPE_UINT64_PVOID): Likewise.
3518 * config/i386/i386-builtin.def: Add CET intrinsics.
3519 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3520 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3521 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3522 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3524 * config/i386/i386.c (rest_of_insert_endbranch): New.
3525 (pass_data_insert_endbranch): Likewise.
3526 (pass_insert_endbranch): Likewise.
3527 (make_pass_insert_endbranch): Likewise.
3528 (ix86_notrack_prefixed_insn_p): Likewise.
3529 (ix86_target_string): Add -mibt, -mshstk flags.
3530 (ix86_option_override_internal): Add flag_cf_protection
3532 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3533 (ix86_print_operand): Add 'notrack' prefix output.
3534 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3535 (ix86_expand_builtin): Expand CET intrinsics.
3536 (x86_output_mi_thunk): Add 'endbranch' instruction.
3537 * config/i386/i386.h (TARGET_IBT): New.
3538 (TARGET_IBT_P): Likewise.
3539 (TARGET_SHSTK): Likewise.
3540 (TARGET_SHSTK_P): Likewise.
3541 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3542 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3543 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3544 (builtin_setjmp_setup): New pattern.
3545 (builtin_longjmp): Likewise.
3546 (rdssp<mode>): Likewise.
3547 (incssp<mode>): Likewise.
3548 (saveprevssp): Likewise.
3549 (rstorssp): Likewise.
3550 (wrss<mode>): Likewise.
3551 (wruss<mode>): Likewise.
3552 (setssbsy): Likewise.
3553 (clrssbsy): Likewise.
3554 (nop_endbr): Likewise.
3555 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3557 * config/i386/immintrin.h: Include <cetintrin.h>.
3558 * config/i386/linux-common.h
3559 (file_end_indicate_exec_stack_and_cet): New prototype.
3560 (TARGET_ASM_FILE_END): New.
3562 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3564 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3565 latencies instead of having separate table; make difference between
3566 integer and float costs.
3567 * i386.h (processor_costs): Remove scalar_stmt_cost,
3568 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3569 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3571 * x86-tune-costs.h: Remove entries which has been removed in
3572 procesor_costs from all tables; make cond_taken_branch_cost
3573 and cond_not_taken_branch_cost COST_N_INSNS based.
3575 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3577 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3579 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3581 * config/i386/i386.md (isa): Remove fma_avx512f.
3582 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3583 <avx512>_fmadd_<mode>_mask3<round_name>,
3584 <avx512>_fmsub_<mode>_mask<round_name>,
3585 <avx512>_fmsub_<mode>_mask3<round_name>,
3586 <avx512>_fnmadd_<mode>_mask<round_name>,
3587 <avx512>_fnmadd_<mode>_mask3<round_name>,
3588 <avx512>_fnmsub_<mode>_mask<round_name>,
3589 <avx512>_fnmsub_<mode>_mask3<round_name>,
3590 <avx512>_fmaddsub_<mode>_mask<round_name>,
3591 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3592 <avx512>_fmsubadd_<mode>_mask<round_name>,
3593 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3594 (*vec_widen_umult_even_v16si<mask_name>,
3595 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3596 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3598 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3600 * extend.texi: Add 'nocf_check' documentation.
3601 * gimple.texi: Add second parameter to
3602 gimple_build_call_from_tree.
3603 * invoke.texi: Add -fcf-protection documentation.
3604 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3606 2017-10-20 Richard Biener <rguenther@suse.de>
3608 PR tree-optimization/82473
3609 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3610 the largest input type.
3612 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3614 * c-attribs.c (handle_nocf_check_attribute): New function.
3615 (c_common_attribute_table): Add 'nocf_check' handling.
3616 * gimple-parser.c: Add second argument NULL to
3617 gimple_build_call_from_tree.
3618 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3619 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3621 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3622 * common.opt: Add fcf-protection flag.
3623 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3624 * flag-types.h: Add enum cf_protection_level.
3625 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3626 Add 'nocf_check' attribute propagation to gimple call.
3627 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3628 (gimple_build_call_from_tree): Update prototype.
3629 (gimple_call_nocf_check_p): New function.
3630 (gimple_call_set_nocf_check): Likewise.
3631 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3632 * ipa-icf.c: Add nocf_check attribute in statement hash.
3633 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3634 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3635 * toplev.c (process_options): Add flag_cf_protection handling.
3637 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3639 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3641 2017-10-20 Richard Biener <rguenther@suse.de>
3643 PR tree-optimization/82603
3644 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3645 remove false predicated stores.
3647 2017-10-20 Richard Biener <rguenther@suse.de>
3649 * graphite-isl-ast-to-gimple.c
3650 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3651 Remove return value and simplify, dump copied stmt after lhs
3653 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3654 Reduce dump verbosity.
3655 (gsi_insert_earliest): Likewise.
3656 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3657 * graphite.c (print_global_statistics): Adjust dumping.
3658 (print_graphite_scop_statistics): Likewise.
3659 (print_graphite_statistics): Do not dump loops here.
3660 (graphite_transform_loops): But here.
3662 2017-10-20 Nicolas Roche <roche@adacore.com>
3664 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3665 * configure: Regenerate.
3667 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3670 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3671 functions when optimizing replace GIMPLE_RETURN stmts with
3672 calls to __builtin_unreachable ().
3675 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3676 for -fsanitize=thread link of executables.
3677 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3678 link of executables.
3681 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3683 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3684 define_insns for logical vector shifts to use VI248_AVX512BW
3685 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3686 condition, useless isa and prefix attributes. Change the first
3687 2 of these define_insns to ...
3688 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3689 define_insn for avx512vl.
3690 (<shift_insn><mode>3): ... and this, new define_insn without
3691 masking for non-avx512vl.
3694 * config/i386/sse.md (*andnot<mode>3,
3695 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3696 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3697 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3698 not applied use empty suffix even for TARGET_AVX512VL.
3699 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3700 is applied, supply evex,evex or evex,evex,evex instead of just
3703 2017-10-20 Julia Koval <julia.koval@intel.com>
3705 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3706 (OPTION_MASK_ISA_GFNI_UNSET): New.
3707 (ix86_handle_option): Handle OPT_mgfni.
3708 * config/i386/cpuid.h (bit_GFNI): New.
3709 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3710 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3711 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3712 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3713 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3714 * config/i386/i386.opt: Add mgfni.
3716 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3718 * config/msp430/msp430.c (msp430_option_override): Disable
3719 -fdelete-null-pointer-checks.
3720 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3722 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3724 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3725 of x87 and SSE instructions.
3727 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3729 * asan.c (create_cond_insert_point): Do not update edge count.
3730 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3731 (afdo_propagate_circuit): Likewise.
3732 (afdo_calculate_branch_prob): Likewise.
3733 (afdo_annotate_cfg): Likewise.
3734 * basic-block.h (struct edge_def): Remove count.
3735 (edge_def::count): New accessor.
3736 * bb-reorder.c (rotate_loop): Update.
3737 (find_traces_1_round): Update.
3738 (connect_traces): Update.
3739 (sanitize_hot_paths): Update.
3740 * cfg.c (unchecked_make_edge): Update.
3741 (make_single_succ_edge): Update.
3742 (check_bb_profile): Update.
3743 (dump_edge_info): Update.
3744 (update_bb_profile_for_threading): Update.
3745 (scale_bbs_frequencies_int): Update.
3746 (scale_bbs_frequencies_gcov_type): Update.
3747 (scale_bbs_frequencies_profile_count): Update.
3748 (scale_bbs_frequencies): Update.
3749 * cfganal.c (connect_infinite_loops_to_exit): Update.
3750 * cfgbuild.c (compute_outgoing_frequencies): Update.
3751 (find_many_sub_basic_blocks): Update.
3752 * cfgcleanup.c (try_forward_edges): Update.
3753 (try_crossjump_to_edge): Update
3754 * cfgexpand.c (expand_gimple_cond): Update
3755 (expand_gimple_tailcall): Update
3756 (construct_exit_block): Update
3757 * cfghooks.c (verify_flow_info): Update
3758 (redirect_edge_succ_nodup): Update
3759 (split_edge): Update
3760 (make_forwarder_block): Update
3761 (duplicate_block): Update
3762 (account_profile_record): Update
3763 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3764 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3765 * cfgloopmanip.c (scale_loop_profile): Update.
3767 (lv_adjust_loop_entry_edge): Update.
3768 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3769 (force_nonfallthru_and_redirect): Update.
3770 (purge_dead_edges): Update.
3771 (rtl_flow_call_edges_add): Update.
3772 * cgraphunit.c (init_lowered_empty_function): Update.
3773 (cgraph_node::expand_thunk): Update.
3774 * gimple-pretty-print.c (dump_probability): Update.
3775 (dump_edge_probability): Update.
3776 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3777 * haifa-sched.c (sched_create_recovery_edges): Update.
3778 * hsa-gen.c (convert_switch_statements): Update.
3779 * ifcvt.c (dead_or_predicable): Update.
3780 * ipa-inline-transform.c (inline_transform): Update.
3781 * ipa-split.c (split_function): Update.
3782 * ipa-utils.c (ipa_merge_profiles): Update.
3783 * loop-doloop.c (add_test): Update.
3784 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3785 * lto-streamer-in.c (input_cfg): Update.
3786 (input_function): Update.
3787 * lto-streamer-out.c (output_cfg): Update.
3788 * modulo-sched.c (sms_schedule): Update.
3789 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3790 * predict.c (maybe_hot_edge_p): Update.
3791 (unlikely_executed_edge_p): Update.
3792 (probably_never_executed_edge_p): Update.
3793 (dump_prediction): Update.
3794 (drop_profile): Update.
3795 (propagate_unlikely_bbs_forward): Update.
3796 (determine_unlikely_bbs): Update.
3797 (force_edge_cold): Update.
3798 * profile.c (compute_branch_probabilities): Update.
3799 * reg-stack.c (better_edge): Update.
3800 * shrink-wrap.c (handle_simple_exit): Update.
3801 * tracer.c (better_p): Update.
3802 * trans-mem.c (expand_transaction): Update.
3803 (split_bb_make_tm_edge): Update.
3804 * tree-call-cdce.c: Update.
3805 * tree-cfg.c (gimple_find_sub_bbs): Update.
3806 (gimple_split_edge): Update.
3807 (gimple_duplicate_sese_region): Update.
3808 (gimple_duplicate_sese_tail): Update.
3809 (gimple_flow_call_edges_add): Update.
3810 (insert_cond_bb): Update.
3811 (execute_fixup_cfg): Update.
3812 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3813 * tree-complex.c (expand_complex_div_wide): Update.
3814 * tree-eh.c (lower_resx): Update.
3815 (unsplit_eh): Update.
3816 (cleanup_empty_eh_move_lp): Update.
3817 * tree-inline.c (copy_edges_for_bb): Update.
3818 (freqs_to_counts): Update.
3819 (copy_cfg_body): Update.
3820 * tree-ssa-dce.c (remove_dead_stmt): Update.
3821 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3822 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3823 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3824 (unloop_loops): Update.
3825 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3826 * tree-ssa-loop-split.c (connect_loops): Update.
3827 (split_loop): Update.
3828 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3829 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3830 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3831 * tree-ssa-reassoc.c (branch_fixup): Update.
3832 * tree-ssa-tail-merge.c (replace_block_by): Update.
3833 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3834 (compute_path_counts): Update.
3835 (update_profile): Update.
3836 (recompute_probabilities): Update.
3837 (update_joiner_offpath_counts): Update.
3838 (estimated_freqs_path): Update.
3839 (freqs_to_counts_path): Update.
3840 (clear_counts_path): Update.
3841 (ssa_fix_duplicate_block_edges): Update.
3842 (duplicate_thread_path): Update.
3843 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3844 (case_bit_test_cmp): Update.
3845 (collect_switch_conv_info): Update.
3846 (gen_inbound_check): Update.
3847 (do_jump_if_equal): Update.
3848 (emit_cmp_and_jump_insns): Update.
3849 * tree-tailcall.c (decrease_profile): Update.
3850 (eliminate_tail_call): Update.
3851 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3852 (vect_do_peeling): Update.
3853 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3854 * ubsan.c (ubsan_expand_null_ifn): Update.
3855 (ubsan_expand_ptr_ifn): Update.
3856 * value-prof.c (gimple_divmod_fixed_value): Update.
3857 (gimple_mod_pow2): Update.
3858 (gimple_mod_subtract): Update.
3859 (gimple_ic): Update.
3860 (gimple_stringop_fixed_value): Update.
3862 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3865 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3867 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3869 PR rtl-optimization/82395
3870 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3871 based on non_spilled_static_chain_regno_p.
3873 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3875 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3876 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3877 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3879 2017-10-19 Martin Sebor <msebor@redhat.com>
3881 PR tree-optimization/82596
3882 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3884 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3886 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3887 (get_mem_refs_of_builtin_call): Likewise.
3888 * builtins.c (expand_builtin_apply): Adjust call to
3889 allocate_dynamic_stack_space.
3890 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3891 the third argument to allocate_dynamic_stack_space, otherwise -1.
3892 (expand_builtin): Deal with all alloca variants.
3893 (is_inexpensive_builtin): Likewise.
3894 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3895 * calls.c (special_function_p): Deal with all alloca variants.
3896 (initialize_argument_information): Adjust call to
3897 allocate_dynamic_stack_space.
3898 (expand_call): Likewise.
3899 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3900 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3901 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3902 use it for the stack usage computation.
3903 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3904 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3905 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3906 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3907 (in_loop_p): Remove first argument and useless check.
3908 (pass_walloca::execute): Remove useless test and adjust call to above.
3909 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3910 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3911 (gimplify_call_expr): Deal with all alloca variants.
3912 * hsa-gen.c (gen_hsa_alloca): Likewise.
3913 (gen_hsa_insns_for_call): Likewise.
3914 * ipa-pure-const.c (special_builtin_state): Likewise.
3915 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3916 * tree-object-size.c (alloc_object_size): Likewise.
3917 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3918 (call_may_clobber_ref_p_1): Likewise.
3919 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3920 (ccp_fold_stmt): Likewise.
3921 (optimize_stack_restore): Likewise.
3922 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3923 (mark_all_reaching_defs_necessary_1): Likewise.
3924 (propagate_necessity): Likewise.
3925 (eliminate_unnecessary_stmts): Likewise.
3926 * tree.c (build_common_builtin_nodes): Build
3927 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3928 (build_alloca_call_expr): New function.
3929 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3930 (CASE_BUILT_IN_ALLOCA): Likewise.
3931 (build_alloca_call_expr): Declare.
3932 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3934 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3937 * dwarf2out.c (new_die_raw): New static inline function.
3938 (new_die): Use it to create the DIE.
3939 (add_AT_external_die_ref): Likewise.
3940 (clone_die): Likewise.
3941 (clone_as_declaration): Likewise.
3942 (dwarf2out_vms_debug_main_pointer): Likewise.
3943 (base_type_die): Likewise. Remove early return for corner cases.
3944 Do not call add_pubtype on the DIE here.
3945 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3946 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3947 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3948 native order exists for base types, attach the DIE manually and call
3949 add_pubtype on it. Do not equate a reverse order DIE to the type.
3951 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3953 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3954 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3956 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3958 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3960 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3961 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3962 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3963 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3964 * opts.c (sanitizer_opts): Add builtin.
3965 * ubsan.c (instrument_builtin): New function.
3966 (pass_ubsan::execute): Call it.
3967 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3968 * doc/invoke.texi: Document -fsanitize=builtin.
3970 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3971 builtins, store max (log2 (align), 0) into uchar field instead of
3972 align into uptr field.
3973 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3974 store uchar 0 field instead of uptr 0 field.
3975 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3976 instead of passing one address of struct with 2 locations pass
3977 two addresses of structs with 1 location each.
3978 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3979 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3980 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3981 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3982 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3983 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3984 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3985 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3987 2017-10-19 Martin Liska <mliska@suse.cz>
3990 * file-find.c (remove_prefix): Remove.
3991 * file-find.h (remove_prefix): Likewise.
3992 * gcc-ar.c: Remove smartness of lookup.
3994 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3996 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3997 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3998 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4000 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4003 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4004 (setcc + and to xor + setcc): New peephole2.
4006 2017-10-19 Tom de Vries <tom@codesourcery.com>
4008 * doc/sourcebuild.texi (Test Directives, Variants of
4009 dg-require-support): Add dg-require-stack-size.
4011 2017-10-19 Martin Liska <mliska@suse.cz>
4014 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4015 that have a large alignment.
4016 (gimplify_target_expr): Likewise.
4018 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4020 PR rtl-optimization/82602
4021 * ira.c (rtx_moveable_p): Return false for volatile asm.
4023 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4026 * config/i386/i386-modes.def (CCGZ): New CC mode.
4027 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4028 * config/i386/predicates.md (ix86_comparison_operator):
4030 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4031 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4032 with double-word subtraction.
4033 (put_condition_code): Handle CCGZmode.
4035 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4037 * wide-int.cc (debug (const wide_int &)): New.
4038 (debug (const wide_int *)): New.
4039 (debug (const widest_int &)): New.
4040 (debug (const widest_int *)): New.
4042 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4045 * lra-constraints.c (curr_insn_transform): Use non-input operand
4046 instead of output one for matched reload.
4048 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4050 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4051 (tree-ssa-loop-ivopts.h): New header file.
4052 (struct builtin_info): New fields.
4053 (classify_builtin_1): Compute and record base and offset parts for
4054 memset builtin partition by calling strip_offset.
4055 (offset_cmp, fuse_memset_builtins): New functions.
4056 (finalize_partitions): Fuse adjacent memset partitions by calling
4058 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4059 Expose the interface.
4060 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4062 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4064 PR tree-optimization/82574
4065 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4066 that data reference must be executed exactly once per iteration
4067 against the outermost loop in nest.
4068 (classify_partition): Update call to above function.
4070 2017-10-18 Richard Biener <rguenther@suse.de>
4072 PR tree-optimization/82591
4073 * graphite.c (graphite_transform_loops): Move code gen message
4075 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4076 Here. Handle scop_to_isl_ast failing.
4077 (scop_to_isl_ast): Limit the number of ISL operations.
4079 2017-10-18 Richard Biener <rguenther@suse.de>
4081 * graphite-isl-ast-to-gimple.c
4082 (translate_isl_ast_to_gimple::set_rename): Simplify.
4083 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4084 (graphite_copy_stmts_from_block): ... here.
4085 (copy_bb_and_scalar_dependences): Simplify.
4086 (add_parameters_to_ivs_params): Canonicalize.
4087 (generate_entry_out_of_ssa_copies): Simplify.
4088 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4089 by passing in ISL dimension.
4090 (parameter_index_in_region_1): Rename to ...
4091 (parameter_index_in_region): ... this.
4092 (extract_affine): Adjust assert, pass down parameter index.
4093 (add_param_constraints): Use range-info when available.
4094 (build_scop_context): Adjust.
4095 * sese.c (new_sese_info): Adjust.
4096 (free_sese_info): Likewise.
4097 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4098 Remove unused typedefs.
4099 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4101 2017-10-18 Martin Liska <mliska@suse.cz>
4103 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4105 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4107 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4108 (s390_sched_init): Do not reset s390_sched_state if we entered the
4109 current basic block via a fallthru edge and all others are unlikely.
4111 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4113 * config/s390/s390.c (NUM_SIDES): New variable.
4114 (LONGRUNNING_THRESHOLD): New variable.
4115 (LATENCY_FACTOR): New variable.
4116 (s390_sched_score): Decrease score for long-running instructions on
4118 (s390_sched_variable_issue): Perform bookkeeping for long-running
4121 2017-10-18 Richard Biener <rguenther@suse.de>
4123 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4124 Simplify with removal of the parameter rename map.
4125 (set_rename): Likewise.
4126 (should_copy_to_new_region): Likewise.
4127 (graphite_copy_stmts_from_block): Likewise.
4128 (copy_bb_and_scalar_dependences): Remove initialization of
4129 unused copied_bb_map.
4131 (copy_internal_parameters): Likewise.
4132 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4133 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4134 Use INTEGRAL_TYPE_P.
4135 (parameter_index_in_region_1): Rename to ...
4136 (assign_parameter_index_in_region): ... this. Assert we have
4137 a parameter we handle.
4138 (scan_tree_for_params): Adjust.
4139 * sese.h (parameter_rename_map_t): Remove.
4140 (struct sese_info_t): Remove unused parameter_rename_map and
4141 copied_bb_map members.
4142 * sese.c (new_sese_info): Adjust.
4143 (free_sese_info): Likewise.
4145 2017-10-18 Martin Liska <mliska@suse.cz>
4148 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4149 on an abnormal edge.
4151 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4153 * doc/invoke.texi (ffunction-sections and fdata-sections):
4156 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4158 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4159 the use statement can throw internally.
4161 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4163 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4164 any RTX present on the RHS of a SET.
4165 * compare-elim.c (try_eliminate_compare): Restore comment.
4167 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4169 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4170 may be also called on tcc_type nodes.
4171 * langhooks.c (lhd_tree_size): Likewise.
4173 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4175 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4176 format_warning_at_substring.
4177 (maybe_warn): Convert source_range * param to a location_t. Pass
4178 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4179 (format_directive): Remove code to extract source_ranges and
4180 source_range * in favor of just a location_t.
4181 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4183 * substring-locations.c (format_warning_va): Convert
4184 source_range * param to a location_t.
4185 (format_warning_at_substring): Likewise.
4186 * substring-locations.h (format_warning_va): Likewise.
4187 (format_warning_at_substring): Likewise.
4189 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4191 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4193 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4194 and scatter/gather ops.
4196 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4197 vec_gather_load and vec_scatter_store.
4198 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4199 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4200 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4201 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4202 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4203 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4205 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4207 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4208 Detect FTST instruction and handle its register pops. Only pop
4209 second operand if can_pop_second_op is true.
4210 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4211 set can_pop_second_op to false in the compare_for_stack_reg call.
4213 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4214 output_fp_compare for stack register operands.
4215 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4216 instructions here. Do not emit stack register pops here. Assert
4217 that FCOMPP pops next to top stack register. Rewrite function.
4219 2017-10-17 Nathan Sidwell <nathan@acm.org>
4222 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4223 use DECL_ASSEMBLER_NAME_RAW.
4226 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4229 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4230 Wilco Dijkstra <wilco.dijkstra@arm.com>
4232 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4233 converstion to Pmode from the buf_addr.
4235 2017-10-17 Richard Biener <rguenther@suse.de>
4237 * graphite-dependences.c (scop_get_reads_and_writes): Change
4238 output parameters to references.
4240 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4242 PR 71026/tree-optimization
4243 * fold-const.c (distribute_real_division): Removed.
4244 (fold_binary_loc): Remove calls to distribute_real_divison.
4246 2017-10-17 Richard Biener <rguenther@suse.de>
4248 * graphite-scop-detection.c
4249 (scop_detection::stmt_has_simple_data_refs_p): Always use
4250 the full nest as region.
4251 (try_generate_gimple_bb): Likewise.
4252 * sese.c (scalar_evolution_in_region): Simplify now that
4253 SCEV can handle instantiation in regions.
4254 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4255 in the non-loop part of a function if requested.
4257 2017-10-17 Richard Biener <rguenther@suse.de>
4259 PR tree-optimization/82563
4260 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4262 (graphite_regenerate_ast_isl): Call it.
4263 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4265 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4267 PR tree-optimization/82549
4268 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4269 Formatting fixes. Instead of calling make_bit_field_ref with negative
4272 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4274 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4275 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4276 _mm_maskz_reduce_ss): New.
4277 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4278 __builtin_ia32_reducess_mask): Ditto..
4279 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4280 * config/i386/sse.md (reduces<mode>): Renamed to ...
4281 (reduces<mode><mask_scalar_name>): ... this.
4282 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4284 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4285 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4287 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4289 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4290 * selftest-run-tests.c (selftest::run_tests): Call
4291 selftest::unique_ptr_tests_cc_tests.
4292 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4293 * unique-ptr-tests.cc: New file.
4295 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4298 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4300 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4301 (make_hard_regno_born, make_hard_regno_dead): Update
4302 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4304 2017-10-16 Jeff Law <law@redhat.com>
4306 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4308 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4310 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4312 2017-10-16 Olivier Hainque <hainque@adacore.com>
4314 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4315 with_cpu if we were configured for an e500v2 target cpu name.
4317 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4319 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4320 * doc/invoke.texi: Document +nodsp as a valid extension for
4323 2017-10-16 Martin Liska <mliska@suse.cz>
4325 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4326 (test_set_range): Likewise.
4327 (test_range_functions): Rename to ...
4328 (test_bit_in_range): ... this.
4329 (sbitmap_c_tests): Add new test.
4331 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4333 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4335 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4336 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4338 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4340 * config/aarch64/aarch64-builtins.c
4341 (aarch64_types_quadopu_lane_qualifiers): New.
4342 (TYPES_QUADOPU_LANE): New.
4343 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4344 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4345 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4346 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4347 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4348 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4349 (Vdottype, DOTPROD): New.
4350 (sur): Add SDOT and UDOT.
4352 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4354 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4355 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4356 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4358 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4359 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4360 Enable TARGET_DOTPROD.
4361 (cortex-a75.cortex-a55): Likewise.
4362 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4364 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4366 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4367 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4369 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4371 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4372 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4373 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4374 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4375 * config/arm/types.md (neon_dot, neon_dot_q): New.
4376 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4378 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4380 * config/arm/arm.h (TARGET_DOTPROD): New.
4381 * config/arm/arm.c (arm_arch_dotprod): New.
4382 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4383 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4384 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4385 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4386 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4387 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4388 * doc/invoke.texi (armv8.2-a): Document dotprod
4390 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4392 * i386.c (ix86_vec_cost): New function.
4393 (ix86_rtx_costs): Handle vector operations better.
4394 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4395 * x86-tune-costs.h: Add new costs to all tables.
4397 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4399 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4401 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4402 divsd, sqrtss and sqrtsd
4403 * x86-tune-costs.h: Add new entries to all costs.
4404 (znver1_cost): Fix to match real instruction latencies.
4406 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4407 Michael Collison <michael.collison@arm.com>
4409 * compare-elim.c: Include emit-rtl.h.
4410 (can_merge_compare_into_arith): New function.
4411 (try_validate_parallel): Likewise.
4412 (try_merge_compare): Likewise.
4413 (try_eliminate_compare): Call the above when no previous clobber
4415 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4418 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4422 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4423 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4427 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4428 to be any operand_equal_p operands. For & (B - 1) require
4429 B to be power of 2. Recognize
4430 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4432 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4435 * optabs.c (expand_memory_blockage): Fix call of
4436 targetm.have_memory_blockage.
4438 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4441 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4442 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4443 objects to extra_objs instead of overwriting it.
4445 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4447 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4448 Use any_fp_register_operand as operand[3] predicate. Simplify
4449 equality test for operands[2] and operands[4] memory location.
4450 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4451 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4452 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4453 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4454 any_fp_register_operand as operand[1] predicate. Simplify
4455 equality test for operands[0] and operands[3] memory location.
4456 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4457 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4458 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4460 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4462 * target-insns.def: Add memory_blockage.
4463 * optabs.c (expand_memory_blockage): New function.
4464 (expand_asm_memory_barrier): Rename ...
4465 (expand_asm_memory_blockage): ... to this.
4466 (expand_mem_thread_fence): Call expand_memory_blockage
4467 instead of expand_asm_memory_barrier.
4468 (expand_mem_singnal_fence): Ditto.
4469 (expand_atomic_load): Ditto.
4470 (expand_atomic_store): Ditto.
4471 * doc/md.texi (Standard Pattern Names For Generation):
4472 Document memory_blockage instruction pattern.
4474 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4476 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4477 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4478 * config/rl78/rl78.md: New define_expand "adddi3".
4480 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4482 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4485 2017-10-13 Jeff Law <law@redhat.com>
4487 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4489 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4492 * internal-fn.c (expand_mul_overflow): If both operands have
4493 the same highpart of -1 or 0 and the topmost bit of lowpart
4494 is different, overflow is if res <= 0 rather than res < 0.
4496 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4498 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4499 TARGET_P9_VECTOR code for unaligned_load case.
4501 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4503 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4505 2017-10-13 Nathan Sidwell <nathan@acm.org>
4507 * tree-core.h (tree_contains_struct): Make bool.
4508 * tree.c (tree_contains_struct): Likewise.
4509 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4510 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4511 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4512 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4513 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4515 2017-10-13 Richard Biener <rguenther@suse.de>
4517 * graphite-isl-ast-to-gimple.c
4518 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4519 parameters and dominance check.
4520 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4521 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4522 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4523 Do not update SSA form here or do intermediate IL verification.
4524 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4525 (graphite_initialize): Remove check on the number of loops in
4526 the function and inline into graphite_transform_loops.
4527 (graphite_finalize): Inline into graphite_transform_loops.
4528 (graphite_transform_loops): Perform SSA update and IL verification
4530 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4532 2017-10-13 Richard Biener <rguenther@suse.de>
4534 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4535 graphite_expression_type_precision): Avoid global constructor
4537 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4538 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4539 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4540 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4541 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4542 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4544 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4547 * config/i386/i386.h (ix86_red_zone_size): New.
4548 * config/i386/i386.md (push peephole2s): Replace
4549 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4551 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4552 Alan Hayward <alan.hayward@arm.com>
4553 David Sherwood <david.sherwood@arm.com>
4555 * combine.c (can_change_dest_mode): Reject changes in
4556 REGMODE_NATURAL_SIZE.
4558 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4559 Alan Hayward <alan.hayward@arm.com>
4560 David Sherwood <david.sherwood@arm.com>
4562 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4563 (expand_debug_source_expr): Likewise.
4564 * combine.c (combine_simplify_rtx): Likewise.
4565 * cse.c (fold_rtx): Likewise.
4566 * fwprop.c (canonicalize_address): Likewise.
4567 * targhooks.c (default_shift_truncation_mask): Likewise.
4569 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4570 Alan Hayward <alan.hayward@arm.com>
4571 David Sherwood <david.sherwood@arm.com>
4573 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4574 (widened_mode): Likewise.
4575 (expand_unop): Likewise.
4576 * ree.c (transform_ifelse): Likewise.
4577 (merge_def_and_ext): Likewise.
4578 (combine_reaching_defs): Likewise.
4579 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4581 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4582 Alan Hayward <alan.hayward@arm.com>
4583 David Sherwood <david.sherwood@arm.com>
4585 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4586 * combine.c (gen_lowpart_for_combine): Likewise.
4587 * dwarf2out.c (rtl_for_decl_location): Likewise.
4588 * final.c (alter_subreg): Likewise.
4589 * rtlhooks.c (gen_lowpart_general): Likewise.
4590 (gen_lowpart_if_possible): Likewise.
4592 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4593 Alan Hayward <alan.hayward@arm.com>
4594 David Sherwood <david.sherwood@arm.com>
4596 * calls.c (expand_call): Use subreg_lowpart_offset.
4597 * cse.c (cse_insn): Likewise.
4598 * regcprop.c (copy_value): Likewise.
4599 (copyprop_hardreg_forward_1): Likewise.
4601 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4604 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4605 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4606 =Q constraints to +Q and into insn condition add check
4607 that operands[0] and operands[1] are equal.
4608 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4609 =Q constraints to +Q and into insn condition add check
4610 that operands[0] is equal to either operands[1] or operands[2].
4613 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4614 instead of handling MINUS_EXPR twice (once for each argument),
4615 canonicalize operand order and handle just once, use rtype where
4616 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4619 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4620 any values of __C while still being pattern recognizable as a simple
4623 2017-10-13 Richard Biener <rguenther@suse.de>
4625 PR tree-optimization/82451
4627 2017-10-02 Richard Biener <rguenther@suse.de>
4629 PR tree-optimization/82355
4630 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4631 a mapping for the enclosing loop but avoid generating one for
4633 (copy_bb_and_scalar_dependences): Remove premature codegen
4634 error on PHIs in blocks duplicated into multiple places.
4635 * graphite-scop-detection.c
4636 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4637 in the region use it as loop and nest to analyze the DR in.
4638 (try_generate_gimple_bb): Likewise.
4639 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4640 (add_loop_constraints): For blocks in a loop not in the region
4641 create a dimension with a single iteration.
4642 * sese.h (gbb_loop_at_index): Remove assert.
4644 * cfgloop.c (loop_preheader_edge): For the loop tree root
4645 return the single successor of the entry block.
4646 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4647 Reset the SCEV hashtable and niters.
4648 * graphite-scop-detection.c
4649 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4650 assert that we only have POLYNOMIAL_CHREC that vary in loops
4651 contained in the region.
4652 (scop_detection::graphite_can_represent_expr): Adjust.
4653 (scop_detection::stmt_has_simple_data_refs_p): For loops
4654 not in the region set loop to NULL. The nest is now the
4655 entry edge to the region.
4656 (try_generate_gimple_bb): Likewise.
4657 * sese.c (scalar_evolution_in_region): Adjust for
4658 instantiate_scev change.
4659 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4660 Make nest parameter the edge into the region.
4661 (create_data_ref): Likewise.
4662 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4663 entry edge into a region and adjust instantiate_scev calls.
4664 (create_data_ref): Likewise.
4665 (graphite_find_data_references_in_stmt): Likewise.
4666 (find_data_references_in_stmt): Pass the loop preheader edge
4667 from the nest argument.
4668 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4669 parameter the edge into the region.
4670 (instantiate_parameters): Use the loop preheader edge as entry.
4671 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4673 (get_instantiated_value_entry): Make instantiate_below parameter
4674 the edge into the region.
4675 (instantiate_scev_name): Likewise. Adjust dominance checks,
4676 when we cannot use loop-based instantiation instantiate by
4677 walking use-def chains.
4678 (instantiate_scev_poly): Adjust.
4679 (instantiate_scev_binary): Likewise.
4680 (instantiate_scev_convert): Likewise.
4681 (instantiate_scev_not): Likewise.
4682 (instantiate_array_ref): Remove.
4683 (instantiate_scev_3): Likewise.
4684 (instantiate_scev_2): Likewise.
4685 (instantiate_scev_1): Likewise.
4686 (instantiate_scev_r): Do not blindly handle N-operand trees.
4687 Do not instantiate array-refs. Handle all constants and invariants.
4688 (instantiate_scev): Make instantiate_below parameter
4689 the edge into the region.
4690 (resolve_mixers): Use the loop preheader edge for the region
4691 parameter to instantiate_scev_r.
4692 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4694 2017-10-13 Richard Biener <rguenther@suse.de>
4696 PR tree-optimization/82525
4697 * graphite-isl-ast-to-gimple.c
4698 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4700 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4701 Fail code generation when we cannot represent the isl integer.
4702 (binary_op_to_tree): Elide modulo operations that are no-ops
4703 in the type we code generate. Remove now superfluous code
4706 2017-10-13 Richard Biener <rguenther@suse.de>
4708 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4709 (scop_detection::harmful_loop_in_region): Remove premature
4710 IV type restriction.
4711 (scop_detection::graphite_can_represent_scev): We can handle
4712 pointer IVs just fine.
4714 2017-10-13 Alan Modra <amodra@gmail.com>
4716 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4717 "Clobbers and Scratch Registers". Add paragraph on
4718 alternative to clobbers for scratch registers and OpenBLAS
4721 2017-10-13 Alan Modra <amodra@gmail.com>
4723 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4724 example of a memory input for a string of known length. Move
4725 commentary out of table. Add a number of new examples
4726 covering array memory inputs.
4728 2017-10-12 Martin Liska <mliska@suse.cz>
4730 PR tree-optimization/82493
4731 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4732 (test_range_functions): New function.
4733 (sbitmap_c_tests): Likewise.
4734 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4735 * selftest.h (sbitmap_c_tests): New function.
4737 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4739 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4741 * config/rs6000/amo.h: Fix spacing issue.
4743 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4746 * config/i386/i386.md (*ashl<mode>3_mask_1,
4747 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4748 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4751 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4753 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4754 (profile_probability): Set max_probability
4755 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4757 * profile-count.c (profile_probability::differs_from_p): Do not
4758 rely on max_probaiblity == 10000
4760 2017-10-12 Jeff Law <law@redhat.com>
4762 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4765 2017-10-12 Martin Sebor <msebor@redhat.com>
4769 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4770 (handle_alias_pairs): Call it.
4771 * common.opt (-Wattribute-alias): New option.
4772 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4773 * doc/invoke.texi (-Wattribute-alias): Document.
4775 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4778 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4780 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4782 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4783 (make_hard_regno_born, make_hard_regno_dead): Update
4784 bb_killed_pseudos and bb_gen_pseudos.
4786 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4788 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4790 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4792 * config/alpha/alpha.c (alpha_split_conditional_move):
4793 Use std::swap instead of manually swapping.
4794 (alpha_stdarg_optimize_hook): Ditto.
4795 (alpha_canonicalize_comparison): Ditto.
4797 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4799 * tree-loop-distribution.c (struct builtin_info): New struct.
4800 (struct partition): Refactor fields into struct builtin_info.
4801 (partition_free): Free struct builtin_info.
4802 (build_size_arg_loc, build_addr_arg_loc): Delete.
4803 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4804 information from struct builtin_info.
4805 (find_single_drs): New function refactored from classify_partition.
4806 Also moved builtin validity checks to this function.
4807 (compute_access_range, alloc_builtin): New functions.
4808 (classify_builtin_st, classify_builtin_ldst): New functions.
4809 (classify_partition): Refactor code into functions find_single_drs,
4810 classify_builtin_st and classify_builtin_ldst.
4811 (distribute_loop): Don't do runtime alias check when distributing
4813 (find_seed_stmts_for_distribution): New function.
4814 (pass_loop_distribution::execute): Refactor code finding seed
4815 stmts into above function. Support distribution for the innermost
4816 two-level loop nest. Adjust dump information.
4818 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4820 * tree-loop-distribution.c: Adjust the general comment.
4821 (NUM_PARTITION_THRESHOLD): New macro.
4822 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4823 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4824 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4825 in call to build_partition_graph.
4826 (finalize_partitions): New parameter. Make loop distribution more
4827 conservative by fusing more partitions.
4828 (distribute_loop): Don't do runtime alias check in case of loop nest
4830 (find_seed_stmts_for_distribution): New function.
4831 (prepare_perfect_loop_nest): New function.
4832 (pass_loop_distribution::execute): Refactor code finding seed stmts
4833 and loop nest into above functions. Support loop nest distribution.
4834 Adjust dump information accordingly.
4836 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4838 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4839 and set PTYPE_SEQUENTIAL for merged partition.
4841 2017-10-12 Richard Biener <rguenther@suse.de>
4843 PR tree-optimization/69728
4845 2017-09-19 Richard Biener <rguenther@suse.de>
4847 PR tree-optimization/69728
4848 * graphite-sese-to-poly.c (schedule_error): New global.
4849 (add_loop_schedule): Handle empty domain by failing the
4851 (build_original_schedule): Handle schedule_error.
4853 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4854 domain by returning an unchanged schedule.
4856 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4858 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4859 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4861 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4863 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4866 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4869 * expr.c (store_field): Don't optimize away bitsize == 0 store
4870 from CALL_EXPR with addressable return type.
4872 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4874 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4875 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4876 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4877 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4878 TARGET_ISEL instead of TARGET_ISEL<sel>.
4880 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4882 * config/rs6000/rs6000.c
4883 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4885 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4887 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4889 (reg_or_cint_operand): Fix comment.
4890 (reg_or_zero_operand): New predicate.
4891 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4892 * config/rs6000/rs6000.c (output_isel): Delete.
4893 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4894 instead of reg_or_cint_operand. Output instruction directly (not via
4896 (isel_unsigned_<mode>): Ditto.
4897 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4898 gpc_reg_operand. Add an instruction alternative for this. Output
4899 instruction directly.
4900 (*isel_reversed_unsigned_<mode>): Ditto.
4902 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4904 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4905 (TARGET_CANONICALIZE_COMPARISON): Define.
4907 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4910 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4911 Check whether the dest is REG before adding REG_EQUIV note.
4913 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4916 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4918 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4919 (make_hard_regno_born, make_hard_regno_dead): Update
4920 bb_killed_pseudos and bb_gen_pseudos.
4922 2017-10-11 Nathan Sidwell <nathan@acm.org>
4924 * incpath.h (enum incpath_kind): Name enum, prefix values.
4925 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4926 * incpath.c (heads, tails): Use INC_MAX.
4927 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4928 (merge_include_chains, split_quote_chain,
4929 register_include_chains): Update incpath_kind names.
4930 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4931 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4933 (add_framework_path, darwin_register_objc_includes): Likewise.
4934 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4936 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4938 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4939 Do not use float_operator operator predicate.
4940 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4941 * config/i386/predicates.md (float_operator): Remove predicate.
4943 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4945 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4946 (*jccxf_i387): Ditto.
4947 (*jcc<mode>_i387): Ditto.
4948 (*jccu<mode>_i387): Ditto.
4949 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4950 (*jcc_*_i387 splitters): Remove.
4951 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4952 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4953 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4956 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4958 * profile-count.h (slow_safe_scale_64bit): New function.
4959 (safe_scale_64bit): New inline.
4960 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4961 * profile-count.c: Include wide-int.h
4962 (slow_safe_scale_64bit): New.
4964 2017-10-11 Nathan Sidwell <nathan@acm.org>
4966 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4967 HAS_DECL_ASSEMBLER_NAME_P.
4968 * gimple-expr.c (gimple_decl_printable_name: Check
4969 HAS_DECL_ASSEMBLER_NAME_P too.
4970 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4971 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4972 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4973 * passes.c (rest_of_decl_compilation): Check
4974 HAS_DECL_ASSEMBLER_NAME_P too.
4975 * recog.c (verify_changes): Likewise.
4976 * tree-pretty-print.c (dump_decl_name): Likewise.
4977 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4979 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4980 (SET_DECL_ASSEMBLER_NAME): Use it.
4981 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4982 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4983 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4985 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4987 * config.gcc (i386, x86_64): Add extra objects.
4988 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4989 (ix86_min_insn_size): Declare.
4990 (ix86_issue_rate): Declare.
4991 (ix86_adjust_cost): Declare.
4992 (ia32_multipass_dfa_lookahead): Declare.
4993 (ix86_macro_fusion_p): Declare.
4994 (ix86_macro_fusion_pair_p): Declare.
4995 (ix86_bd_has_dispatch): Declare.
4996 (ix86_bd_do_dispatch): Declare.
4997 (ix86_core2i7_init_hooks): Declare.
4998 (ix86_atom_sched_reorder): Declare.
4999 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5000 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5001 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5002 (rip_relative_addr_p): Rename to ...
5003 (ix86_rip_relative_addr_p): ... this one; export.
5004 (memory_address_length): Update.
5005 (ix86_issue_rate): Move to x86-tune-sched.c.
5006 (ix86_flags_dependent): Move to x86-tune-sched.c.
5007 (ix86_agi_dependent): Move to x86-tune-sched.c.
5008 (exact_dependency_1): Move to x86-tune-sched.c.
5009 (exact_store_load_dependency): Move to x86-tune-sched.c.
5010 (ix86_adjust_cost): Move to x86-tune-sched.c.
5011 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5012 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5013 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5014 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5015 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5016 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5017 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5018 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5019 (min_insn_size): Rename to ...
5020 (ix86_min_insn_size): ... this one; export.
5021 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5022 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5023 (core2i7_first_cycle_multipass_backtrack): Move to
5024 x86-tune-sched-core.c.
5025 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5026 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5027 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5028 (ix86_avoid_jump_mispredicts): Update.
5029 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5030 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5031 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5032 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5033 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5034 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5035 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5036 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5037 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5038 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5039 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5040 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5041 (BIG): Move to ix86-tune-sched-bd.c.
5042 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5043 (enum insn_path): Move to ix86-tune-sched-bd.c.
5044 (get_mem_group): Move to ix86-tune-sched-bd.c.
5045 (is_cmp): Move to ix86-tune-sched-bd.c.
5046 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5047 (is_branch): Move to ix86-tune-sched-bd.c.
5048 (is_prefetch): Move to ix86-tune-sched-bd.c.
5049 (init_window): Move to ix86-tune-sched-bd.c.
5050 (allocate_window): Move to ix86-tune-sched-bd.c.
5051 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5052 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5053 (process_end_window): Move to ix86-tune-sched-bd.c.
5054 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5055 (find_constant): Move to ix86-tune-sched-bd.c.
5056 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5057 (has_immediate): Move to ix86-tune-sched-bd.c.
5058 (get_insn_path): Move to ix86-tune-sched-bd.c.
5059 (get_insn_group): Move to ix86-tune-sched-bd.c.
5060 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5061 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5062 (add_insn_window): Move to ix86-tune-sched-bd.c.
5063 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5064 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5065 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5066 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5067 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5068 (do_dispatch): Move to ix86-tune-sched-bd.c.
5069 (has_dispatch): Move to ix86-tune-sched-bd.c.
5070 * config/i386/t-i386: Add new object files.
5071 * config/i386/x86-tune-costs.h: New file.
5072 * config/i386/x86-tune-sched-atom.c: New file.
5073 * config/i386/x86-tune-sched-bd.c: New file.
5074 * config/i386/x86-tune-sched-core.c: New file.
5075 * config/i386/x86-tune-sched.c: New file.
5077 2017-10-11 Liu Hao <lh_mouse@126.com>
5079 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5080 the generic version below instead.
5081 (should_colorize): Recognize Windows consoles as terminals
5083 * pretty-print.c [__MINGW32__] (write_all): New function.
5084 [__MINGW32__] (find_esc_head): Likewise.
5085 [__MINGW32__] (find_esc_terminator): Likewise.
5086 [__MINGW32__] (eat_esc_sequence): Likewise.
5087 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5089 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5092 2017-10-11 Richard Biener <rguenther@suse.de>
5094 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5095 Properly call analyze_scalar_evolution with the loop of the stmt.
5097 2017-10-11 Richard Biener <rguenther@suse.de>
5099 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5100 * tree-core.h (tree_base): Add chrec_var union member.
5101 * tree.h (CHREC_VAR): Remove.
5102 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5103 * tree-chrec.h (build_polynomial_chrec): Adjust.
5104 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5105 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5107 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5109 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5110 * match.pd: ... here.
5111 ((T) X == (T) Y): Relax condition.
5113 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5115 PR tree-optimization/82472
5116 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5118 (break_alias_scc_partitions): Update postorder number.
5120 2017-10-11 Martin Liska <mliska@suse.cz>
5123 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5125 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5127 2017-10-11 Martin Liska <mliska@suse.cz>
5129 * print-rtl.c (print_insn): Move declaration of idbuf
5130 to same scope as name.
5132 2017-10-11 Martin Liska <mliska@suse.cz>
5137 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5140 2017-10-11 Martin Liska <mliska@suse.cz>
5143 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5146 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5148 * config.gcc (default_gnu_indirect_function): Default to yes for
5149 arm*-*-linux* with glibc.
5151 2017-10-11 Richard Biener <rguenther@suse.de>
5153 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5154 default-defs and types we do not want to analyze.
5155 (interpret_loop_phi): Replace unreachable code with an assert.
5156 (compute_scalar_evolution_in_loop): Remove and inline ...
5157 (analyze_scalar_evolution_1): ... here, replacing condition with
5158 what makes the intent clearer. Remove handling of cases
5159 get_scalar_evolution now handles.
5161 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5163 PR rtl-optimization/81434
5164 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5165 comment for main loop. In sched_group_found if, also add checks for
5166 pass and min_cost_group.
5168 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5170 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5171 (rs6000_insn_cost): New function.
5172 * config/rs6000/rs6000.md (cost): New attribute.
5174 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5175 H.J. Lu <hongjiu.lu@intel.com>
5179 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5180 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5181 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5182 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5183 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5184 OPTION_MASK_ISA_MMX and some other bit set require both
5185 mmx and the ISAs without the mmx bit.
5186 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5187 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5188 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5189 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5190 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5191 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5192 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5193 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5194 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5195 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5196 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5197 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5198 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5199 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5200 Add OPTION_MASK_ISA_MMX.
5202 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5204 * config.gcc (armv7*-*-freebsd*): New target.
5205 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5207 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5209 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5210 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5213 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5215 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5216 a template parameter.
5217 (WIDE_INT_REF_FOR): Update accordingly.
5218 * tree.h (wi::int_traits <const_tree>): Delete.
5219 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5220 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5221 (wi::tree_to_wide_ref): New typedef.
5222 (wi::to_wide): New function.
5223 * calls.c (get_size_range): Use wi::to_wide when operating on
5225 * cgraph.c (cgraph_node::create_thunk): Likewise.
5226 * config/i386/i386.c (ix86_data_alignment): Likewise.
5227 (ix86_local_alignment): Likewise.
5228 * dbxout.c (stabstr_O): Likewise.
5229 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5230 * expr.c (const_vector_from_tree): Likewise.
5231 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5232 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5233 (fold_negate_expr_1, int_const_binop_1, const_binop)
5234 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5235 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5236 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5237 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5238 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5239 (fold_not_const, round_up_loc): Likewise.
5240 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5241 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5242 (alloca_call_type): Likewise.
5243 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5244 * godump.c (go_output_typedef): Likewise.
5245 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5246 * internal-fn.c (get_min_precision): Likewise.
5247 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5248 * ipa-polymorphic-call.c
5249 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5250 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5251 (ipa_modify_call_arguments): Likewise.
5252 * match.pd: Likewise.
5253 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5254 * print-tree.c (print_node_brief, print_node): Likewise.
5255 * stmt.c (expand_case): Likewise.
5256 * stor-layout.c (layout_type): Likewise.
5257 * tree-affine.c (tree_to_aff_combination): Likewise.
5258 * tree-cfg.c (group_case_labels_stmt): Likewise.
5259 * tree-data-ref.c (dr_analyze_indices): Likewise.
5260 (prune_runtime_alias_test_list): Likewise.
5261 * tree-dump.c (dequeue_and_dump): Likewise.
5262 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5263 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5264 * tree-pretty-print.c (dump_generic_node): Likewise.
5265 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5266 (simple_iv_with_niters): Likewise.
5267 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5268 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5269 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5270 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5271 (refine_value_range_using_guard, number_of_iterations_ne_max)
5272 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5273 (get_cst_init_from_scev, record_nonwrapping_iv)
5274 (scev_var_range_cant_overflow): Likewise.
5275 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5276 * tree-ssa-pre.c (compute_avail): Likewise.
5277 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5278 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5279 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5280 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5281 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5282 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5283 (dump_case_nodes, try_switch_expansion): Likewise.
5284 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5285 (vect_do_peeling): Likewise.
5286 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5287 * tree-vect-stmts.c (vectorizable_load): Likewise.
5288 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5289 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5290 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5291 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5292 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5293 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5294 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5295 (evrp_dom_walker::before_dom_children): Likewise.
5296 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5297 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5298 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5299 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5300 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5301 (get_range_pos_neg): Likewise.
5302 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5303 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5304 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5305 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5306 * config/avr/avr.c (avr_fold_builtin): Likewise.
5307 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5308 * config/msp430/msp430.c (msp430_attr): Likewise.
5309 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5310 * config/powerpcspe/powerpcspe-c.c
5311 (altivec_resolve_overloaded_builtin): Likewise.
5312 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5313 (rs6000_expand_ternop_builtin): Likewise.
5314 * config/rs6000/rs6000-c.c
5315 (altivec_resolve_overloaded_builtin): Likewise.
5316 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5317 (rs6000_expand_ternop_builtin): Likewise.
5318 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5320 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5322 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5323 when copying loop nest with only one inner loop.
5325 2017-10-10 Richard Biener <rguenther@suse.de>
5327 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5328 blocks if SCEV is active.
5329 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5331 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5332 (scev_initialize): Assert we are not yet initialized.
5334 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5336 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5337 inner loop's exit stmt by making it always exit the loop, otherwise
5338 we would generate an infinite empty loop.
5340 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5342 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5343 renaming variables in new preheader if it's deleted.
5345 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5347 * tree-loop-distribution.c (struct partition): Remove unused field
5348 loops of the structure.
5349 (partition_alloc, partition_free): Ditto.
5350 (build_rdg_partition_for_vertex): Ditto.
5352 2017-10-09 Jeff Law <law@redhat.com>
5354 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5355 return type to match prototype and documentation.
5357 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5359 * config/rs6000/rs6000.c (processor_costs): Move to ...
5360 * config/rs6000/rs6000.h: ... here.
5361 (rs6000_cost): Declare.
5363 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5365 * except.c (setjmp_fn): New global variable.
5366 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5367 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5368 if DONT_USE_BUILTIN_SETJMP is defined.
5370 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5372 * target.def (insn_cost): New hook.
5373 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5374 * doc/tm.texi: Regenerate.
5375 * rtlanal.c (insn_cost): Use the new hook.
5377 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5379 * combine.c (combine_validate_cost): Compute the new insn_cost,
5380 not just pattern_cost.
5381 (try_combine): Adjust comment.
5383 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5385 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5387 * combine.c (uid_insn_cost): Adjust comment.
5388 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5390 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5391 * dse.c (find_shift_sequence): Ditto.
5392 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5393 (bb_valid_for_noce_process_p): Use pattern_cost.
5394 * rtl.h (insn_rtx_cost): Delete.
5395 (pattern_cost): New prototype.
5396 (insn_cost): New prototype.
5397 * rtlanal.c (insn_rtx_cost): Rename to...
5398 (pattern_cost): ... this.
5401 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5403 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5404 (*jcc<mode>_0_r_i387): Ditto.
5405 (*jccxf_r_i387): Ditto.
5406 (*jcc<mode>_r_i387): Ditto.
5407 (*jccu<mode>_r_i387): Ditto.
5408 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5409 (*jcc): Rename from *jcc_1.
5411 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5413 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5414 deferred rescans after the lvx/stvx recombination pre-pass.
5416 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5418 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5419 memory operation instruction support.
5420 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5421 (rs6000-ibm-aix[789]*): Likewise.
5422 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5423 Document new functions.
5425 2017-10-09 Richard Biener <rguenther@suse.de>
5427 PR tree-optimization/82397
5428 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5429 equality only for semantically equal trees.
5431 2017-10-09 Richard Biener <rguenther@suse.de>
5433 PR tree-optimization/82449
5434 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5435 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5436 allow constant addresses.
5437 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5440 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5442 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5445 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5448 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5451 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5454 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5456 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5459 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5460 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5462 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5464 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5465 (WI_BINARY_PREDICATE_RESULT): Likewise.
5466 (wi::binary_traits::operator_result): New type.
5467 (wi::binary_traits::predicate_result): Likewise.
5468 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5469 (generic_wide_int::operator==, generic_wide_int::operator!=)
5470 (generic_wide_int::operator&, generic_wide_int::and_not)
5471 (generic_wide_int::operator|, generic_wide_int::or_not)
5472 (generic_wide_int::operator^, generic_wide_int::operator+
5473 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5475 (operator~, unary operator-, operator==, operator!=, operator&)
5476 (operator|, operator^, operator+, binary operator-, operator*): New
5478 * expr.c (get_inner_reference): Use wi::bit_and_not.
5479 * fold-const.c (fold_binary_loc): Likewise.
5480 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5481 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5482 (bit_value_binop): Likewise.
5483 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5484 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5485 (extract_range_from_binary_expr_1): Likewise.
5486 (masked_increment): Likewise.
5487 (simplify_bit_ops_using_ranges): Likewise.
5489 2017-10-09 Martin Jambor <mjambor@suse.cz>
5492 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5493 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5494 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5496 (hsa_fixup_mov_insn_type): New function.
5497 (hsa_op_with_type::get_in_type): Use it.
5498 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5499 immediates in an assert.
5500 (hsa_op_with_type::extend_int_to_32bit): New method.
5501 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5502 types. Convert to dest type if necessary.
5503 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5504 (reg_for_gimple_ssa): Pass false as min32int to
5505 hsa_type_for_scalar_tree_type.
5506 (gen_hsa_addr): Fixup type when creating addresable temporary.
5507 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5508 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5509 necessary. Call hsa_fixup_mov_insn_type.
5510 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5511 extend operands and convert to dest type if necessary.
5512 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5513 to dest type if necessary.
5514 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5515 if conversion nt necessary and size matches.
5516 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5517 to dest type if necessary.
5518 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5519 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5521 (gen_hsa_clrsb): Likewise.
5522 (gen_hsa_ffs): Likewise.
5523 (gen_hsa_divmod): Extend operands and convert to dest type if
5525 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5527 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5529 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5530 Remove empty default arguments. Use a brace block as output
5532 (conditional return): Ditto.
5534 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5535 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5537 (group_ending_nop): Ditto.
5538 (doloop_end): Ditto.
5539 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5540 (splitters for those): Ditto.
5542 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5544 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5545 a conditional jump (and the compare for it) so that pc_rtx is the
5547 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5548 for the deleted and renamed ctr<mode>_internal[234] patterns.
5549 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5550 Delete second conditional return pattern.
5551 (ctr<mode>_internal2): Delete this second bdnz pattern.
5552 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5553 (ctr<mode>_internal4): Delete this second bdz pattern.
5555 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5557 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5558 (always_initialized_rtx_for_ssa_name_p): New predicate.
5559 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5560 (finish_out_of_ssa): Free new field of SA.
5561 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5562 * tree-ssa-coalesce.c: Include tree-ssa.h.
5563 (get_parm_default_def_partitions): Remove extern keyword.
5564 (get_undefined_value_partitions): New function.
5565 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5566 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5568 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5570 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5572 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5574 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5576 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5577 for targets that preffer 128bit.
5579 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5581 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5583 2017-10-08 Olivier Hainque <hainque@adacore.com>
5585 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5586 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5588 (thumb_set_return_address): Likewise.
5590 2017-10-08 Olivier Hainque <hainque@adacore.com>
5592 * common/config/arm/arm-common.c (arm_except_unwind_info):
5593 Handle DWARF2_UNWIND_INFO.
5595 2017-10-07 Michael Collison <michael.collison@arm.com>
5597 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5600 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5602 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5603 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5604 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5605 defined, force the creation of a new block for a dispatch label.
5607 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5609 * invoke.texi (Wsuggest-attribute=cold): Document.
5610 * common.opt (Wsuggest-attribute=cold): New
5611 * ipa-pure-const.c (warn_function_cold): New function.
5612 * predict.c (compute_function_frequency): Use it.
5613 * predict.h (warn_function_cold): Declare.
5615 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5617 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5620 2017-10-06 Martin Liska <mliska@suse.cz>
5622 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5623 keyword for member functions.
5624 (struct sanopt_tree_couple): New struct.
5625 (struct sanopt_tree_couple_hash): New function.
5626 (struct sanopt_ctx): Add new hash_map.
5627 (has_dominating_ubsan_ptr_check): New function.
5628 (record_ubsan_ptr_check_stmt): Likewise.
5629 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5630 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5631 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5633 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5636 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5637 aarch64_simd_valid_immediate on CONST_VECTORs.
5638 (aarch64_reg_or_bic_imm): Likewise.
5640 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5642 PR rtl-optimization/82396
5643 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5645 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5647 * graphite-dependences.c (scop_get_reads): Move code to...
5648 (scop_get_must_writes): Move code to...
5649 (scop_get_may_writes): Move code to...
5650 (scop_get_reads_and_writes): ... here.
5651 (scop_get_dependences): Call scop_get_reads_and_writes.
5653 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5655 PR tree-optimization/82434
5656 * fold-const.h (can_native_encode_type_p,
5657 can_native_encode_string_p): Remove.
5658 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5659 don't encode anything, just return what would be otherwise returned.
5660 (native_encode_fixed, native_encode_complex, native_encode_vector):
5662 (native_encode_string): Likewise. Inline by hand
5663 can_native_encode_string_p.
5664 (can_native_encode_type_p): Remove.
5665 (can_native_encode_string_p): Remove.
5666 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5667 STRING_CSTs using can_native_encode_string_p, test all
5668 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5669 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5670 argument from native_encode_expr.
5671 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5672 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5675 2017-10-06 Richard Biener <rguenther@suse.de>
5677 PR tree-optimization/82397
5678 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5679 operand_equal_p but rely on data_ref_compare_tree for detecting
5681 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5682 to match up with dr_group_sort_cmp.
5684 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5687 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5689 * config/s390/s390-builtin-types.def: Regenerate.
5691 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5694 * config/s390/s390-builtin-types.def: Regenerate.
5695 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5696 Change flag from B_VXE to B_VX.
5697 (s390_vec_min_dbl): Remove B_VXE flag.
5699 2017-10-06 Richard Biener <rguenther@suse.de>
5701 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5702 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5703 translate_isl_ast_to_gimple::is_valid_rename,
5704 translate_isl_ast_to_gimple::get_rename,
5705 translate_isl_ast_to_gimple::get_def_bb_for_const,
5706 translate_isl_ast_to_gimple::get_new_name,
5707 translate_isl_ast_to_gimple::collect_all_ssa_names,
5708 translate_isl_ast_to_gimple::copy_loop_phi_args,
5709 translate_isl_ast_to_gimple::collect_all_ssa_names,
5710 translate_isl_ast_to_gimple::copy_loop_phi_args,
5711 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5712 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5713 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5714 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5715 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5716 translate_isl_ast_to_gimple::copy_cond_phi_args,
5717 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5718 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5719 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5720 translate_isl_ast_to_gimple::rename_uses,
5721 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5722 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5723 (set_rename_for_each_def): Likewise.
5724 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5725 here. Handle rewriting SCEV analyzable uses here.
5726 (copy_bb_and_scalar_dependences): Generate code for PHI
5728 (graphite_regenerate_ast_isl): Adjust.
5729 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5730 (add_write, add_read): New functions.
5731 (build_cross_bb_scalars_def): Use it and simplify.
5732 (build_cross_bb_scalars_use): Likewise.
5733 (graphite_find_cross_bb_scalar_vars): Inline into...
5734 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5735 simulating out-of-SSA. Compute liveout and add dependencies.
5736 (build_scops): Force an empty entry block.
5737 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5739 (sese_build_liveouts): Declare.
5740 (sese_trivially_empty_bb_p): Likewise.
5741 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5742 compute liveout and debug_liveout.
5743 (sese_bad_liveouts_use): Remove.
5744 (sese_reset_debug_liveouts_bb): Likewise.
5745 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5746 (sese_build_liveouts): Build liveout and debug_liveout and store
5748 (new_sese_info): Adjust.
5749 (free_sese_info): Likewise.
5750 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5751 do not build liveout here.
5752 (move_sese_in_condition): Adjust region entry.
5753 (scev_analyzable_p): Match up with chrec_apply requirements.
5754 (sese_trivially_empty_bb_p): New.
5755 * tree-into-ssa.c (get_reaching_def): Properly support generating
5756 default-defs for incremental rewrite of anonymous names.
5758 2017-10-06 Richard Biener <rguenther@suse.de>
5760 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5761 precision do not perform modulo reduction.
5763 2017-10-06 Richard Biener <rguenther@suse.de>
5765 PR tree-optimization/82436
5766 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5767 conservatively choose the vectorization factor when checking
5768 whether we can perform the required load permutation.
5769 (vect_transform_slp_perm_load): Assert when we may not fail.
5771 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5773 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5774 message for incompatible -msdata=* and -mcall-* options.
5776 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5778 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5779 rate for post-reload scheduling.
5781 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5783 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5785 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5787 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5788 to improve monte carlo in scimark.
5790 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5792 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5793 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5794 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5795 pentium4_cost, nocona_cost): Set reassociation width to 1.
5796 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5797 width to 2 for fp operations and 1 otherwise.
5798 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5800 (atom_cost): Set reassociation width to 2.
5801 (slm_cost, generic_cost): Set fp reassociation width
5802 to 2 and 1 otherwise.
5803 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5804 (core_cost): Set fp reassociation width to 4 and vector to 2.
5805 (ix86_reassociation_width): Rewrite using cost table; special case
5806 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5807 and TARGET_AVX128_OPTIMAL.
5808 * config/i386/i386.h (processor_costs): Add
5809 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5810 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5811 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5812 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5813 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5814 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5816 2017-10-05 Nathan Sidwell <nathan@acm.org>
5818 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5820 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5822 * config/arm/arm.c (arm_test_fpu_data): New.
5823 (arm_run_selftests): Call arm_test_fpu_data.
5825 2017-10-04 Nathan Sidwell <nathan@acm.org>
5827 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5829 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5831 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5833 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5834 check type for aarch64_simd_valid_immediate.
5835 (aarch64_output_simd_mov_immediate): Update prototype.
5836 (aarch64_simd_valid_immediate): Update prototype.
5837 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5838 support for ORR-immediate.
5839 (and<mode>3): modified pattern to add support for BIC-immediate.
5840 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5841 now checks for valid immediate for BIC and ORR based on new enum
5843 (aarch64_output_simd_mov_immediate): Function now used to output
5844 BIC/ORR imm as well based on new enum argument.
5845 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5847 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5848 (aarch64_reg_or_bic_imm): Likewise.
5850 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5852 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5853 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5855 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5859 PR rtl-optimization/82396
5860 * haifa-sched.c (autopref_multipass_init): Simplify
5862 (autopref_rank_data): Simplify sort order.
5863 * sched-int.h (autopref_multipass_data_): Remove
5864 multi_mem_insn_p, min_offset and max_offset.
5866 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5868 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5870 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5872 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5875 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5877 * doc/sourcebuild.texi: Document vect_long_mult.
5879 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5881 PR tree-optimization/82413
5882 * fold-const.c (build_range_check): Use widest_int when comparing
5883 the maximum ETYPE value with HIGH.
5885 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5887 PR rtl-optimization/82396
5888 * haifa-sched.c (autopref_multipass_init): Simplify
5890 (autopref_rank_data): Simplify sort order.
5891 * sched-int.h (autopref_multipass_data_): Remove
5892 multi_mem_insn_p, min_offset and max_offset.
5894 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5896 PR tree-optimization/82381
5897 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5898 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5901 PR tree-optimization/82374
5902 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5903 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5904 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5905 current_function_decl to the new decl.
5907 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5909 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5910 helper macro for IEEE float128 hardware built-in functions.
5911 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5913 (TRUNCF128_ODD): Likewise.
5914 (ADDF128_ODD): Likewise.
5915 (SUBF128_ODD): Likewise.
5916 (MULF128_ODD): Likewise.
5917 (DIVF128_ODD): Likewise.
5918 (FMAF128_ODD): Likewise.
5919 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5920 UNSPEC_TRUNC_ROUND_TO_ODD.
5921 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5922 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5923 floating point round to odd instructions.
5924 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5925 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5926 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5927 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5928 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5929 (trunc<mode>sf2_hw): Change the truncate with round to odd
5930 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5931 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5932 to odd hardware instructions.
5933 (sub<mode>3_odd): Likewise.
5934 (mul<mode>3_odd): Likewise.
5935 (div<mode>3_odd): Likewise.
5936 (sqrt<mode>2_odd): Likewise.
5937 (fma<mode>4_odd): Likewise.
5938 (fms<mode>4_odd): Likewise.
5939 (nfma<mode>4_odd): Likewise.
5940 (nfms<mode>4_odd): Likewise.
5941 (trunc<mode>df2_odd): Change the truncate with round to odd
5942 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5944 * doc/extend.texi (PowerPC built-in functions): Update documentation
5945 for existing IEEE float128-bit built-in functions. Add built-in
5946 functions that generate the IEEE 128-bit floating point round to
5949 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5951 PR rtl-optimization/77729
5952 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5953 to (X&(C1&~C2))|C2 transformations.
5955 2017-10-03 Martin Jambor <mjambor@suse.cz>
5957 PR tree-optimization/82363
5958 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5959 mismatch, mark lacc written regardless of racc.
5961 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5963 PR tree-optimization/82381
5964 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5965 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5966 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5967 fallthrough into reassoc_stmt_dominates_stmt_p.
5970 * combine.c (combine_instructions): Don't combine in unreachable
5973 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5976 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5977 function to not use the have_cpu variable. Do not set cpu_index,
5978 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5980 (rs6000_valid_attribute_p): Remove duplicate initializations of
5981 old_optimize and func_optimize.
5982 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5983 (rs6000_activate_target_options): Make global.
5984 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5987 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5989 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5990 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5991 Set *poffset to 0 and *psize and *pmax_size to -1 if
5992 *poffset + *psize overflows in HOST_WIDE_INT.
5994 PR tree-optimization/82387
5995 PR tree-optimization/82388
5996 PR tree-optimization/82389
5997 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5998 instead of live_bytes non-NULL.
6000 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6003 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6006 2017-10-02 Richard Biener <rguenther@suse.de>
6008 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6009 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6010 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6012 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6014 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6015 requested precision matches the type's.
6016 * calls.c (alloc_max_size): Calculate the new candidate size as
6017 a widest_int and use wi::to_widest when comparing it with the
6018 current candidate size.
6019 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6020 zero rather than integer_zero_node.
6021 * match.pd: Check for a no-op conversion before using wi::add
6022 rather than after. Use tree_to_uhwi when summing small shift
6023 counts into an unsigned int.
6025 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6026 Alan Hayward <alan.hayward@arm.com>
6027 David Sherwood <david.sherwood@arm.com>
6030 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6031 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6032 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6033 POINTER_AND_FP_REGS.
6035 2017-10-02 Richard Biener <rguenther@suse.de>
6037 PR tree-optimization/82355
6038 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6039 a mapping for the enclosing loop but avoid generating one for
6041 (copy_bb_and_scalar_dependences): Remove premature codegen
6042 error on PHIs in blocks duplicated into multiple places.
6043 * graphite-scop-detection.c
6044 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6045 in the region use it as loop and nest to analyze the DR in.
6046 (try_generate_gimple_bb): Likewise.
6047 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6048 (add_loop_constraints): For blocks in a loop not in the region
6049 create a dimension with a single iteration.
6050 * sese.h (gbb_loop_at_index): Remove assert.
6052 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6054 * omp-expand.c (adjust_context_scope): New function.
6055 (expand_parallel_call): Call adjust_context_scope.
6057 2017-10-01 Jeff Law <law@redhat.com>
6059 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6060 dom_opt_dom_walker class with direct access to private members.
6061 Add comments. Call test_for_singularity.
6062 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6063 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6064 m_dummy_cond anymore.
6065 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6067 (pass_dominator:execute): Build the dummy_cond here and pass it
6068 to the dom_opt_dom_walker ctor.
6069 (test_for_singularity): New function.
6071 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6072 Maya Rashish <coypu@sdf.org>
6074 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6075 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6076 (alpha*-*-netbsd*) Use nbsd_tm_file.
6077 (arm*-*-netbsdelf*) Likewise.
6078 (i[34567]86-*-netbsdelf*) Likewise.
6079 (x86_64-*-netbsd*) Likewise.
6080 (mips*-*-netbsd*) Likewise.
6081 (powerpc-*-netbsd*) Likewise.
6082 (sh*-*-netbsd*) Likewise.
6083 (sparc-*-netbsdelf*) Likewise.
6084 (sparc64-*-netbsd*) Likewise.
6085 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6087 (vax-*-netbsdelf*) Likewise.
6088 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6089 (UINT_FAST8_TYPE) Likewise.
6090 (INT_FAST16_TYPE) Check CHAR_FAST16.
6091 (UINT_FAST16_TYPE) Likewise.
6093 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6096 * config/i386/i386.md
6097 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6098 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6099 *divmodsi4_zext_2): New define_insn_and_split.
6100 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6101 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6102 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6103 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6104 New define_insn_and_split.
6105 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6106 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6107 operands[1] having DImode when mode is SImode.
6109 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6110 always SImode for DIV and MOD in REG_EQUAL notes.
6112 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6115 * match.pd: Fix handling of NaNs in pattern.
6117 2017-09-29 Jeff Law <law@redhat.com>
6119 * sbitmap.c (bitmap_bit_in_range_p): New function.
6120 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6121 * tree-ssa-dse.c (live_bytes_read): New function.
6122 (dse_classify_store): Ignore reads of dead bytes.
6124 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6125 typos and whitespace errors.
6126 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6127 * config/s390/s390.c (s390_emit_prologue): Likewise.
6129 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6132 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6133 with a symbol for LRA.
6135 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6137 PR rtl-optimization/82338
6138 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6140 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6142 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6143 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6144 (qsort_chk): Declare.
6145 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6146 (qsort_chk): New function.
6148 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6150 PR tree-optimization/82337
6151 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6152 phi definition if the PHI result appears in an abnormal PHI.
6153 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6154 basis appears in an abnormal PHI.
6156 2017-09-29 Richard Biener <rguenther@suse.de>
6158 * graphite-isl-ast-to-gimple.c
6159 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6160 (binary_op_to_tree): Use it.
6161 (get_rename_from_scev): Likewise.
6162 (copy_loop_phi_nodes): Likewise.
6163 (copy_bb_and_scalar_dependences): Likewise.
6164 (translate_pending_phi_nodes): Likewise.
6166 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6169 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6170 for movabsq $(i32 << shift), r64.
6172 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6174 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6175 index when encoding %esp as %rsp to avoid 0x67 prefix.
6177 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6179 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6180 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6182 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6184 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6185 Extensions with more than 16 double VFP registers.
6186 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6187 to_clear_mask and all code related to it. Replace the remaining
6188 entry by a sbitmap and adapt code accordingly.
6190 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6192 * brig-builtins.def: Change pure attributes to const.
6194 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6196 * config.gcc (default_gnu_indirect_function): Default to yes for
6197 sparc*-*-linux* with glibc.
6199 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6201 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6202 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6203 when creating .init_array and .fini_array sections with priority
6206 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6209 * config/aarch64/aarch64.c
6210 (aarch64_builtin_support_vector_misalignment): Always return false
6211 when misalignment is unknown.
6213 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6215 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6216 this function to return false if the definition used by the swap
6217 instruction is artificial, or if the memory address from which the
6218 constant value is loaded is not represented by a base address held
6219 in a register or if the base address register is a frame or stack
6220 pointer. Additionally, return false if the base address of the
6221 loaded constant is a SYMBOL_REF but is not considered to be a
6223 (replace_swapped_load_constant): New function.
6224 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6225 loaded constant vector with a load of a swapped constant vector.
6227 2017-09-27 Carl Love <cel@us.ibm.com>
6229 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6230 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6231 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6234 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6236 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6237 first, always call autopref_rank_data otherwise.
6239 2017-09-27 Richard Biener <rguenther@suse.de>
6241 * graphite-scop-detection.c (find_scop_parameters): Move
6242 loop bound handling ...
6243 (gather_bbs::before_dom_children): ... here, avoiding the need
6244 to build scop_info->loop_nest.
6245 (record_loop_in_sese): Remove.
6246 * sese.h (sese_info_t::loop_nest): Remove.
6247 * sese.c (new_sese_info): Do not allocate loop_nest.
6248 (free_sese_info): Do not free loop_nest.
6250 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6253 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6254 lhs from calls if the lhs has addressable type.
6256 2017-09-27 Richard Biener <rguenther@suse.de>
6258 * graphite.h (scop::max_alias_set): New member.
6259 * graphite-scop-detection.c: Remove references to non-existing
6260 --param in comments.
6261 (build_alias_sets): Record the maximum alias set used for drs.
6262 (build_scops): Support zero as unlimited for
6263 --param graphite-max-arrays-per-scop.
6264 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6266 (build_poly_sr_1): ... here. Compute alias set based on the
6267 maximum alias set used for drs rather than
6268 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6270 2017-09-27 Richard Biener <rguenther@suse.de>
6272 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6273 --param loop-block-tile-size=0 to disable tiling.
6275 2017-09-27 Richard Biener <rguenther@suse.de>
6277 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6278 (graphite-max-nb-scop-params): Document special value zero.
6279 * domwalk.h (dom_walker::STOP): New symbolical constant.
6280 (dom_walker::dom_walker): Add optional parameter for bb to
6282 (dom_walker::~dom_walker): Declare.
6283 (dom_walker::before_dom_children): Document STOP return value.
6284 (dom_walker::m_user_bb_to_rpo): New member.
6285 (dom_walker::m_bb_to_rpo): Likewise.
6286 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6287 mapping here if not provided by the user.
6288 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6289 provided by the user.
6290 (dom_walker::STOP): Define.
6291 (dom_walker::walk): Do not compute bb to RPO mapping here.
6292 Support STOP return value from before_dom_children to stop
6294 * graphite-optimize-isl.c (optimize_isl): If the schedule
6295 is the same still generate code if -fgraphite-identity
6296 or -floop-parallelize-all are given.
6297 * graphite-scop-detection.c: Include cfganal.h.
6298 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6300 (gather_bbs::before_dom_children): Return STOP for BBs
6302 (build_scops): Compute bb to RPO mapping and pass it to
6303 the domwalk. Treat --param graphite-max-nb-scop-params=0
6304 as not limiting the number of params.
6305 * graphite.c (graphite_initialize): Remove limit on the
6306 number of basic-blocks in a function.
6307 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6308 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6309 default value of 10.
6311 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6313 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6314 Adjust code to eliminate needing to do the shift right 32-bits
6315 operation after XSCVDPSPN.
6317 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6319 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6320 ((X / Y) != 0 -> X >= Y): Likewise.
6322 2017-09-26 Carl Love <cel@us.ibm.com>
6324 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6325 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6326 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6327 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6328 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6329 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6330 definitions and overloading.
6331 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6332 statement for P9V_BUILTIN_XST_LEN_R.
6333 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6334 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6335 define_expand and define_insn for the instructions and builtins.
6336 * doc/extend.texi: Update the built-in documentation file for the new
6338 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6339 define_insn for the instructions
6341 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6344 * gcc/config/netbsd-protos.h: New file.
6345 * gcc/config/netbsd.c: New file.
6346 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6347 * gcc/config/t-netbsd: New file.
6348 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6349 (tmake_file) Add t-netbsd.
6350 (extra_objs) Add netbsd.o.
6352 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6356 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6358 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6360 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6361 sign extension from a vector register to a GPR by doing a 32-bit
6362 direct move and then an EXTSW.
6363 (extendsi<mode>2 splitter): Likewise.
6364 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6365 right or vector extract after doing XSCVDPSPN. Use
6366 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6368 (movdi_from_sf_zero_ext): Likewise.
6369 (reload_gpr_from_vsxsf): Likewise.
6370 (p8_mfvsrd_4_disf): Delete, no longer used.
6371 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6372 then needing to move the SFmode to a GPR to use the XSCVDPSP
6373 instruction instead of FRSP and XSCVDPSPN.
6374 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6375 it is adjacent to the other XSCVSPDP insns.
6376 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6377 SFmode to be in traditional Altivec registers.
6378 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6379 (vsx_xscvspdpn): Likewise.
6380 (vsx_xscvspdpn_scalar): Likewise.
6382 2017-09-26 Martin Jambor <mjambor@suse.cz>
6384 * tree-sra.c (compare_access_positions): Put integral types first,
6385 stabilize sorting of integral types, remove conditions putting
6386 non-full-precision integers last.
6387 (sort_and_splice_var_accesses): Disable scalarization if a
6388 non-integert would be represented by a non-full-precision integer.
6390 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6392 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6393 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6394 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6395 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6396 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6397 conditionals inside the function instead of around it. Call
6398 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6399 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6401 2017-09-26 Richard Biener <rguenther@suse.de>
6403 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6405 (scop_detection::build_scop_breadth): ... this. Removed.
6406 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6407 (scop_detection::harmful_stmt_in_bb): Likewise.
6408 (scop_detection::graphite_can_represent_stmt): Likewise.
6409 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6410 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6411 (scop_detection::can_represent_loop_1): ... this. Removed.
6412 (scop_detection::harmful_loop_in_region): Simplify after inlining
6413 the above and remove more quadraticness.
6414 (build_scops): Adjust.
6415 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6418 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6421 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6422 REGNO (base) == SP_REG if base is a REG.
6425 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6426 if it is different SSA_NAME.
6427 (optimize_range_tests_cmp_bitwise): New function.
6428 (optimize_range_tests): Call it.
6430 2017-09-26 Richard Biener <rguenther@suse.de>
6432 PR tree-optimization/82321
6433 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6434 for the def being inside the loop.
6436 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6438 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6440 * config/s390/s390-builtins.def: Fix constraint on op4.
6442 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6444 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6445 independent expanders.
6446 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6447 ("vec_ordered", "vec_unordered"): New expanders.
6449 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6451 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6454 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6456 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6457 vec_unpacks_lo_v16qi.
6458 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6460 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6462 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6463 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6464 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6466 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6468 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6470 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6472 ("*vec_slb<mode>"): New insn pattern.
6473 ("vec_shr_<mode>"): New expander.
6474 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6475 and force the shift count operand to V16QImode.
6476 ("vec_srb<mode>"): Set shift count mode to V16QI.
6478 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6480 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6481 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6482 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6484 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6487 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6489 2017-09-26 Richard Biener <rguenther@suse.de>
6491 PR tree-optimization/82320
6492 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6495 2017-09-25 Jeff Law <law@redhat.com>
6497 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6498 prototype for new argument.
6499 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6500 mostly extracted from rs6000_emit_allocate_stack.
6501 (rs6000_emit_probe_stack_range_stack_clash): New function.
6502 (rs6000_emit_allocate_stack): Call
6503 rs6000_emit_probe_stack_range_stack_clash as needed.
6504 (rs6000_emit_probe_stack_range): Add additional argument
6505 to call to gen_probe_stack_range{si,di}.
6506 (output_probe_stack_range): New.
6507 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6508 (output_probe_stack_range_stack_clash): New.
6509 (rs6000_emit_prologue): Emit notes into dump file as requested.
6510 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6511 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6512 Add additional operand and pass it to output_probe_stack_range.
6514 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6516 PR tree-optimization/82163
6517 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6518 (checking_verify_loop_closed_ssa): New parameter.
6519 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6520 (check_loop_closed_ssa_stmt): Delete.
6521 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6522 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6523 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6526 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6528 * brig-builtins.def: Treat HSAIL barrier builtins as
6529 setjmp/longjump style functions.
6531 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6533 * target.def (constant_alignment): New hook.
6534 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6535 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6536 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6537 * doc/tm.texi: Regenerate.
6538 * targhooks.h (default_constant_alignment): Declare.
6539 (constant_alignment_word_strings): Likewise.
6540 * targhooks.c (default_constant_alignment): New function.
6541 (constant_alignment_word_strings): Likewise.
6542 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6543 instead of CONSTANT_ALIGNMENT.
6544 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6545 (force_const_mem): Likewise.
6546 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6547 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6548 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6549 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6550 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6552 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6553 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6554 constant_alignment_word_strings.
6555 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6556 (CONSTANT_ALIGNMENT): Likewise.
6557 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6558 (arm_constant_alignment): New function.
6559 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6560 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6561 constant_alignment_word_strings.
6562 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6563 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6564 constant_alignment_word_strings.
6565 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6566 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6567 (cris_constant_alignment): New function.
6568 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6569 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6570 (epiphany_constant_alignment): New function.
6571 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6572 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6573 constant_alignment_word_strings.
6574 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6575 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6576 constant_alignment_word_strings.
6577 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6578 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6579 constant_alignment_word_strings.
6580 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6581 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6582 * config/i386/i386.c (ix86_constant_alignment): Make static.
6583 Use the same interface as the target hook.
6584 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6585 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6586 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6587 constant_alignment_word_strings.
6588 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6589 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6590 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6591 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6592 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6593 constant_alignment_word_strings.
6594 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6595 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6596 constant_alignment_word_strings.
6597 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6598 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6599 constant_alignment_word_strings.
6600 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6601 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6603 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6604 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6605 * config/mips/mips.c (mips_constant_alignment): New function.
6606 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6607 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6608 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6609 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6610 (mmix_constant_alignment): Make static. Use the same interface
6612 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6613 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6614 constant_alignment_word_strings.
6615 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6616 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6617 constant_alignment_word_strings.
6618 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6619 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6620 constant_alignment_word_strings.
6621 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6622 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6623 (rs6000_constant_alignment): New function.
6624 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6625 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6626 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6627 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6628 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6629 (rs6000_constant_alignment): New function.
6630 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6631 * config/s390/s390.c (s390_constant_alignment): New function.
6632 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6633 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6634 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6635 constant_alignment_word_strings.
6636 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6637 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6638 (sparc_constant_alignment): New function.
6639 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6640 * config/spu/spu.c (spu_constant_alignment): New function.
6641 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6642 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6643 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6644 constant_alignment_word_strings.
6645 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6646 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6647 constant_alignment_word_strings.
6648 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6649 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6650 constant_alignment_word_strings.
6651 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6652 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6653 (visium_constant_alignment): New function.
6654 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6655 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6656 (xtensa_constant_alignment): New function.
6657 * system.h (CONSTANT_ALIGNMENT): Poison.
6659 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6661 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6662 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6663 (rs6000_builtin_valid_without_lhs): New helper function.
6664 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6665 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6667 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6669 * target.h (vec_perm_indices): Use unsigned short rather than
6671 (auto_vec_perm_indices): Likewise.
6672 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6673 Use unsigned int rather than unsigned char.
6674 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6676 2017-09-25 Richard Biener <rguenther@suse.de>
6678 * cfgloop.h (sort_sibling_loops): Declare.
6679 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6680 (sort_sibling_loops): New function sorting the sibling loop list
6682 * graphite.c (graphite_transform_loops): Sort sibling loops.
6684 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6686 * target.def (vec_perm_const_ok): Change sel parameter to
6688 * optabs-query.c (can_vec_perm_p): Update accordingly.
6689 * doc/tm.texi: Regenerate.
6690 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6691 auto_vec_perm_indices and remove separate nelt field.
6692 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6693 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6694 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6695 (aarch64_expand_vec_perm_const): Update accordingly.
6696 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6697 to vec_perm_indices.
6698 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6699 auto_vec_perm_indices and remove separate nelt field.
6700 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6701 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6702 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6704 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6705 to vec_perm_indices.
6706 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6707 sel to vec_perm_indices.
6708 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6709 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6710 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6712 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6715 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6718 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6719 on the FUNCTION_DECL function context if it has a DIE that is a
6722 2017-09-25 Richard Biener <rguenther@suse.de>
6724 PR tree-optimization/82285
6725 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6728 2017-09-25 Tom de Vries <tom@codesourcery.com>
6732 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6735 2017-09-25 Richard Biener <rguenther@suse.de>
6737 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6738 ISL errors other than isl_error_quota happen. Dump if the
6739 schedule is the same.
6740 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6741 errors instead of aborting inside ISL.
6743 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6746 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6747 of libgcc_eh for m64.
6748 * config/i386/darwin64.h: Likewise.
6750 2017-09-25 Richard Biener <rguenther@suse.de>
6753 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6754 attribute for incomplete types nor twice for complete ones.
6756 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6759 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6760 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6763 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6766 * config/i386/i386.opt (mprefer-avx256): Use
6767 ix86_target_flags variable.
6768 * config/i386/i386.c (ix86_target_string): Move
6769 -mprefer-avx256 to flag2_opts.
6771 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6774 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6775 and x != -1 | y != -1 into (x & y) != -1.
6777 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6779 * config.gcc: Add new case statement to set
6780 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6781 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6782 s390x-*-linux* case statements. Added aarch64 to the list of
6783 supported architectures.
6785 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6787 PR tree-optimization/82289
6788 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6789 STMT_VINFO_RELEVANT_P.
6791 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6792 Alan Hayward <alan.hayward@arm.com>
6793 David Sherwood <david.sherwood@arm.com>
6795 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6796 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6797 (extract_range_from_binary_expr_1): Don't call
6798 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6800 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6801 Alan Hayward <alan.hayward@arm.com>
6802 David Sherwood <david.sherwood@arm.com>
6804 * target.def (preferred_vector_alignment): New hook.
6805 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6807 * doc/tm.texi: Regenerate.
6808 * targhooks.h (default_preferred_vector_alignment): Declare.
6809 * targhooks.c (default_preferred_vector_alignment): New function.
6810 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6812 (DR_TARGET_ALIGNMENT): New macro.
6813 (aligned_access_p): Update commentary.
6814 (vect_known_alignment_in_bytes): New function.
6815 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6817 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6818 Calculate the misalignment based on the target alignment rather than
6820 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6821 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6822 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6823 the target alignment, rather than masking the element misalignment
6824 with the number of elements in a vector. Also use the target
6825 alignment when calculating the maximum number of peels.
6826 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6827 instead of TYPE_ALIGN_UNIT.
6828 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6829 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6830 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6831 (vect_create_data_ref_ptr): Likewise.
6832 (vect_setup_realignment): Realign by ANDing with
6833 -DR_TARGET_MISALIGNMENT.
6834 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6835 the number of peels based on DR_TARGET_ALIGNMENT.
6836 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6837 with the guaranteed alignment boundary when deciding whether
6839 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6840 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6841 (ensure_base_align): Remove stmt_info parameter. Get the
6842 target base alignment from DR_TARGET_ALIGNMENT.
6843 (vectorizable_store): Update call accordingly. Interpret
6844 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6846 (vectorizable_load): Likewise.
6848 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6849 Alan Hayward <alan.hayward@arm.com>
6850 David Sherwood <david.sherwood@arm.com>
6852 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6853 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6854 (vect_enhance_data_refs_alignment): Likewise.
6856 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6858 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6859 error. Only quit immediately if parsing is complete.
6860 (BEGIN): Initialize fatal_err and parse_done.
6861 (begin fpu, end fpu): Check number of arguments.
6862 (begin arch, end arch): Likewise.
6863 (begin cpu, end cpu): Likewise.
6864 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6865 (optalias): Likewise.
6867 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6869 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6870 * config/arm/arm-isa.h: Delete. Move definitions to ...
6871 * arm-cpus.in: ... here. Use new feature and fgroup values.
6872 * config/arm/arm.c (arm_option_override): Use lower case for feature
6874 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6875 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6876 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6878 (print_isa_bits_for): New function.
6879 (gen_isa): New function.
6880 (gen_comm_data): Use print_isa_bits_for.
6881 (define feature): New keyword.
6882 (define fgroup): New keyword.
6883 * config/arm/t-arm (TM_H): Remove.
6884 (GTM_H): Add arm-isa.h.
6885 (arm-isa.h): Add rule to generate file.
6886 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6887 case for feature bit names.
6889 2017-09-22 Richard Biener <rguenther@suse.de>
6891 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6893 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6894 print of no dependency loops ...
6895 * graphite.c (graphite_transform_loops): ... here.
6896 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6898 (same_close_phi_node, remove_duplicate_close_phi,
6899 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6900 (canonicalize_loop_closed_ssa): ... here and simplify.
6901 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6902 (optimize_isl): Use dump_printf_loc to tell when we stopped
6903 optimizing because of an ISL timeout.
6905 2017-09-22 Richard Biener <rguenther@suse.de>
6907 PR tree-optimization/82291
6908 * tree-if-conv.c (predicate_mem_writes): Make sure to
6909 remove writes in blocks predicated with false.
6911 2017-09-22 Richard Biener <rguenther@suse.de>
6913 * sese.c: Include cfganal.h.
6914 (if_region_set_false_region): Remove.
6915 (create_if_region_on_edge): Likewise.
6916 (move_sese_in_condition): Re-implement without destroying
6919 2017-09-22 Richard Biener <rguenther@suse.de>
6921 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6922 Verify both BBs contain loop PHI nodes before dispatching to
6924 (graphite_regenerate_ast_isl): Do not recompute dominators,
6925 do not verify three times. Restructure for clarity.
6926 * graphite-scop-detection.c (same_close_phi_node,
6927 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6928 defined_in_loop_p, canonicalize_loop_closed_ssa,
6929 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6930 checking and SSA rewrite, move to ...
6931 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6932 (graphite_initialize): Do not pass in ctx, do not reset the
6933 SCEV cache, compute only dominators.
6934 (graphite_transform_loops): Allocate ISL ctx after
6935 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6936 Maintain post-dominators only around build_scops.
6937 * sese.c (if_region_set_false_region): Make static. Free
6938 and recompute dominators.
6939 (move_sese_in_condition): Assert we don't get called with
6940 post-dominators computed.
6941 * sese.h (if_region_set_false_region): Remove.
6943 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6945 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6946 mode attribute for TARGET_AVX512VL.
6948 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6950 * config/i386/i386.opt (mprefer-avx256): New option.
6951 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6953 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6954 for TARGET_PREFER_AVX256.
6955 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6957 2017-09-21 Jeff Law <law@redhat.com>
6959 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6960 Fix dump output if the only stack space is for pushed registers.
6962 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6964 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6967 2017-09-21 Martin Sebor <msebor@redhat.com>
6970 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6971 code (in C++) or code that triggers warnings.
6973 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6975 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6977 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6979 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6980 * sched-rgn.c: Ditto.
6981 * sel-sched-ir.c: Ditto.
6983 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6985 * toplev.h (set_random_seed): Adjust return type.
6986 * toplev.c (init_local_tick): Move eager initialization of random_seed
6987 to get_random_seed. Adjust comment.
6988 (init_random_seed): Inline to get_random_seed, delete.
6989 (get_random_seed): Initialize random_seed lazily.
6990 (set_random_seed): Do not return previous value.
6991 (print_switch_value): Do not call get_random_seed.
6993 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6995 * cgraph.c (delete_function_version): New, broken out from...
6996 (cgraph_node::delete_function_version): ...here. Rename to
6997 cgraph_node::delete_function_version_by_decl. Update all uses.
6998 (cgraph_node::remove): Call delete_function_version.
7000 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7003 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7004 VAR_DECLs to which addressable non-volatile parameters are mapped
7005 and for id->retvar after the return value assignment. Clear
7006 id->retval and id->retbnd after inlining.
7008 2017-09-21 Richard Biener <rguenther@suse.de>
7010 PR tree-optimization/82276
7011 PR tree-optimization/82244
7012 * tree-vrp.c (build_assert_expr_for): Set
7013 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7015 (remove_range_assertions): Revert earlier change.
7017 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7020 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7022 2017-09-21 Richard Biener <rguenther@suse.de>
7024 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7025 Restore valid IL after code generation errors.
7026 * graphite.c (graphite_transform_loops): Diagnose code
7027 generation issues as MSG_MISSED_OPTIMIZATION and continue
7028 with processing SCOPs.
7030 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7031 Alan Hayward <alan.hayward@arm.com>
7032 David Sherwood <david.sherwood@arm.com>
7034 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7035 rather than choosing between plus_constant and gen_rtx_<CODE>.
7036 * expr.c (emit_push_insn): Likewise.
7037 (expand_expr_real_2): Likewise.
7039 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7040 Alan Hayward <alan.hayward@arm.com>
7041 David Sherwood <david.sherwood@arm.com>
7043 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7045 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7046 Alan Hayward <alan.hayward@arm.com>
7047 David Sherwood <david.sherwood@arm.com>
7049 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7050 calling tree_to_uhwi.
7052 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7053 Alan Hayward <alan.hayward@arm.com>
7054 David Sherwood <david.sherwood@arm.com>
7056 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7057 INTEGER_CST rather than a negative test for ADDR_EXPR.
7059 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7060 Alan Hayward <alan.hayward@arm.com>
7061 David Sherwood <david.sherwood@arm.com>
7063 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7064 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7066 2017-09-21 Richard Biener <rguenther@suse.de>
7068 PR tree-optimization/71351
7069 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7070 graphite_create_new_loop_guard): Remove, fold remaining parts
7072 (translate_isl_ast_node_for): ... here and simplify.
7074 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7077 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7078 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7079 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7080 alternative always use QI mode, for -Os imov (=R,R) alternative
7081 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7084 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7085 Jeff Law <law@redhat.com>
7087 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7088 (allocate_stack_space): New function, partially extracted from
7090 (s390_emit_prologue): Track offset to most recent stack probe.
7091 Code to allocate space moved into allocate_stack_space.
7092 Dump actions when no stack is allocated.
7093 (s390_prologue_plus_offset): New function.
7094 (s390_emit_stack_probe): Likewise.
7096 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7098 * common.opt (Wa, Wl, Wp, g, gz=): Add
7100 (gno-column-info): Remove.
7101 (gcolumn-info): Drop RejectNegative.
7103 (gno-record-gcc-switches): Remove.
7104 (grecord-gcc-switches): Drop RejectNegative.
7105 (gno-split-dwarf): Remove.
7106 (gsplit-dwarf): Drop RejectNegative.
7107 (gno-strict-dwarf): Remove.
7108 (gstrict-dwarf): Drop RejectNegative.
7109 * config/darwin.opt (gfull, gused): Add RejectNegative.
7110 * dwarf2out.c (gen_producer_string): Drop
7111 gno-record-gcc-switches handler.
7112 * optc-gen.awk: Add g to prefixes with negative forms.
7113 * opts-common.c (remapping_prefix_p): New.
7114 (find_opt): Check it.
7115 (generate_canonical_option): Test g prefix.
7116 (option_map): Add -gno- mapping.
7117 (add_misspelling_candidates): Check remapping_prefix_p.
7119 2017-09-20 Jeff Law <law@redhat.com>
7121 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7122 thinko in stack clash protection support.
7124 * explow.c (compute_stack_clash_protection_loop_data): Use
7125 CONST_INT_P instead of explicit test. Verify object is a
7126 CONST_INT_P before looking at INTVAL.
7127 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7128 instead of explicit test.
7130 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7133 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7134 address instead of to r1 and r11.
7136 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7138 * config.gcc: Support "knm".
7139 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7140 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7142 * config/i386/i386.c (m_KNM): Define.
7143 (processor_target_table): Add "knm".
7145 (ix86_option_override_internal): Add "knm".
7146 (ix86_issue_rate): Add PROCESSOR_KNM.
7147 (ix86_adjust_cost): Ditto.
7148 (ia32_multipass_dfa_lookahead): Ditto.
7149 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7150 (fold_builtin_cpu): Add M_INTEL_KNM.
7151 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7152 (processor_type): Add PROCESSOR_KNM.
7153 * config/i386/x86-tune.def: Add m_KNM.
7154 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7156 2017-09-20 Richard Biener <rguenther@suse.de>
7158 PR tree-optimization/80213
7159 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7160 are allowed in empty BBs as well.
7161 (canonicalize_loop_closed_ssa): Also look for other complex
7163 (scop_detection::get_sese): Include the loop-closed PHI block
7165 (scop_detection::merge_sese): Remove code adding extra blocks.
7166 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7167 (build_scops): Assert the final returned scop is invalid.
7169 2017-09-20 Richard Biener <rguenther@suse.de>
7171 PR tree-optimization/82264
7172 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7174 (vn_phi_lookup): Likewise.
7175 (vn_phi_insert): Likewise.
7177 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7179 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7180 that fits into uhwi or shwi, add DW_AT_const_value regardless
7181 of early_dwarf without going through RTL, using add_AT_unsigned
7184 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7185 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7186 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7187 (DEBUG_MACRO_SECTION): Likewise.
7188 (DEBUG_LTO_MACRO_SECTION): Likewise.
7189 (DEBUG_STR_DWO_SECTION): Likewise.
7190 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7191 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7192 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7193 (DEBUG_LTO_LINE_STR_SECTION): Define.
7194 (init_sections_and_labels): Initialize debug_line_str_section
7195 variable. Initialize debug_loc_section for -gdwarf-5 to
7196 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7198 2017-09-20 Richard Biener <rguenther@suse.de>
7200 * graphite-sese-to-poly.c (extract_affine): Properly handle
7201 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7203 2017-09-20 Richard Biener <rguenther@suse.de>
7205 PR tree-optimization/81373
7206 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7207 Force SESE live-out defs to be handled even if they are
7210 2017-09-19 Jeff Law <law@redhat.com>
7212 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7213 nothing for stack adjustments with REG_STACK_CHECK.
7214 * sched-deps.c (parse_add_or_inc): Reject insns with
7215 REG_STACK_CHECK from dependency breaking.
7216 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7217 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7218 * reg-notes.def (STACK_CHECK): New note.
7220 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7221 (ix86_expand_prologue): Dump stack clash info as needed.
7222 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7224 * function.c (dump_stack_clash_frame_info): New function.
7225 * function.h (dump_stack_clash_frame_info): Prototype.
7226 (enum stack_clash_probes): New enum.
7228 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7229 flag_stack_clash_protection.
7230 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7231 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7232 (arm_frame_pointer_required): Likewise.
7233 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7234 (ia64_expand_prologue): Likewise.
7235 * config/mips/mips.c (mips_expand_prologue): Likewise.
7236 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7237 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7238 (sparc_flat_expand_prologue): Likewise.
7239 * config/spu/spu.c (spu_expand_prologue): Likewise.
7241 * explow.c: Include "params.h".
7242 (anti_adjust_stack_and_probe_stack_clash): New function.
7243 (get_stack_check_protect): Likewise.
7244 (compute_stack_clash_protection_loop_data): Likewise.
7245 (emit_stack_clash_protection_loop_start): Likewise.
7246 (emit_stack_clash_protection_loop_end): Likewise.
7247 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7248 Use anti_adjust_stack_and_probe_stack_clash.
7249 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7250 (emit_stack_clash_protection_loop_start): Likewise.
7251 (emit_stack_clash_protection_loop_end): Likewise.
7252 * rtl.h (get_stack_check_protect): Prototype.
7253 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7254 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7255 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7257 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7259 * doc/tm.texi: Rebuilt.
7260 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7261 get_stack_check_protect.
7262 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7263 * config/arm/arm.c (arm_expand_prologue): Likewise.
7264 (arm_frame_pointer_required): Likewise.
7265 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7266 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7267 * config/mips/mips.c (mips_expand_prologue): Likewise.
7268 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7269 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7270 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7271 (sparc_flat_expand_prologue): Likewise.
7273 * common.opt (-fstack-clash-protection): New option.
7274 * flag-types.h (enum stack_check_type): Note difference between
7275 -fstack-check= and -fstack-clash-protection.
7276 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7277 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7278 * toplev.c (process_options): Issue warnings/errors for cases
7279 not handled with -fstack-clash-protection.
7280 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7281 (-fstack-check): Note additional problem with -fstack-check=generic.
7282 Note that -fstack-check is primarily for Ada and refer users
7283 to -fstack-clash-protection for stack-clash-protection.
7284 Document new params for stack clash protection.
7286 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7288 * config/i386/i386.c (ix86_split_long_move): Do not handle
7289 address used for LEA in a special way.
7291 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7293 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7295 2017-09-19 Martin Sebor <msebor@redhat.com>
7298 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7299 of incompatible types.
7301 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7304 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7306 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7308 2017-09-19 Richard Biener <rguenther@suse.de>
7310 PR tree-optimization/82244
7311 * tree-vrp.c (remove_range_assertions): Do not propagate
7312 a constant to abnormals but replace the assert with a copy.
7314 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7316 PR rtl-optimization/57878
7317 PR rtl-optimization/68988
7318 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7319 avoidance test involving non_reload_pseudos. Move frequency test
7320 below the general fragmentation avoidance test.
7322 2017-09-19 Richard Biener <rguenther@suse.de>
7324 PR tree-optimization/69728
7325 * graphite-sese-to-poly.c (schedule_error): New global.
7326 (add_loop_schedule): Handle empty domain by failing the
7328 (build_original_schedule): Handle schedule_error.
7330 2017-09-19 Richard Biener <rguenther@suse.de>
7332 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7333 Do not iterate to sibling loops but only to siblings of inner
7336 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7339 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7340 that the registers are different.
7342 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7344 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7345 to processor_model and "amdfam17h" to arch_names_table.
7346 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7348 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7351 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7353 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7354 Alan Hayward <alan.hayward@arm.com>
7355 David Sherwood <david.sherwood@arm.com>
7357 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7359 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7361 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7362 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7363 vect_schedule_slp_instance.
7364 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7365 Update call to vect_slp_analyze_node_operations. Simplify return
7367 (vect_slp_analyze_bb_1): Update call accordingly.
7368 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7369 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7370 (vect_schedule_slp): Update call accordingly.
7372 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7373 Alan Hayward <alan.hayward@arm.com>
7374 David Sherwood <david.sherwood@arm.com>
7376 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7377 with types that aren't in fact scalar.
7379 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7381 * tree-vect-slp.c (vect_record_max_nunits): New function,
7383 (vect_build_slp_tree_1): ...here.
7384 (vect_build_slp_tree_2): Call it for phis too.
7386 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7388 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7389 to vect_get_vec_def_for_operand when getting the mask operand.
7391 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7392 Alan Hayward <alan.hayward@arm.com>
7393 David Sherwood <david.sherwood@arm.com>
7395 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7398 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7399 Alan Hayward <alan.hayward@arm.com>
7400 David Sherwood <david.sherwood@arm.com>
7402 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7403 calculation for vector booleans.
7405 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7406 Alan Hayward <alan.hayward@arm.com>
7407 David Sherwood <david.sherwood@arm.com>
7409 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7411 (vect_transform_stmt): ...here.
7412 (vect_analyze_stmt): Use it instead of calling
7413 vectorizable_live_operation directly.
7415 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7417 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7418 non-SIMT targets in acc vector loops.
7420 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7422 * configure.ac: Add arc and check if assembler supports gdwarf2.
7423 * configure: Regenerate.
7425 2017-09-18 Richard Biener <rguenther@suse.de>
7427 PR tree-optimization/82220
7428 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7429 epilogue niters from the min_profitable_iters compute.
7431 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7434 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7435 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7436 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7438 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7441 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7442 switching to a new text section.
7444 2017-09-18 Richard Biener <rguenther@suse.de>
7446 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7448 (build_alias_set): Reject aliases with no access function.
7450 2017-09-18 Richard Biener <rguenther@suse.de>
7452 PR tree-optimization/79622
7453 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7455 (build_cross_bb_scalars_use): Likewise.
7457 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7459 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7461 2017-09-18 Alan Modra <amodra@gmail.com>
7464 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7465 stack_pointer_rtx for count 0. Update comments. Break up
7466 large rtl expression.
7468 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7470 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7471 Increase to 20 bytes.
7472 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7473 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7474 or avx version of the stub.
7476 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7479 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7480 compute the minimum stack alignment. Also update preferred stack
7481 boundary for leaf functions.
7483 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7485 PR tree-optimization/82228
7486 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7489 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7491 * common/config/nds32/nds32-common.c
7492 (nds32_option_optimization_table): Refine formatting.
7493 (nds32_option_optimization_table): Use -fsched-pressure and
7494 -fomit-frame-pointer for specific optimization level.
7496 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7498 * config/nds32/nds32.c: Refine formatting and comments.
7499 * config/nds32/nds32.h: Likewise.
7500 * config/nds32/nds32.md: Likewise.
7501 * config/nds32/nds32-cost.c: Likewise.
7502 * config/nds32/nds32-isr.c: Likewise.
7503 * config/nds32/nds32-md-auxiliary.c: Likewise.
7504 * config/nds32/nds32-multiple.md: Likewise.
7505 * config/nds32/nds32-predicates.c: Likewise.
7507 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7508 Jakub Jelinek <jakub@redhat.com>
7510 Add support for -std=c++2a.
7511 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7513 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7515 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7518 * doc/extend.texi (Common Function Attributes): Add
7519 references to ARM, AArch64, and S/390 specific attributes.
7520 (Function Specific Option Pragmas): Add AArch64 and S/390
7521 to list of back ends that support the target pragma.
7523 2017-09-15 Nathan Sidwell <nathan@acm.org>
7525 * doc/standards.texi: Fix C++17 description. Update URLs for
7528 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7530 * common.opt (Wcast-align=strict): New warning option.
7531 * doc/invoke.texi: Document -Wcast-align=strict.
7533 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7535 * cgraph.h (cgraph_thunk_info): Add comments.
7536 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7537 assert for VIRTUAL_* arguments stricter.
7539 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7541 PR tree-optimization/71026
7542 * match.pd: Move RDIV patterns from fold-const.c
7543 * fold-const.c (distribute_real_division): Removed.
7544 (fold_binary_loc): Remove calls to distribute_real_divison.
7546 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7548 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7549 c++1z and gnu++1z as deprecated. Change other references to
7550 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7551 Change -Wc++1z-compat to -Wc++17-compat.
7552 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7553 * dwarf2out.c (highest_c_language): Handle C++17.
7554 (gen_compile_unit_die): Likewise.
7556 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7558 PR rtl-optimization/82192
7559 * combine.c (make_extraction): Don't look through non-paradoxical
7560 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7563 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7564 Alan Hayward <alan.hayward@arm.com>
7565 David Sherwood <david.sherwood@arm.com>
7567 * target.def (function_arg_offset): New hook.
7568 * targhooks.h (default_function_arg_offset): Declare.
7569 * targhooks.c (default_function_arg_offset): New function.
7570 * function.c (locate_and_pad_parm): Use
7571 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7572 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7573 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7574 * doc/tm.texi: Regenerate.
7575 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7576 * config/spu/spu.c (spu_function_arg_offset): New function.
7577 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7578 * system.h (FUNCTION_ARG_OFFSET): Poison.
7580 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7581 Alan Hayard <alan.hayward@arm.com>
7582 David Sherwood <david.sherwood@arm.com>
7584 * target.def (truly_noop_truncation): New hook.
7585 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7586 than TRULY_NOOP_TRUNCATION.
7587 * hooks.h (hook_bool_uint_uint_true): Declare.
7588 * hooks.c (hook_bool_uint_uint_true): New function.
7589 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7590 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7591 * doc/tm.texi: Regenerate.
7592 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7593 rather than TRULY_NOOP_TRUNCATION in comments.
7594 (simplify_comparison): Likewise.
7595 (record_truncated_value): Likewise.
7596 * expmed.c (extract_bit_field_1): Likewise.
7597 (extract_split_bit_field): Likewise.
7598 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7599 instead of TRULY_NOOP_TRUNCATION.
7600 * function.c (assign_parm_setup_block): Likewise.
7601 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7602 * rtlhooks.c: Include target.h.
7603 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7604 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7605 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7606 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7607 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7608 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7609 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7610 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7611 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7612 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7613 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7614 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7615 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7616 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7617 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7618 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7619 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7620 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7621 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7622 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7623 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7624 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7625 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7626 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7627 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7628 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7629 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7630 rather than TRULY_NOOP_TRUNCATION in comments.
7631 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7632 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7633 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7634 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7635 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7636 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7637 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7638 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7639 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7640 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7641 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7642 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7643 rather than TRULY_NOOP_TRUNCATION in comments.
7644 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7645 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7646 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7647 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7648 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7649 TRULY_NOOP_TRUNCATION condition.
7650 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7651 (TRULY_NOOP_TRUNCATION): Delete.
7652 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7653 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7654 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7655 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7656 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7657 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7658 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7659 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7660 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7661 rather than TRULY_NOOP_TRUNCATION in comments.
7662 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7663 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7664 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7665 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7666 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7667 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7669 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7672 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7673 (*cmp_ior): Likewise.
7674 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7675 (*ior_scc_scc_cmp): Likewise.
7676 (*and_scc_scc): Likewise.
7677 (*and_scc_scc_cmp): Likewise.
7679 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7680 Alan Hayard <alan.hayward@arm.com>
7681 David Sherwood <david.sherwood@arm.com>
7683 * target.def (can_change_mode_class): New hook.
7684 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7685 (hard_regno_nregs): Likewise.
7686 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7687 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7688 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7689 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7690 (LOAD_EXTEND_OP): Update accordingly.
7691 * doc/tm.texi: Regenerate.
7692 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7693 CANNOT_CHANGE_MODE_CLASS.
7694 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7695 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7696 * combine.c (simplify_set): Update accordingly.
7697 * emit-rtl.c (validate_subreg): Likewise.
7698 * recog.c (general_operand): Likewise.
7699 * regcprop.c (mode_change_ok): Likewise.
7700 * reload1.c (choose_reload_regs): Likewise.
7701 (inherit_piecemeal_p): Likewise.
7702 * rtlanal.c (simplify_subreg_regno): Likewise.
7703 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7704 instead of CANNOT_CHANGE_MODE_CLASS.
7705 (reload_cse_simplify_operands): Likewise.
7706 * reload.c (push_reload): Use targetm.can_change_mode_class
7707 instead of CANNOT_CHANGE_MODE_CLASS.
7708 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7709 REG_CANNOT_CHANGE_MODE_P.
7710 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7711 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7712 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7713 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7714 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7715 (arm_can_change_mode_class): New function.
7716 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7717 than CANNOT_CHANGE_MODE_CLASS in comments.
7718 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7719 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7720 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7721 (ix86_can_change_mode_class): ...this new function, inverting the
7722 sense of the return value.
7723 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7724 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7725 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7726 (ia64_can_change_mode_class): New function.
7727 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7728 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7729 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7730 (m32c_can_change_mode_class): ...this new function, inverting the
7731 sense of the return value.
7732 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7733 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7734 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7735 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7736 (mips_can_change_mode_class): ...this new function, inverting the
7737 sense of the return value.
7738 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7739 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7740 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7741 (msp430_can_change_mode_class): New function.
7742 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7743 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7744 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7745 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7746 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7747 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7748 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7749 (pa_cannot_change_mode_class): Replace with...
7750 (pa_can_change_mode_class): ...this new function, inverting the
7751 sense of the return value.
7752 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7753 than CANNOT_CHANGE_MODE_CLASS in comments.
7754 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7755 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7756 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7757 (pdp11_cannot_change_mode_class): Replace with...
7758 (pdp11_can_change_mode_class): ...this new function, inverting the
7759 sense of the return value.
7760 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7761 * config/powerpcspe/powerpcspe-protos.h
7762 (rs6000_cannot_change_mode_class_ptr): Delete.
7763 * config/powerpcspe/powerpcspe.c
7764 (rs6000_cannot_change_mode_class_ptr): Delete.
7765 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7766 (rs6000_option_override_internal): Assign to
7767 targetm.can_change_mode_class instead of
7768 rs6000_cannot_change_mode_class_ptr.
7769 (rs6000_cannot_change_mode_class): Replace with...
7770 (rs6000_can_change_mode_class): ...this new function, inverting the
7771 sense of the return value.
7772 (rs6000_debug_cannot_change_mode_class): Replace with...
7773 (rs6000_debug_can_change_mode_class): ...this new function.
7774 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7775 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7776 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7777 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7778 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7780 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7781 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7782 (rs6000_option_override_internal): Assign to
7783 targetm.can_change_mode_class instead of
7784 rs6000_cannot_change_mode_class_ptr.
7785 (rs6000_cannot_change_mode_class): Replace with...
7786 (rs6000_can_change_mode_class): ...this new function, inverting the
7787 sense of the return value.
7788 (rs6000_debug_cannot_change_mode_class): Replace with...
7789 (rs6000_debug_can_change_mode_class): ...this new function.
7790 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7791 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7792 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7793 (s390_can_change_mode_class): ...this new function, inverting the
7794 sense of the return value.
7795 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7796 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7797 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7798 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7799 (sh_cannot_change_mode_class): Replace with...
7800 (sh_can_change_mode_class): ...this new function, inverting the
7801 sense of the return value.
7802 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7803 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7804 (sparc_can_change_mode_class): New function.
7805 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7806 * config/spu/spu.c (spu_can_change_mode_class): New function.
7807 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7808 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7809 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7810 (visium_can_change_mode_class): New function.
7811 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7813 2017-09-15 Richard Biener <rguenther@suse.de>
7815 PR tree-optimization/82217
7816 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7817 but not undefined case.
7819 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7822 * postreload.c (reload_cse_simplify_operands): Skip
7823 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7825 2017-09-15 Richard Biener <rguenther@suse.de>
7827 PR tree-optimization/68823
7828 * graphite-scop-detection.c (build_alias_set): If we have a
7829 possible dependence check whether we can handle them by just
7830 looking at the DRs DR_ACCESS_FNs.
7831 (build_scops): If build_alias_set fails, fail the SCOP.
7833 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7835 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7836 to support float128 built-in functions that require the ISA 3.0
7838 (BU_FLOAT128_3_HW): Likewise.
7839 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7841 (FMAF128): Likewise.
7843 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7844 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7845 floating point instructions.
7846 (rs6000_invalid_builtin): Likewise.
7847 (rs6000_builtin_mask_names): Likewise.
7848 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7849 (RS6000_BTM_FLOAT128_HW): Likewise.
7850 (RS6000_BTM_COMMON): Likewise.
7851 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7853 * doc/extend.texi (RS/6000 built-in functions): Document the
7854 IEEE 128-bit floating point square root and fused multiply-add
7857 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7859 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7860 reg (r2) isn't in the set of registers defined in the prologue.
7862 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7863 Alan Hayward <alan.hayward@arm.com>
7864 David Sherwood <david.sherwood@arm.com>
7866 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7867 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7868 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7869 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7870 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7872 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7873 max_vectorization_factor.
7874 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7876 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7877 Alan Hayward <alan.hayward@arm.com>
7878 David Sherwood <david.sherwood@arm.com>
7880 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7881 (vect_worthwhile_without_simd_p): Declare.
7882 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7883 (vectorizable_reduction): Use it.
7884 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7885 (vectorizable_operation): Likewise.
7887 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7888 Alan Hayward <alan.hayward@arm.com>
7889 David Sherwood <david.sherwood@arm.com>
7891 * tree-vectorizer.h (vect_get_num_copies): New function.
7892 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7893 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7894 (vectorizable_induction): Likewise.
7895 (vectorizable_live_operation): Likewise.
7896 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7897 (vectorizable_bswap): Likewise.
7898 (vectorizable_call): Likewise.
7899 (vectorizable_conversion): Likewise.
7900 (vectorizable_assignment): Likewise.
7901 (vectorizable_shift): Likewise.
7902 (vectorizable_operation): Likewise.
7903 (vectorizable_store): Likewise.
7904 (vectorizable_load): Likewise.
7905 (vectorizable_condition): Likewise.
7906 (vectorizable_comparison): Likewise.
7907 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7909 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7910 Alan Hayward <alan.hayward@arm.com>
7911 David Sherwood <david.sherwood@arm.com>
7913 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7914 of vect_init_vector.
7916 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7917 Alan Hayward <alan.hayward@arm.com>
7918 David Sherwood <david.sherwood@arm.com>
7920 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7921 an inline wrapper that provides a location.
7922 (gimple_build_vector): Likewise.
7923 * gimple-fold.c (gimple_build_vector_from_val): New function.
7924 (gimple_build_vector): Likewise.
7925 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7926 functions to build the initial value. Always return a gimple value.
7927 (get_initial_defs_for_reduction): Likewise. Only compute
7929 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7930 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7931 (vectorizable_induction): Use gimple_build_vector rather than
7934 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7935 Alan Hayward <alan.hayward@arm.com>
7936 David Sherwood <david.sherwood@arm.com>
7938 * target.h (vec_perm_indices): New typedef.
7939 (auto_vec_perm_indices): Likewise.
7940 * optabs-query.h: Include target.h
7941 (can_vec_perm_p): Take a vec_perm_indices *.
7942 * optabs-query.c (can_vec_perm_p): Likewise.
7943 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7944 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7945 * tree-vect-generic.c (lower_vec_perm): Likewise.
7946 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7947 (vect_grouped_load_supported): Likewise.
7948 (vect_shift_permute_load_chain): Likewise.
7949 (vect_permute_store_chain): Use auto_vec_perm_indices.
7950 (vect_permute_load_chain): Likewise.
7951 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7952 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7953 Update uses of can_vec_perm_p.
7954 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7955 mode with a number of elements. Take a vec_perm_indices *.
7956 (vect_create_epilog_for_reduction): Update accordingly.
7957 Use auto_vec_perm_indices.
7958 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7959 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7960 (vect_transform_slp_perm_load): Likewise.
7961 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7962 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7963 (vect_gen_perm_mask_checked): Likewise.
7964 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7965 (vect_gen_perm_mask_checked): Likewise.
7966 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7967 (vectorizable_store): Likewise.
7968 (vectorizable_load): Likewise.
7969 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7970 (vectorizable_bswap): Likewise.
7972 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7973 Alan Hayward <alan.hayward@arm.com>
7974 David Sherwood <david.sherwood@arm.com>
7976 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7977 * tree.c (build_vector): Likewise.
7978 (build_vector_from_ctor): Update accordingly.
7979 (build_vector_from_val): Likewise.
7980 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7981 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7982 * tree-vect-generic.c (add_rshift): Likewise.
7983 (expand_vector_divmod): Likewise.
7984 (optimize_vector_constructor): Likewise.
7985 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7986 (vect_transform_slp_perm_load): Likewise.
7987 (vect_schedule_slp_instance): Likewise.
7988 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7989 (vectorizable_call): Likewise.
7990 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7991 * expmed.c (make_tree): Likewise.
7992 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7993 a vector passed to build_vector.
7994 (fold_convert_const): Likewise.
7995 (exact_inverse): Likewise.
7996 (fold_ternary_loc): Likewise.
7997 (fold_relational_const): Likewise.
7998 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7999 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8000 (const_unop): Likewise. Store the reduction accumulator in a
8001 variable rather than an array.
8002 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8003 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8004 the new vector, rather than constructing it after the input arrays.
8005 (native_interpret_vector): Use auto_vec<tree> when building
8006 a vector passed to build_vector. Add elements in order.
8007 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8008 auto_vec<tree> when building a vector passed to build_vector.
8009 (vect_create_epilog_for_reduction): Likewise.
8010 (vectorizable_induction): Likewise.
8011 (get_initial_def_for_reduction): Likewise. Fix indentation of
8013 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8015 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8016 passed to build_vector.
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-core.h (tree_base::u): Add an "nelts" field.
8023 (tree_vector): Use VECTOR_CST_NELTS as the length.
8024 * tree.c (tree_size): Likewise.
8025 (make_vector): Initialize VECTOR_CST_NELTS.
8026 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8027 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8028 TYPE_VECTOR_SUBPARTS.
8029 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8030 as the number of units, setting it from VECTOR_CST_NELTS.
8031 (const_vector_from_tree): Likewise.
8032 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8033 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8034 (fold_negate_expr_1): Likewise.
8035 (fold_convert_const): Likewise.
8036 (const_binop): Likewise. Differentiate the number of output and
8038 (const_unop): Likewise.
8039 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8040 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8041 in cases that did the opposite.
8043 2017-09-14 Richard Biener <rguenther@suse.de>
8045 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8048 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8050 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8052 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8055 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8056 if and where to split a bb, except for splitting before debug insn
8057 sequences followed by non-label real insn. Delete debug insns
8058 in between basic blocks.
8060 * combine.c (make_compound_operation_int): Formatting fixes.
8062 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8063 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8064 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8065 * config/sol2.h (LINK_EH_SPEC): Likewise.
8066 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8067 * config/s390/linux.h (LINK_SPEC): Likewise.
8068 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8069 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8070 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8071 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8072 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8073 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8074 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8075 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8077 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8079 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8081 (ENDFILE_LINUX_SPEC): Likewise.
8082 (LINK_EH_SPEC): Likewise.
8083 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8084 (LINK_OS_LINUX_SPEC32): Likewise.
8085 (LINK_OS_LINUX_SPEC64): Likewise.
8086 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8087 (LINK_OS_LINUX_SPEC): Likewise.
8089 2017-09-13 Martin Liska <mliska@suse.cz>
8092 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8093 CASE_HIGH is NULL_TREE.
8095 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8096 Alan Hayward <alan.hayward@arm.com>
8097 David Sherwood <david.sherwood@arm.com>
8099 * target.def (secondary_memory_needed): New hook.
8100 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8101 instead of SECONDARY_MEMORY_NEEDED.
8102 (secondary_memory_needed_mode): Likewise.
8103 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8104 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8105 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8106 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8107 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8108 * doc/tm.texi: Regenerate.
8109 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8110 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8111 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8112 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8113 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8114 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8115 mode argument first and change the reg_class arguments to reg_class_t.
8116 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8117 Make static. Update the call to inline_secondary_memory_needed.
8118 (ix86_register_move_cost): Update the call to
8119 inline_secondary_memory_needed.
8120 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8121 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8123 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8124 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8126 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8127 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8128 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8129 and match hook interface. Add comment from mips.h.
8130 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8131 * config/mmix/mmix.md (truncdfsf2): Refer to
8132 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8134 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8135 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8136 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8137 (pa_secondary_memory_needed): New function.
8138 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8139 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8140 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8141 (pdp11_secondary_memory_needed): Make static and match hook interface.
8142 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8143 * config/powerpcspe/powerpcspe-protos.h
8144 (rs6000_secondary_memory_needed_ptr): Delete.
8145 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8147 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8148 (rs6000_option_override_internal): Assign to
8149 targetm.secondary_memory_needed rather than
8150 rs6000_secondary_memory_needed_ptr.
8151 (rs6000_secondary_memory_needed): Match hook interface.
8152 (rs6000_debug_secondary_memory_needed): Likewise.
8153 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8154 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8155 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8156 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8157 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8158 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8160 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8161 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8162 (rs6000_option_override_internal): Assign to
8163 targetm.secondary_memory_needed rather than
8164 rs6000_secondary_memory_needed_ptr.
8165 (rs6000_secondary_memory_needed): Match hook interface.
8166 (rs6000_debug_secondary_memory_needed): Likewise.
8167 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8168 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8169 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8170 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8171 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8172 (sparc_secondary_memory_needed): New function.
8173 * lra-constraints.c (check_and_process_move): Refer to
8174 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8176 (curr_insn_transform): Likewise.
8177 (process_alt_operands): Use targetm.secondary_memory_needed
8178 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8179 (check_secondary_memory_needed_p): Likewise.
8180 (choose_split_class): Likewise.
8181 * reload.c: Unconditionally include code that was previously
8182 conditional on SECONDARY_MEMORY_NEEDED.
8183 (push_secondary_reload): Use targetm.secondary_memory_needed
8184 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8185 (push_reload): Likewise.
8186 * reload1.c: Unconditionally include code that was previously
8187 conditional on SECONDARY_MEMORY_NEEDED.
8188 (choose_reload_regs): Use targetm.secondary_memory_needed
8189 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8190 (gen_reload): Likewise.
8191 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8193 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8194 Alan Hayward <alan.hayward@arm.com>
8195 David Sherwood <david.sherwood@arm.com>
8197 * target.def (secondary_memory_needed_mode): New hook:
8198 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8199 * targhooks.h (default_secondary_memory_needed_mode): New function.
8200 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8201 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8202 * doc/tm.texi: Regenerate.
8203 * lra-constraints.c (check_and_process_move): Use
8204 targetm.secondary_memory_needed_mode instead of
8205 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8206 (curr_insn_transform): Likewise.
8207 * reload.c (get_secondary_mem): Likewise.
8208 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8209 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8211 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8212 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8213 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8214 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8215 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8217 * config/powerpcspe/powerpcspe-protos.h
8218 (rs6000_secondary_memory_needed_mode): Delete.
8219 * config/powerpcspe/powerpcspe.c
8220 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8221 (rs6000_secondary_memory_needed_mode): Make static.
8222 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8223 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8225 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8227 (rs6000_secondary_memory_needed_mode): Make static.
8228 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8229 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8230 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8231 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8232 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8234 (sparc_secondary_memory_needed_mode): New function.
8235 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8237 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8239 * config/aarch64/constraints.md (Umq): New constraint.
8240 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8242 (mov<mode>): Update condition.
8244 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8246 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8247 when bitposition is the same.
8249 2017-09-13 Richard Biener <rguenther@suse.de>
8251 * dwarf2out.c (output_die_symbol): Remove.
8252 (output_die): Do not output a DIEs symbol.
8254 2017-09-13 Richard Biener <rguenther@suse.de>
8257 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8258 default-def to avoid breaking iterator update with the weird
8259 interaction with cgraph_update_edges_for_call_stmt_node.
8261 2017-09-13 Richard Biener <rguenther@suse.de>
8263 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8264 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8265 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8266 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8268 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8270 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8271 Disable pc relative literal load irrespective of
8272 TARGET_FIX_ERR_A53_84341 for default.
8274 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8276 * config/sparc/sparc.c (output_return): Output the source location of
8277 the insn in the delay slot, if any.
8278 (output_sibcall): Likewise.
8280 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8283 * common.opt (-static-pie): New alias.
8284 (shared): Negate static-pie.
8285 (-no-pie): Update help text.
8287 (static-pie): New option.
8288 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8289 -static-pie support.
8290 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8291 (LINK_EH_SPEC): Likewise.
8292 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8293 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8294 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8295 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8296 (init_gcc_specs): Likewise.
8297 (init_spec): Likewise.
8298 (display_help): Update help message for -pie.
8299 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8302 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8304 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8305 (movdi_aarch64): Likewise.
8306 (movti_aarch64): Likewise.
8308 2017-09-12 Simon Wright <simon@pushface.org>
8311 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8312 calculation of the minor version, always output as 0.
8314 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8317 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8318 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8319 on it early, rather than manual conversion late. For
8320 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8321 instead of performing manual conversion.
8323 2017-09-12 Carl Love <cel@us.ibm.com>
8325 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8326 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8328 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8329 VMULOSW): Add definitions.
8330 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8331 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8332 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8333 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8334 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8336 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8338 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8340 (movti_aarch64): Likewise.
8341 (movdf_aarch64): Likewise.
8342 (movtf_aarch64): Likewise.
8343 (load_pairdi): Likewise.
8344 (store_pairdi): Likewise.
8345 (load_pairdf): Likewise.
8346 (store_pairdf): Likewise.
8347 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8348 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8349 (ldr_got_small_<mode>): Likewise.
8350 (ldr_got_small_28k_<mode>): Likewise.
8351 (ldr_got_tiny): Likewise.
8352 * config/aarch64/iterators.md (ldst_sz): New.
8353 (ldpstp_sz): Likewise.
8354 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8356 (thunderx_load): Split load_8 to load_16.
8357 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8359 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8360 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8361 (xgene1_store_pair): Split store_8 to store_16.
8362 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8363 (falkor_st_0_st_sd): Split store_8 to store_16.
8365 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8367 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8368 and store1/2/3/4 to store_4/8/12/16.
8369 * config/aarch64/aarch64.md: Update for rename.
8370 * config/arm/arm.md: Likewise.: Likewise.
8371 * config/arm/arm.c: Likewise.
8372 * config/arm/thumb1.md: Likewise.
8373 * config/arm/thumb2.md: Likewise.
8374 * config/arm/vfp.md: Likewise.
8375 * config/arm/arm-generic.md: Likewise.
8376 * config/arm/arm1020e.md: Likewise.
8377 * config/arm/arm1026ejs.md: Likewise.
8378 * config/arm/arm1136jfs.md: Likewise.
8379 * config/arm/arm926ejs.md: Likewise.
8380 * config/arm/cortex-a15.md: Likewise.
8381 * config/arm/cortex-a17.md: Likewise.
8382 * config/arm/cortex-a5.md: Likewise.
8383 * config/arm/cortex-a53.md: Likewise.
8384 * config/arm/cortex-a57.md: Likewise.
8385 * config/arm/cortex-a7.md: Likewise.
8386 * config/arm/cortex-a8.md: Likewise.
8387 * config/arm/cortex-a9.md: Likewise.
8388 * config/arm/cortex-m4.md: Likewise.
8389 * config/arm/cortex-m7.md: Likewise.
8390 * config/arm/cortex-r4.md: Likewise.
8391 * config/arm/exynos-m1.md: Likewise.
8392 * config/arm/fa526.md: Likewise.
8393 * config/arm/fa606te.md: Likewise.
8394 * config/arm/fa626te.md: Likewise.
8395 * config/arm/fa726te.md: Likewise.
8396 * config/arm/fmp626.md: Likewise.
8397 * config/arm/iwmmxt.md: Likewise.
8398 * config/arm/ldmstm.md: Likewise.
8399 * config/arm/marvell-pj4.md: Likewise.
8400 * config/arm/xgene1.md: Likewise.
8401 * config/aarch64/thunderx.md: Likewise.
8402 * config/aarch64/thunderx2t99.md: Likewise.
8403 * config/aarch64/falkor.md: Likewise.
8405 2017-09-12 Martin Liska <mliska@suse.cz>
8407 * attribs.c (private_lookup_attribute): New function.
8408 * attribs.h (private_lookup_attribute): Declared here.
8409 (lookup_attribute): Called from this place.
8411 2017-09-12 Richard Biener <rguenther@suse.de>
8413 PR tree-optimization/82157
8414 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8415 stmts with side-effects.
8417 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8418 Alan Hayward <alan.hayward@arm.com>
8419 David Sherwood <david.sherwood@arm.com>
8421 * target.def (hard_regno_nregs): New hook.
8422 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8423 * targhooks.h (default_hard_regno_nregs): Declare.
8424 * targhooks.c (default_hard_regno_nregs): New function.
8425 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8426 (TARGET_HARD_REGNO_NREGS): ...this hook.
8427 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8428 (CLASS_MAX_NREGS): Likewise.
8429 * doc/tm.texi: Regenerate.
8430 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8431 instead of HARD_REGNO_NREGS.
8432 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8433 HARD_REGNO_NREGS in the comment.
8434 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8435 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8436 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8437 Return an unsigned int.
8438 (TARGET_HARD_REGNO_NREGS): Redefine.
8439 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8440 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8441 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8442 (arc_hard_regno_nregs): New function.
8443 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8444 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8445 (arm_hard_regno_nregs): New function.
8446 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8447 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8448 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8449 (TARGET_HARD_REGNO_NREGS): Redefine.
8450 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8451 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8452 (HARD_REGNO_NREGS): Delete.
8453 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8454 (cr16_hard_regno_nregs): New function.
8455 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8456 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8457 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8458 (cris_hard_regno_nregs): New function.
8459 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8460 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8461 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8462 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8463 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8464 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8465 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8466 (frv_hard_regno_nregs): Make static. Take and return an
8468 (frv_class_max_nregs): Remove outdated copy of documentation.
8469 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8470 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8471 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8472 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8473 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8474 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8475 (TARGET_HARD_REGNO_NREGS): Redefine.
8476 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8477 (CLASS_MAX_NREGS): Update comment.
8478 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8479 (ia64_hard_regno_nregs): New function.
8480 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8481 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8482 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8483 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8484 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8486 (m32c_hard_regno_nregs): Likewise. Make static.
8487 (TARGET_HARD_REGNO_NREGS): Redefine.
8488 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8489 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8490 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8491 (m68k_hard_regno_nregs): New function.
8492 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8493 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8494 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8495 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8496 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8497 Take and return an unsigned int.
8498 (TARGET_HARD_REGNO_NREGS): Redefine.
8499 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8500 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8501 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8502 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8503 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8504 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8505 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8506 (msp430_hard_regno_nregs): Make static. Take and return an
8508 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8509 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8510 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8511 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8512 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8513 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8514 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8515 (TARGET_HARD_REGNO_NREGS): Redefine.
8516 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8517 (PA_HARD_REGNO_NREGS): ...this.
8518 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8519 (PA_HARD_REGNO_NREGS): ...this.
8520 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8521 (pa_hard_regno_nregs): New function.
8522 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8523 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8524 (pdp11_hard_regno_nregs): New function.
8525 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8526 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8527 (rs6000_hard_regno_nregs_hook): New function.
8528 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8529 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8530 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8531 Take and return an unsigned int. Move earlier in file.
8532 (TARGET_HARD_REGNO_NREGS): Redefine.
8533 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8534 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8535 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8536 (rl78_hard_regno_nregs): Make static. Take and return an
8538 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8539 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8540 (rs6000_hard_regno_nregs_hook): New function.
8541 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8542 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8543 (TARGET_HARD_REGNO_NREGS): Redefine.
8544 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8545 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8546 instead of HARD_REGNO_NREGS.
8547 (s390_hard_regno_nregs): New function.
8548 (s390_hard_regno_mode_ok): Add comment from s390.h.
8549 (TARGET_HARD_REGNO_NREGS): Redefine.
8550 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8551 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8552 (sh_hard_regno_nregs): New function.
8553 (sh_pass_in_reg_p): Use it.
8554 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8555 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8556 (sparc_hard_regno_nregs): New function.
8557 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8558 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8559 (spu_function_arg_advance): Use it, supplying a valid register number.
8560 (TARGET_HARD_REGNO_NREGS): Redefine.
8561 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8562 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8563 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8564 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8565 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8566 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8567 (CLASS_MAX_NREGS): Remove copy of old documentation.
8568 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8569 (visium_hard_regno_nregs): New function.
8570 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8571 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8572 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8573 xtensa_hard_regno_nregs): New function.
8574 * system.h (HARD_REGNO_NREGS): Poison.
8576 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8578 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8579 hard_regno_nregs instead of HARD_REGNO_NREGS.
8580 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8581 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8582 (c6x_expand_epilogue): Likewise.
8583 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8584 (frv_read_iacc_argument): Likewise.
8585 * config/sh/sh.c: Include regs.h.
8586 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8587 (regs_used): Likewise.
8588 (output_stack_adjust): Likewise.
8589 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8590 * expmed.c: Include regs.h.
8591 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8592 * ree.c: Include regs.h.
8593 (combine_reaching_defs): Use hard_regno_nregs instead of
8595 (add_removable_extension): Likewise.
8597 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8599 * regs.h (hard_regno_nregs): Turn into a function.
8600 (end_hard_regno): Update accordingly.
8601 * caller-save.c (setup_save_areas): Likewise.
8602 (save_call_clobbered_regs): Likewise.
8603 (replace_reg_with_saved_mem): Likewise.
8604 (insert_restore): Likewise.
8605 (insert_save): Likewise.
8606 * combine.c (can_change_dest_mode): Likewise.
8607 (move_deaths): Likewise.
8608 (distribute_notes): Likewise.
8609 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8610 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8611 (rs6000_split_multireg_move): Likewise.
8612 (rs6000_register_move_cost): Likewise.
8613 (rs6000_memory_move_cost): Likewise.
8614 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8615 (rs6000_split_multireg_move): Likewise.
8616 (rs6000_register_move_cost): Likewise.
8617 (rs6000_memory_move_cost): Likewise.
8618 * cselib.c (cselib_reset_table): Likewise.
8619 (cselib_lookup_1): Likewise.
8620 * emit-rtl.c (set_mode_and_regno): Likewise.
8621 * function.c (aggregate_value_p): Likewise.
8622 * ira-color.c (setup_profitable_hard_regs): Likewise.
8623 (check_hard_reg_p): Likewise.
8624 (calculate_saved_nregs): Likewise.
8625 (assign_hard_reg): Likewise.
8626 (improve_allocation): Likewise.
8627 (calculate_spill_cost): Likewise.
8628 * ira-emit.c (modify_move_list): Likewise.
8629 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8630 (ira_hard_reg_in_set_p): Likewise.
8631 * ira.c (setup_reg_mode_hard_regset): Likewise.
8632 (clarify_prohibited_class_mode_regs): Likewise.
8633 (check_allocation): Likewise.
8634 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8635 (lra_setup_reg_renumber): Likewise.
8636 (setup_try_hard_regno_pseudos): Likewise.
8637 (spill_for): Likewise.
8638 (assign_hard_regno): Likewise.
8639 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8640 * lra-constraints.c (in_class_p): Likewise.
8641 (lra_constraint_offset): Likewise.
8642 (simplify_operand_subreg): Likewise.
8643 (lra_constraints): Likewise.
8644 (split_reg): Likewise.
8645 (split_if_necessary): Likewise.
8646 (invariant_p): Likewise.
8647 (inherit_in_ebb): Likewise.
8648 * lra-lives.c (process_bb_lives): Likewise.
8649 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8650 (get_hard_regs): Likewise.
8651 (do_remat): Likewise.
8652 * lra-spills.c (assign_spill_hard_regs): Likewise.
8653 * mode-switching.c (create_pre_exit): Likewise.
8654 * postreload.c (reload_combine_recognize_pattern): Likewise.
8655 * recog.c (peep2_find_free_register): Likewise.
8656 * regcprop.c (kill_value_regno): Likewise.
8657 (set_value_regno): Likewise.
8658 (copy_value): Likewise.
8659 (maybe_mode_change): Likewise.
8660 (find_oldest_value_reg): Likewise.
8661 (copyprop_hardreg_forward_1): Likewise.
8662 * regrename.c (check_new_reg_p): Likewise.
8663 (regrename_do_replace): Likewise.
8664 * reload.c (push_reload): Likewise.
8665 (combine_reloads): Likewise.
8666 (find_dummy_reload): Likewise.
8667 (operands_match_p): Likewise.
8668 (find_reloads): Likewise.
8669 (find_equiv_reg): Likewise.
8670 (reload_adjust_reg_for_mode): Likewise.
8671 * reload1.c (count_pseudo): Likewise.
8672 (count_spilled_pseudo): Likewise.
8673 (find_reg): Likewise.
8674 (clear_reload_reg_in_use): Likewise.
8675 (free_for_value_p): Likewise.
8676 (allocate_reload_reg): Likewise.
8677 (choose_reload_regs): Likewise.
8678 (reload_adjust_reg_for_temp): Likewise.
8679 (emit_reload_insns): Likewise.
8680 (delete_output_reload): Likewise.
8681 * rtlanal.c (subreg_get_info): Likewise.
8682 * sched-deps.c (sched_analyze_reg): Likewise.
8683 * sel-sched.c (init_regs_for_mode): Likewise.
8684 (mark_unavailable_hard_regs): Likewise.
8685 (choose_best_reg_1): Likewise.
8686 (verify_target_availability): Likewise.
8687 * valtrack.c (dead_debug_insert_temp): Likewise.
8688 * var-tracking.c (track_loc_p): Likewise.
8689 (emit_note_insn_var_location): Likewise.
8690 * varasm.c (make_decl_rtl): Likewise.
8691 * reginfo.c (choose_hard_reg_mode): Likewise.
8692 (init_reg_modes_target): Refer directly to
8693 this_target_regs->x_hard_regno_nregs.
8695 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8697 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8698 instead of hard_regno_nregs.
8700 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8702 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8703 end_hard_regno instead of hard_regno_nregs.
8704 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8705 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8706 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8707 * ira-color.c (improve_allocation): Likewise.
8708 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8709 * lra-lives.c (mark_regno_live): Likewise.
8710 (mark_regno_dead): Likewise.
8711 * lra-remat.c (operand_to_remat): Likewise.
8712 * lra.c (collect_non_operand_hard_regs): Likewise.
8713 * postreload.c (reload_combine_note_store): Likewise.
8714 (move2add_valid_value_p): Likewise.
8715 * reload.c (regno_clobbered_p): Likewise.
8717 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8719 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8721 * config/v850/v850.c (v850_reorg): Likewise.
8722 * reload.c (refers_to_regno_for_reload_p): Likewise.
8723 (find_equiv_reg): Likewise.
8724 * reload1.c (reload_reg_reaches_end_p): Likewise.
8726 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8728 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8730 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8731 * config/arm/arm.c (output_move_neon): Likewise.
8732 (arm_attr_length_move_neon): Likewise.
8733 (neon_split_vcombine): Likewise.
8734 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8735 (c6x_mark_reg_written): Likewise.
8736 (c6x_dwarf_register_span): Likewise.
8737 * config/i386/i386.c (ix86_save_reg): Likewise.
8738 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8739 (rws_access_reg): Likewise.
8740 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8741 * mode-switching.c (create_pre_exit): Likewise.
8742 * ree.c (combine_reaching_defs): Likewise.
8743 (add_removable_extension): Likewise.
8744 * regcprop.c (find_oldest_value_reg): Likewise.
8745 (copyprop_hardreg_forward_1): Likewise.
8746 * reload.c (reload_inner_reg_of_subreg): Likewise.
8747 (push_reload): Likewise.
8748 (combine_reloads): Likewise.
8749 (find_dummy_reload): Likewise.
8750 (reload_adjust_reg_for_mode): Likewise.
8751 * reload1.c (find_reload_regs): Likewise.
8752 (forget_old_reloads_1): Likewise.
8753 (reload_reg_free_for_value_p): Likewise.
8754 (reload_adjust_reg_for_temp): Likewise.
8755 (emit_reload_insns): Likewise.
8756 (delete_output_reload): Likewise.
8757 * sel-sched.c (choose_best_reg_1): Likewise.
8758 (choose_best_pseudo_reg): Likewise.
8760 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8761 Alan Hayward <alan.hayward@arm.com>
8762 David Sherwood <david.sherwood@arm.com>
8764 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8765 * target.def (slow_unaligned_access): New hook.
8766 * targhooks.h (default_slow_unaligned_access): Declare.
8767 * targhooks.c (default_slow_unaligned_access): New function.
8768 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8769 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8770 * doc/tm.texi: Regenerate.
8771 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8772 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8773 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8775 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8776 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8778 (rs6000_slow_unaligned_access): New function.
8779 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8780 (expand_block_compare): Likewise.
8781 (expand_strn_compare): Likewise.
8782 (rs6000_rtx_costs): Likewise.
8783 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8784 (riscv_slow_unaligned_access): Likewise.
8785 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8786 (riscv_slow_unaligned_access_p): ...this and make static.
8787 (riscv_option_override): Update accordingly.
8788 (riscv_slow_unaligned_access): New function.
8789 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8790 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8791 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8792 (rs6000_slow_unaligned_access): New function.
8793 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8794 (rs6000_rtx_costs): Likewise.
8795 * config/rs6000/rs6000-string.c (expand_block_compare)
8796 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8797 of SLOW_UNALIGNED_ACCESS.
8798 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8799 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8800 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8801 of SLOW_UNALIGNED_ACCESS.
8802 * expmed.c (simple_mem_bitfield_p): Likewise.
8803 * expr.c (alignment_for_piecewise_move): Likewise.
8804 (emit_group_load_1): Likewise.
8805 (emit_group_store): Likewise.
8806 (copy_blkmode_from_reg): Likewise.
8807 (emit_push_insn): Likewise.
8808 (expand_assignment): Likewise.
8809 (store_field): Likewise.
8810 (expand_expr_real_1): Likewise.
8811 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8812 * lra-constraints.c (simplify_operand_subreg): Likewise.
8813 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8814 * gimple-ssa-store-merging.c: Likewise in block comment at start
8816 * tree-ssa-strlen.c: Include target.h.
8817 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8818 of SLOW_UNALIGNED_ACCESS.
8819 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8821 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8823 PR rtl-optimization/82185
8824 * expmed.c (emit_store_flag_int): Only test tem if it has been
8827 2017-09-12 Richard Biener <rguenther@suse.de>
8830 * match.pd ((FTYPE) N CMP CST): Fix typo.
8832 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8834 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8836 (mips_near_type_p): Add 'short_call' attribute as a synonym
8838 * doc/extend.texi (short_call): Document new function attribute.
8840 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8843 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8844 assertion check that in the condition.
8845 (get_atomic_generic_size): Likewise. Before testing if parameter
8846 has pointer type, if it has array type, call for C++
8847 default_conversion to perform array-to-pointer conversion.
8849 2017-09-12 Richard Biener <rguenther@suse.de>
8851 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8852 for operations we cannot scalarize.
8854 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8856 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8857 vectors heap vectors. Clean up comments.
8858 Make visited_bbs a reference.
8859 (profitable_jump_thread_path): Make GC
8860 vectors heap vectors. Clean up comments.
8862 (convert_and_register_jump_thread_path): Make GC vectors heap
8864 (check_subpath_and_update_thread_path): Same. Clean up comments.
8865 Make visited_bbs a reference.
8866 (handle_phi): Abstract common code to to
8867 register_jump_thread_path_if_profitable.
8868 Rename VAR_BB to DEF_BB.
8870 Make GC vectors heap vectors.
8871 Make visited_bbs a reference.
8872 (handle_assignment): Same.
8873 (register_jump_thread_path_if_profitable): New.
8874 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8876 Make GC vectors heap vectors. Clean up comments.
8877 Make visited_bbs a reference.
8878 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8879 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8882 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8885 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8886 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8888 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8890 Revert r251800 and r251799.
8892 2017-09-11 Martin Jambor <mjambor@suse.cz>
8895 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8896 arguments in advance.
8897 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8898 use it to find predecessor edges.
8899 (naive_outof_ssa): Collect vector of predecessors.
8901 2017-09-08 Jason Merrill <jason@redhat.com>
8903 PR c++/70029 - ICE with ref-qualifier and -flto
8904 * langhooks.h (struct lang_hooks_for_types): Add
8905 copy_lang_qualifiers.
8906 * attribs.c (build_type_attribute_qual_variant): Use it.
8907 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8909 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8910 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8912 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8915 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8916 (*mulsi3_sp64): New instruction.
8917 (mulsi3): New expander.
8919 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8921 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8923 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8925 * sancov.c: Include memmodel.h.
8927 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8930 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8933 2017-09-07 Carl Love <cel@us.ibm.com>
8935 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8936 the sldi instruction.
8938 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8940 * sancov.c: Include tm_p.h.
8942 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8945 * output.h (switch_to_other_text_partition): New declaration.
8946 * varasm.c (switch_to_other_text_partition): New function.
8947 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8948 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8949 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8950 to the other text partition before emitting LCL label and switch back
8951 after emitting the word after it.
8953 2017-09-07 Richard Biener <rguenther@suse.de>
8955 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8956 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8957 critical edges here, after loop init.
8958 (pass_data_pre): Remove PROP_no_crit_edges flags.
8959 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8960 for valueization of call args to avoid leaking VN_TOP.
8961 (visit_use): Assert we do not visit default defs.
8962 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8963 Use error_mark_node to more easily detect leaking VN_TOP.
8964 All default-defs are varying, not VN_TOP. Mark them visited.
8965 (run_scc_vn): Make code match comment.
8967 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8969 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8970 OPTION_MASK_FLOAT128_KEYWORD.
8971 (POWERPC_MASKS): Likewise.
8972 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8973 support for the -mfloat128-type option, and make -mfloat128
8974 default on PowerPC Linux systems. Define or undefine
8975 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8976 Define __float128 to be __ieee128 if IEEE 128-bit support is
8977 enabled, or undefine it.
8978 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8979 Delete defining __FLOAT128_TYPE__.
8980 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8981 -mfloat128-type option and make -mfloat128 default on PowerPC
8983 (TARGET_FLOAT128_TYPE): Likewise.
8984 (-mfloat128-type): Likewise.
8985 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8986 Delete the -mfloat128-type option and make -mfloat128 default on
8987 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8988 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8989 128-bit floating point is enabled. Change tests from using
8990 -mfloat128-type to -mfloat128.
8991 (rs6000_mangle_type): Use the correct mangling for the __float128
8992 type even if normal long double is restricted to 64-bits.
8993 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8995 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8996 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8997 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8999 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9000 documentation for -mfloat128.
9002 2017-09-06 Olivier Hainque <hainque@adacore.com>
9004 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9006 2017-09-06 Wish Wu <wishwu007@gmail.com>
9007 Jakub Jelinek <jakub@redhat.com>
9009 * asan.c (initialize_sanitizer_builtins): Add
9010 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9011 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9012 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9013 BT_FN_VOID_UINT64_PTR variables.
9014 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9015 (BT_FN_VOID_UINT16_UINT16): Likewise.
9016 (BT_FN_VOID_UINT32_UINT32): Likewise.
9017 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9018 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9019 (BT_FN_VOID_UINT64_PTR): Likewise.
9020 * common.opt (flag_sanitize_coverage): New variable.
9021 (fsanitize-coverage=trace-pc): Remove.
9022 (fsanitize-coverage=): Add.
9023 * flag-types.h (enum sanitize_coverage_code): New enum.
9024 * fold-const.c (fold_range_test): Disable non-short-circuit
9025 optimization if flag_sanitize_coverage.
9026 (fold_truth_andor): Likewise.
9027 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9028 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9029 (coverage_sanitizer_opts): New array.
9030 (get_closest_sanitizer_option): Add OPTS argument, handle also
9031 OPT_fsanitize_coverage_.
9032 (parse_sanitizer_options): Adjusted to also handle
9033 OPT_fsanitize_coverage_.
9034 (common_handle_option): Add OPT_fsanitize_coverage_.
9035 * sancov.c (instrument_comparison, instrument_switch): New function.
9036 (sancov_pass): Add trace-cmp support.
9037 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9038 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9039 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9040 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9041 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9042 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9043 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9044 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9045 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9046 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9048 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9050 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9051 error. Only quit immediately if parsing is complete.
9052 (BEGIN): Initialize fatal_err and parse_done.
9053 (begin fpu, end fpu): Check number of arguments.
9054 (begin arch, end arch): Likewise.
9055 (begin cpu, end cpu): Likewise.
9056 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9057 (optalias): Likewise.
9059 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9061 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9062 * config/arm/arm-isa.h: Delete. Move definitions to ...
9063 * arm-cpus.in: ... here. Use new feature and fgroup values.
9064 * config/arm/arm.c (arm_option_override): Use lower case for feature
9066 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9067 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9068 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9070 (print_isa_bits_for): New function.
9071 (gen_isa): New function.
9072 (gen_comm_data): Use print_isa_bits_for.
9073 (define feature): New keyword.
9074 (define fgroup): New keyword.
9075 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9076 (arm-isa.h): Add rule to generate file.
9077 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9078 case for feature bit names.
9080 2017-09-06 Richard Biener <rguenther@suse.de>
9082 * tree-ssa-pre.c (NECESSARY): Remove.
9083 (create_expression_by_pieces): Do not touch pass-local flags.
9084 (insert_into_preds_of_block): Likewise.
9085 (do_pre_regular_insertion): Likewise.
9086 (eliminate_insert): Likewise.
9087 (eliminate_dom_walker::before_dom_children): Likewise.
9088 (fini_eliminate): Do not look at inserted_exprs.
9089 (mark_operand_necessary): Remove.
9090 (remove_dead_inserted_code): Replace with simple work-list
9091 algorithm based on inserted_exprs and SSA uses.
9092 (pass_pre::execute): Re-order fini_eliminate and
9093 remove_dead_inserted_code.
9095 2017-09-06 Olivier Hainque <hainque@adacore.com>
9097 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9098 for VxWorks 7. Adjust surrounding comments.
9100 2017-09-06 Richard Biener <rguenther@suse.de>
9102 * gimple-ssa-strength-reduction.c
9103 (find_candidates_dom_walker::before_dom_children): Also allow
9106 2017-09-06 Richard Biener <rguenther@suse.de>
9108 PR tree-optimization/82108
9109 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9110 for gap in the non-permutation SLP case.
9112 2017-09-06 Martin Jambor <mjambor@suse.cz>
9114 PR tree-optimization/82078
9115 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9116 add_access_to_work_queue...
9117 (build_accesses_from_assign): ...here.
9118 (propagate_all_subaccesses): Make sure racc is the group
9119 representative, if there is one.
9121 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9124 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9127 2017-09-06 Richard Biener <rguenther@suse.de>
9129 * gimple-ssa-strength-reduction.c
9130 (find_candidates_dom_walker::before_doom_children): Use a
9131 type and not a mode check.
9133 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9136 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9138 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9140 2017-09-05 Jeff Law <law@redhat.com>
9142 PR tree-optimization/64910
9143 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9144 cases where we have 3 or more operands.
9146 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9149 * omp-low.c (lower_omp_for): Recompute tree invariant if
9150 gimple_omp_for_initial/final is ADDR_EXPR.
9153 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9154 into gimple val before gimplification fo the COND_EXPR.
9156 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9158 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9159 REGION_COPY argument.
9160 (thread_through_all_blocks): Remove unused argument to
9161 duplicate_thread_path.
9163 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9164 Alan Hayward <alan.hayward@arm.com>
9165 David Sherwood <david.sherwood@arm.com>
9167 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9168 Take a scalar_mode rather than a machine_mode.
9169 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9170 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9171 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9172 (aarch64_gen_adjusted_ldpstp): Likewise.
9173 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9176 Alan Hayward <alan.hayward@arm.com>
9177 David Sherwood <david.sherwood@arm.com>
9179 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9180 Take a scalar_int_mode instead of a machine_mode.
9181 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9182 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9183 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9184 (aarch64_simd_attr_length_rglist): Delete.
9185 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9186 a scalar_int_mode instead of a machine_mode.
9187 (aarch64_add_offset): Likewise.
9188 (aarch64_internal_mov_immediate): Likewise
9189 (aarch64_add_constant_internal): Likewise.
9190 (aarch64_add_constant): Likewise.
9191 (aarch64_movw_imm): Likewise.
9192 (aarch64_rtx_arith_op_extract_p): Likewise.
9193 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9194 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9195 Remove assert that the mode isn't a vector.
9196 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9197 (aarch64_expand_mov_immediate): Update calls after above changes.
9198 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9199 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9200 (aarch64_move_imm): Likewise.
9201 (aarch64_can_const_movi_rtx_p): Likewise.
9202 (aarch64_strip_extend): Likewise.
9203 (aarch64_extr_rtx_p): Likewise.
9204 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9205 a CONST_INT when the mode parameter is VOIDmode.
9206 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9208 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9210 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9211 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9212 (bitwise_type_for_mode): Update accordingly.
9214 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9216 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9217 * stor-layout.c (mode_for_size_tree): Likewise.
9218 (mode_for_array): Update accordingly.
9219 (layout_decl): Likewise.
9220 (compute_record_mode): Likewise. Only set the mode once.
9222 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9224 * target.def (get_mask_mode): Change return type to opt_mode.
9226 * doc/tm.texi: Regenerate.
9227 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9228 * targhooks.c (default_get_mask_mode): Likewise.
9229 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9230 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9231 targetm.get_mask_mode.
9232 * tree.c (build_truth_vector_type): Likewise.
9234 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9236 * machmode.h (mode_for_vector): Return an opt_mode.
9237 * stor-layout.c (mode_for_vector): Likewise.
9238 (mode_for_int_vector): Update accordingly.
9239 (layout_type): Likewise.
9240 * config/i386/i386.c (emit_memmov): Likewise.
9241 (ix86_expand_set_or_movmem): Likewise.
9242 (ix86_expand_vector_init): Likewise.
9243 (ix86_get_mask_mode): Likewise.
9244 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9246 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9247 * expmed.c (extract_bit_field_1): Likewise.
9248 * expr.c (expand_expr_real_2): Likewise.
9249 * optabs-query.c (can_vec_perm_p): Likewise.
9250 (can_vec_mask_load_store_p): Likewise.
9251 * optabs.c (expand_vec_perm): Likewise.
9252 * targhooks.c (default_get_mask_mode): Likewise.
9253 * tree-vect-stmts.c (vectorizable_store): Likewise.
9254 (vectorizable_load): Likewise.
9255 (get_vectype_for_scalar_type_and_size): Likewise.
9257 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9259 * machmode.h (mode_for_int_vector): New function.
9260 * stor-layout.c (mode_for_int_vector): Likewise.
9261 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9262 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9263 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9264 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9265 (s390_expand_vcond): Likewise.
9267 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9269 * machmode.h (opt_machine_mode): New type.
9270 (opt_mode<T>): Allow construction from anything that can be
9272 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9273 (mode_for_size): Return an opt_machine_mode.
9274 * stor-layout.c (mode_for_size): Likewise.
9275 (mode_for_size_tree): Update call accordingly.
9276 (bitwise_mode_for_mode): Likewise.
9277 (make_fract_type): Likewise.
9278 (make_accum_type): Likewise.
9279 * caller-save.c (replace_reg_with_saved_mem): Update call
9281 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9282 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9283 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9284 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9285 * expmed.c (extract_bit_field_1): Likewise.
9286 * reload.c (get_secondary_mem): Likewise.
9287 * varasm.c (assemble_integer): Likewise.
9288 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9291 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9293 * machmode.h (decimal_float_mode_for_size): New function.
9294 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9295 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9296 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9297 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9298 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9299 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9301 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9303 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9304 (get_builtin_sync_mode): Likewise.
9305 (expand_ifn_atomic_compare_exchange): Likewise.
9306 (expand_builtin_atomic_clear): Likewise.
9307 (expand_builtin_atomic_test_and_set): Likewise.
9308 (fold_builtin_atomic_always_lock_free): Likewise.
9309 * calls.c (compute_argument_addresses): Likewise.
9310 (emit_library_call_value_1): Likewise.
9311 (store_one_arg): Likewise.
9312 * combine.c (combine_instructions): Likewise.
9313 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9314 * config/arm/arm.c (arm_function_value): Likewise.
9315 (aapcs_allocate_return_reg): Likewise.
9316 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9317 * config/i386/i386.c (construct_container): Likewise.
9318 (ix86_gimplify_va_arg): Likewise.
9319 (ix86_expand_sse_cmp): Likewise.
9320 (emit_memmov): Likewise.
9321 (emit_memset): Likewise.
9322 (expand_small_movmem_or_setmem): Likewise.
9323 (ix86_expand_pextr): Likewise.
9324 (ix86_expand_pinsr): Likewise.
9325 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9326 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9328 * config/mips/mips.c (mips_function_value_1) Likewise.
9329 (mips_block_move_straight): Likewise.
9330 (mips_expand_ins_as_unaligned_store): Likewise.
9331 * config/powerpcspe/powerpcspe.c
9332 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9333 (rs6000_darwin64_record_arg_flush): Likewise.
9334 * config/rs6000/rs6000.c
9335 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9336 (rs6000_darwin64_record_arg_flush): Likewise.
9337 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9338 (sparc_function_value_1): Likewise.
9339 * config/spu/spu.c (adjust_operand): Likewise.
9340 (spu_emit_branch_or_set): Likewise.
9341 (arith_immediate_p): Likewise.
9342 * emit-rtl.c (gen_lowpart_common): Likewise.
9343 * expr.c (expand_expr_real_1): Likewise.
9344 * function.c (assign_parm_setup_block): Likewise.
9345 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9346 * reload1.c (alter_reg): Likewise.
9347 * stor-layout.c (mode_for_vector): Likewise.
9348 (layout_type): Likewise.
9350 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9352 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9353 (spu_convert_move): Likewise.
9354 * lower-subreg.c (resolve_simple_move): Likewise.
9356 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9359 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9360 define_insn to a define_expand.
9361 (altivec_vsum2sws_direct): New define_insn.
9362 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9364 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9366 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9369 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9374 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9375 realign stack if stack alignment needed is less than incoming
9378 2017-09-05 Marek Polacek <polacek@redhat.com>
9381 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9384 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9386 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9388 2017-09-05 Richard Biener <rguenther@suse.de>
9390 PR tree-optimization/82084
9391 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9393 2017-09-05 Richard Biener <rguenther@suse.de>
9395 PR tree-optimization/82102
9396 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9398 2017-09-05 Martin Liska <mliska@suse.cz>
9400 PR tree-optimization/82032
9401 * tree-cfg.c (generate_range_test): New function.
9402 * tree-cfg.h (generate_range_test): Declared here.
9403 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9404 (cleanup_control_expr_graph): Use it.
9405 * tree-switch-conversion.c (try_switch_expansion): Remove
9407 (emit_case_nodes): Use generate_range_test.
9409 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9412 * config/i386/i386.md (*<btsc><mode>_mask): Add
9413 TARGET_USE_BT to insn constraint.
9414 (*btr<mode>_mask): Ditto.
9416 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9418 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9419 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9421 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9424 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9425 TARGET_NEON and TARGET_IWMMXT.
9426 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9427 TARGET_NEON and TARGET_IWMMXT.
9428 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9430 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9432 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9433 (ix86_rewrite_tls_address): Ditto.
9434 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9435 (ix86_rewrite_tls_address_1): Ditto.
9436 (ix86_rewrite_tls_address): Ditto.
9437 * config/i386/predicates.md (tls_address_pattern): New predicate.
9438 * config/i386/i386.md (TLS address splitter): New splitter.
9440 2017-09-04 Richard Biener <rguenther@suse.de>
9442 PR tree-optimization/82084
9443 * fold-const.h (can_native_encode_string_p): Declare.
9444 * fold-const.c (can_native_encode_string_p): Factor out from ...
9445 (native_encode_string): ... here.
9446 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9447 vectorizing stores from constants we later cannot handle.
9449 2017-09-04 Marek Polacek <polacek@redhat.com>
9452 * doc/invoke.texi: Update -Wtautological-compare documentation.
9454 2017-09-04 Jeff Law <law@redhat.com>
9456 PR tree-optimization/64910
9457 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9458 swap the first and last operand if the last is a constant.
9460 2017-09-04 Marek Polacek <polacek@redhat.com>
9463 * convert.c (do_narrow): When sanitizing signed integer overflows,
9464 bail out for signed types.
9465 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9467 2017-09-04 Richard Biener <rguenther@suse.de>
9469 PR tree-optimization/82060
9470 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9471 Move devirtualization after stmt folding and before EH/AB/noreturn
9472 cleanup to get the stmt refs canonicalized. Use a bool instead
9473 of gimple_modified_p since that doesn't work for NOPs. Schedule
9474 NOPs generated by folding for removal.
9476 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9477 Alan Hayward <alan.hayward@arm.com>
9478 David Sherwood <david.sherwood@arm.com>
9480 * coretypes.h (pad_direction): New enum.
9481 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9482 (FUNCTION_ARG_PADDING): Likewise.
9483 * target.def (function_arg_padding): New hook.
9484 * targhooks.h (default_function_arg_padding): Declare.
9485 * targhooks.c (default_function_arg_padding): New function.
9486 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9487 (TARGET_FUNCTION_ARG_PADDING): ...this.
9488 * doc/tm.texi: Regenerate.
9489 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9490 instead of direction.
9491 (compute_argument_addresses): Likewise.
9492 (load_register_parameters): Likewise.
9493 (emit_library_call_value_1): Likewise.
9494 (store_one_arg): Use targetm.calls.function_arg_padding instead
9495 of FUNCTION_ARG_PADDING.
9496 (must_pass_in_stack_var_size_or_pad): Likewise.
9497 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9498 (emit_group_store): Likewise.
9499 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9500 instead of FUNCTION_ARG_PADDING.
9501 (emit_push_insn): Likewise, and propagate enum change throughout
9503 * function.h (direction): Delete.
9504 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9506 * function.c (assign_parm_find_stack_rtl): Likewise.
9507 (assign_parm_setup_block_p): Likewise.
9508 (assign_parm_setup_block): Likewise.
9509 (gimplify_parameters): Likewise.
9510 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9511 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9513 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9514 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9515 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9516 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9517 (aarch64_function_arg_padding): ...this new function.
9518 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9519 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9520 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9521 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9522 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9523 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9524 (arm_pad_arg_upward): Replace with...
9525 (arm_function_arg_padding): ...this new function.
9526 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9528 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9529 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9530 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9531 (ia64_hpux_function_arg_padding): Replace with...
9532 (ia64_function_arg_padding): ...this new function. Use pad_direction
9533 instead of direction. Check for TARGET_HPUX.
9534 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9535 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9536 (iq2000_function_arg_padding): New function.
9537 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9538 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9539 (mips_function_arg_padding): ...this new function.
9540 (mips_pad_reg_upward): Update accordingly.
9541 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9542 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9543 targetm.calls.function_arg_padding.
9544 (FUNCTION_ARG_PADDING): Delete.
9545 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9546 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9547 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9548 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9549 (nios2_block_reg_padding): Return pad_direction instead of direction.
9550 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9551 instead of direction.
9552 (nios2_function_arg_padding): Likewise. Make static.
9553 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9554 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9555 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9556 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9557 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9558 (pa_function_arg_padding): Make static. Return pad_direction instead
9560 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9561 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9562 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9563 instead of direction. Use targetm.calls.function_arg_padding.
9564 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9565 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9566 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9567 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9568 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9570 (function_arg_padding): Rename to...
9571 (rs6000_function_arg_padding): ...this. Make static. Return
9572 pad_direction instead of direction.
9573 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9574 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9575 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9576 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9577 instead of direction. Use targetm.calls.function_arg_padding.
9578 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9579 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9580 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9581 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9582 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9583 (function_arg_padding): Rename to...
9584 (rs6000_function_arg_padding): ...this. Make static. Return
9585 pad_direction instead of direction.
9586 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9587 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9588 * config/s390/s390.c (s390_function_arg_padding): New function.
9589 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9590 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9591 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9592 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9593 (function_arg_padding): Rename to...
9594 (sparc_function_arg_padding): ...this. Make static. Return
9595 pad_direction instead of direction.
9596 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9597 * config/spu/spu.c (spu_function_arg_padding): New function.
9598 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9599 * system.h (FUNCTION_ARG_PADDING): Poison.
9601 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9602 Alan Hayward <alan.hayward@arm.com>
9603 David Sherwood <david.sherwood@arm.com>
9605 * target.def (modes_tieable_p): New hook.
9606 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9607 (TARGET_MODES_TIEABLE_P): ...this.
9608 * doc/tm.texi.in: Regenerate.
9609 * hooks.h (hook_bool_mode_mode_true): Declare.
9610 * hooks.c (hook_bool_mode_mode_true): New function.
9611 * combine.c (subst): Use targetm.modes_tieable_p instead of
9613 * dse.c (find_shift_sequence): Likewise.
9614 * expmed.c (extract_low_bits): Likewise.
9615 * lower-subreg.c: Include target.h.
9616 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9618 * rtlanal.c (rtx_cost): Likewise.
9619 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9620 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9621 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9622 (TARGET_MODES_TIEABLE_P): Redefine.
9623 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9624 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9625 (TARGET_MODES_TIEABLE_P): Redefine.
9626 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9627 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9628 (arc_modes_tieable_p): New function.
9629 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9630 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9631 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9632 (arm_modes_tieable_p): Make static.
9633 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9634 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9635 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9636 (TARGET_MODES_TIEABLE_P): Redefine.
9637 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9638 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9639 (TARGET_MODES_TIEABLE_P): Redefine.
9640 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9641 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9642 (cr16_modes_tieable_p): New function.
9643 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9644 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9645 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9646 (TRULY_NOOP_TRUNCATION): Update comment.
9647 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9648 (TRULY_NOOP_TRUNCATION): Update comment.
9649 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9650 (frv_modes_tieable_p): New function.
9651 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9652 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9653 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9654 (TARGET_MODES_TIEABLE_P): Redefine.
9655 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9656 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9657 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9658 (TARGET_MODES_TIEABLE_P): Redefine.
9659 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9660 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9661 (ia64_modes_tieable_p): New function.
9662 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9663 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9664 (iq2000_modes_tieable_p): New function.
9665 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9666 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9667 (lm32_modes_tieable_p): New function.
9668 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9669 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9670 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9671 (TARGET_MODES_TIEABLE_P): Redefine.
9672 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9673 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9674 (m32r_modes_tieable_p): New function.
9675 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9676 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9677 (m68k_modes_tieable_p): New function.
9678 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9679 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9680 (mcore_modes_tieable_p): New function.
9681 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9682 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9684 (TARGET_MODES_TIEABLE_P): Redefine.
9685 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9686 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9687 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9688 (TARGET_MODES_TIEABLE_P): Redefine.
9689 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9690 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9691 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9692 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9693 (mn10300_modes_tieable_p): ...this and make static.
9694 (TARGET_MODES_TIEABLE_P): Redefine.
9695 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9696 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9697 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9698 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9699 (msp430_modes_tieable_p): Make static.
9700 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9701 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9702 (TARGET_MODES_TIEABLE_P): Redefine.
9703 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9704 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9705 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9706 (TARGET_MODES_TIEABLE_P): Redefine.
9707 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9708 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9709 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9710 (TARGET_MODES_TIEABLE_P): Redefine.
9711 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9712 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9713 (pdp11_modes_tieable_p): New function.
9714 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9715 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9716 (rs6000_modes_tieable_p): New function.
9717 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9718 * config/powerpcspe/powerpcspe.md: Update comment.
9719 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9720 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9721 (TARGET_MODES_TIEABLE_P): Redefine.
9722 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9723 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9724 (rl78_modes_tieable_p): New function.
9725 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9726 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9727 (rs6000_modes_tieable_p): New function.
9728 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9729 * config/rs6000/rs6000.md: Update comment.
9730 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9731 * config/rx/rx.c (rx_modes_tieable_p): New function.
9732 (TARGET_MODES_TIEABLE_P): Redefine.
9733 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9734 * config/s390/s390.c (s390_modes_tieable_p): New function.
9735 (TARGET_MODES_TIEABLE_P): Redefine.
9736 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9737 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9738 (sh_modes_tieable_p): New function.
9739 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9740 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9741 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9742 (sparc_modes_tieable_p): Make static.
9743 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9744 * config/spu/spu.c (spu_modes_tieable_p): New function.
9745 (TARGET_MODES_TIEABLE_P): Redefine.
9746 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9747 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9748 (TARGET_MODES_TIEABLE_P): Redefine.
9749 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9750 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9751 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9752 * config/v850/v850.c (v850_modes_tieable_p): New function.
9753 (TARGET_MODES_TIEABLE_P): Redefine.
9754 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9755 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9756 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9757 (visium_modes_tieable_p): New function.
9758 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9759 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9760 (xtensa_modes_tieable_p): New function.
9761 * system.h (MODES_TIEABLE_P): Poison.
9763 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9764 Alan Hayward <alan.hayward@arm.com>
9765 David Sherwood <david.sherwood@arm.com>
9767 * target.def (hard_regno_mode_ok): New hook.
9768 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9769 (TARGET_HARD_REGNO_MODE_OK): ...this.
9770 * doc/tm.texi.in: Regenerate.
9771 * hooks.h (hook_bool_uint_mode_true): Declare.
9772 * hooks.c (hook_bool_uint_mode_true): New function.
9773 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9775 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9776 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9777 instead of HARD_REGNO_MODE_OK.
9778 * caller-save.c: Include target.h.
9779 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9781 * combine.c (can_combine_p): Likewise.
9782 (combinable_i3pat): Likewise.
9783 (can_change_dest_mode): Likewise.
9784 * expr.c (init_expr_target): Likewise.
9785 (convert_move): Likewise.
9786 (convert_modes): Likewise.
9787 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9788 (setup_prohibited_mode_move_regs): Likewise.
9789 * ira.h (target_ira): Likewise.
9790 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9791 * lra-constraints.c (process_alt_operands): Likewise.
9792 (split_reg): Likewise.
9793 * recog.c (peep2_find_free_register): Likewise.
9794 * ree.c (combine_reaching_defs): Likewise.
9795 * regcprop.c (maybe_mode_change): Likewise.
9796 * reginfo.c (init_reg_sets_1): Likewise.
9797 (choose_hard_reg_mode): Likewise.
9798 (simplifiable_subregs): Likewise.
9799 * regrename.c (check_new_reg_p): Likewise.
9800 * reload.c (find_valid_class): Likewise.
9801 (find_valid_class_1): Likewise.
9802 (reload_inner_reg_of_subreg): Likewise.
9803 (push_reload): Likewise.
9804 (combine_reloads): Likewise.
9805 (find_dummy_reload): Likewise.
9806 (find_reloads): Likewise.
9807 * reload1.c (find_reg): Likewise.
9808 (set_reload_reg): Likewise.
9809 (allocate_reload_reg): Likewise.
9810 (choose_reload_regs): Likewise.
9811 (reload_adjust_reg_for_temp): Likewise.
9812 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9813 (simplify_subreg_regno): Likewise.
9814 * sel-sched.c (init_regs_for_mode): Likewise.
9815 * varasm.c (make_decl_rtl): Likewise.
9816 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9817 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9819 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9820 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9821 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9822 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9823 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9825 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9826 (arc_mode_class): Delete.
9827 (HARD_REGNO_MODE_OK): Delete.
9828 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9829 (arc_hard_regno_mode_ok): Rename old array to...
9830 (arc_hard_regno_mode_ok_modes): ...this.
9831 (arc_conditional_register_usage): Update accordingly.
9832 (arc_mode_class): Make static.
9833 (arc_hard_regno_mode_ok): New function.
9834 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9835 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9836 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9837 (arm_hard_regno_mode_ok): Make static.
9838 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9840 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9841 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9842 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9845 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9846 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9847 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9848 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9850 * config/bfin/predicates.md (valid_reg_operand): Use
9851 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9852 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9853 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9854 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9855 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9856 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9857 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9858 (cr16_hard_regno_mode_ok): Make static and return a bool.
9859 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9860 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9861 (cris_hard_regno_mode_ok): New function.
9862 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9863 (epiphany_mode_class): Delete.
9864 (HARD_REGNO_MODE_OK): Delete.
9865 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9866 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9867 (hard_regno_mode_ok): Rename to...
9868 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9869 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9870 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9872 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9873 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9874 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9875 (frv_hard_regno_mode_ok): Make static and return a bool.
9876 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9878 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9879 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9880 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9881 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9884 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9885 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9886 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9889 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9890 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9891 (ia64_hard_regno_mode_ok): New function.
9892 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9893 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9894 (iq2000_hard_regno_mode_ok): New function.
9895 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9896 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9897 (lm32_hard_regno_mode_ok): New function.
9898 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9899 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9900 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9901 instead of HARD_REGNO_MODE_OK.
9902 (m32c_hard_regno_ok): Rename to...
9903 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9904 (m32c_cannot_change_mode_class): Update accordingly.
9905 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9906 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9907 (m32r_mode_class): Delete.
9908 (HARD_REGNO_MODE_OK): Delete.
9909 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9910 (m32r_hard_regno_mode_ok): Rename to...
9911 (m32r_hard_regno_modes): ...this.
9912 (m32r_mode_class): Make static.
9913 (m32r_hard_regno_mode_ok): New function.
9914 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9915 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9916 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9917 (m68k_hard_regno_mode_ok): Make static.
9918 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9919 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9920 (mcore_hard_regno_mode_ok): New function.
9921 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9922 (HARD_REGNO_MODE_OK): Delete.
9923 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9925 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9926 (microblaze_hard_regno_mode_ok): New function.
9927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9928 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9929 (mips_hard_regno_mode_ok): Delete.
9930 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9931 (mips_hard_regno_mode_ok_p): ...this and make static.
9932 (mips_hard_regno_mode_ok_p): Rename to...
9933 (mips_hard_regno_mode_ok_uncached): ...this.
9934 (mips_hard_regno_mode_ok): New function.
9935 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9936 of HARD_REGNO_MODE_OK.
9937 (mips_option_override): Update after above name changes.
9938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9939 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9940 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9941 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9942 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9943 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9944 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9945 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9946 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9947 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9948 (msp430_hard_regno_mode_ok): Make static and return a bool.
9949 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9950 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9951 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9953 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9954 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9955 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9956 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9957 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9958 (PA_HARD_REGNO_MODE_OK): ...this
9959 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9960 (PA_HARD_REGNO_MODE_OK): ...this.
9961 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9962 (pa_hard_regno_mode_ok): New function.
9963 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9964 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9965 (pdp11_hard_regno_mode_ok): New function.
9966 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9967 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9969 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9972 (rs6000_hard_regno_mode_ok): Rename to...
9973 (rs6000_hard_regno_mode_ok_uncached): ...this.
9974 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9975 (rs6000_hard_regno_mode_ok): New function.
9976 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9977 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9978 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9979 (riscv_hard_regno_mode_ok): ...this and make static.
9980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9981 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9982 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9983 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9984 (rl78_hard_regno_mode_ok): Make static and return bool.
9985 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9986 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9988 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9989 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9990 (rs6000_hard_regno_mode_ok): Rename to...
9991 (rs6000_hard_regno_mode_ok_uncached): ...this.
9992 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9993 (rs6000_hard_regno_mode_ok): New function.
9994 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9995 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9996 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9997 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9998 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9999 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10001 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10002 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10003 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10004 (sh_hard_regno_mode_ok): Make static.
10005 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10006 instead of HARD_REGNO_MODE_OK.
10007 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10008 (sparc_mode_class): Delete.
10009 (HARD_REGNO_MODE_OK): Delete.
10010 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10011 (hard_regno_mode_classes): Make static.
10012 (sparc_mode_class): Likewise.
10013 (sparc_hard_regno_mode_ok): New function.
10014 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10015 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10016 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10019 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10020 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10021 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10022 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10023 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10024 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10025 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10026 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10027 (visium_hard_regno_mode_ok): New function.
10028 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10029 instead of HARD_REGNO_MODE_OK.
10030 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10031 (HARD_REGNO_MODE_OK): Delete.
10032 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10033 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10034 (xtensa_option_override): Update accordingly.
10035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10036 (xtensa_hard_regno_mode_ok): New function.
10037 * system.h (HARD_REGNO_MODE_OK): Poison.
10039 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10040 Alan Hayward <alan.hayward@arm.com>
10041 David Sherwood <david.sherwood@arm.com>
10043 * target.def (hard_regno_call_part_clobbered): New hook.
10044 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10045 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10046 * doc/tm.texi: Regenerate.
10047 * hooks.h (hook_bool_uint_mode_false): Declare.
10048 * hooks.c (hook_bool_uint_mode_false): New function.
10049 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10050 * cselib.c (cselib_process_insn): Use
10051 targetm.hard_regno_call_part_clobbered instead of
10052 HARD_REGNO_CALL_PART_CLOBBERED.
10053 * ira-conflicts.c (ira_build_conflicts): Likewise.
10054 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10055 * lra-constraints.c (need_for_call_save_p): Likewise.
10056 * lra-lives.c: Include target.h.
10057 (check_pseudos_live_through_calls): Use
10058 targetm.hard_regno_call_part_clobbered instead of
10059 HARD_REGNO_CALL_PART_CLOBBERED.
10060 * regcprop.c: Include target.h.
10061 (copyprop_hardreg_forward_1): Use
10062 targetm.hard_regno_call_part_clobbered instead of
10063 HARD_REGNO_CALL_PART_CLOBBERED.
10064 * reginfo.c (choose_hard_reg_mode): Likewise.
10065 * regrename.c (check_new_reg_p): Likewise.
10066 * reload.c (find_equiv_reg): Likewise.
10067 * reload1.c (emit_reload_insns): Likewise.
10068 * sched-deps.c (deps_analyze_insn): Likewise.
10069 * sel-sched.c (init_regs_for_mode): Likewise.
10070 (mark_unavailable_hard_regs): Likewise.
10071 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10072 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10073 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10075 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10076 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10077 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10079 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10081 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10082 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10083 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10085 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10086 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10087 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10089 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10090 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10092 * config/powerpcspe/powerpcspe.c
10093 (rs6000_hard_regno_call_part_clobbered): New function.
10094 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10095 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10096 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10098 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10099 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10100 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10102 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10103 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10104 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10106 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10107 Alan Hayward <alan.hayward@arm.com>
10108 David Sherwood <david.sherwood@arm.com>
10110 * rtl.h (subreg_memory_offset): Declare.
10111 * emit-rtl.c (subreg_memory_offset): New function.
10112 * expmed.c (store_bit_field_1): Use it.
10113 * expr.c (undefined_operand_subword_p): Likewise.
10114 * simplify-rtx.c (simplify_subreg): Likewise.
10116 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10118 PR rtl-optimization/57448
10121 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10122 using atomic_load pattern.
10123 (expand_atomic_store): Likewise.
10125 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10128 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10129 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10130 handling. Use replace_call_with_value with NULL instead of
10131 gsi_replace, unlink_stmt_vdef and release_defs.
10133 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10136 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10138 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10141 * rtl.h (emit_library_call_value_1): Declare.
10142 (emit_library_call): Replace declaration with a series of overloads.
10143 Remove the parameter count argument.
10144 (emit_library_call_value): Likewise.
10145 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10146 with an "rtx_mode_t *".
10147 (emit_library_call_value): Delete.
10148 (emit_library_call): Likewise.
10149 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10150 (asan_emit_allocas_unpoison): Likewise.
10151 * builtins.c (expand_builtin_powi): Likewise.
10152 (expand_asan_emit_allocas_unpoison): Likewise.
10153 * cfgexpand.c (expand_main_function): Likewise.
10154 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10155 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10156 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10157 * config/arm/arm.c (arm_trampoline_init): Likewise.
10158 (arm_call_tls_get_addr): Likewise.
10159 (arm_expand_divmod_libfunc): Likewise.
10160 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10161 (smulsi3_highpart): Likewise.
10162 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10163 (c6x_expand_compare): Likewise.
10164 (c6x_expand_movmem): Likewise.
10165 * config/frv/frv.c (frv_trampoline_init): Likewise.
10166 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10167 (ix86_expand_divmod_libfunc): Likewise.
10168 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10169 (ia64_expand_compare): Likewise.
10170 (ia64_profile_hook): Likewise.
10171 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10172 (nonlocal_goto): Likewise.
10173 (restore_stack_nonlocal): Likewise.
10174 * config/m32r/m32r.c (block_move_call): Likewise.
10175 (m32r_trampoline_init): Likewise.
10176 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10177 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10178 (m68k_call_m68k_read_tp): Likewise.
10179 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10180 (microblaze_expand_divide): Likewise.
10181 * config/mips/mips.h (mips_args): Likewise.
10182 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10183 (MIPS_ICACHE_SYNC): Likewise.
10184 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10185 (nios2_trampoline_init): Likewise.
10186 * config/pa/pa.c (hppa_tls_call): Likewise.
10187 (pa_trampoline_init): Likewise.
10188 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10189 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10190 (expand_strn_compare): Likewise.
10191 (rs6000_generate_compare): Likewise.
10192 (rs6000_expand_float128_convert): Likewise.
10193 (output_profile_hook): Likewise.
10194 (rs6000_trampoline_init): Likewise.
10195 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10196 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10197 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10198 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10199 (rs6000_generate_compare): Likewise.
10200 (rs6000_expand_float128_convert): Likewise.
10201 (output_profile_hook): Likewise.
10202 (rs6000_trampoline_init): Likewise.
10203 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10204 * config/sh/sh.c (sh_trampoline_init): Likewise.
10205 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10206 (sparc_emit_float_lib_cmp): Likewise.
10207 (sparc32_initialize_trampoline): Likewise.
10208 (sparc64_initialize_trampoline): Likewise.
10209 (sparc_profile_hook): Likewise.
10210 * config/spu/spu.c (ea_load_store): Likewise.
10211 * config/spu/spu.md (floatunssidf2): Likewise.
10212 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10213 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10214 * config/visium/visium.c (expand_block_move_4): Likewise.
10215 (expand_block_move_2): Likewise.
10216 (expand_block_move_1): Likewise.
10217 (expand_block_set_4): Likewise.
10218 (expand_block_set_2): Likewise.
10219 (expand_block_set_1): Likewise.
10220 (visium_trampoline_init): Likewise.
10221 (visium_profile_hook): Likewise.
10222 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10223 (xtensa_setup_frame_addresses): Likewise.
10224 (xtensa_trampoline_init): Likewise.
10225 * except.c (sjlj_emit_function_enter): Likewise.
10226 (sjlj_emit_function_exit): Likewise.
10227 * explow.c (allocate_dynamic_stack_space): Likewise.
10228 (probe_stack_range): Likewise.
10229 * expr.c (convert_mode_scalar): Likewise.
10230 * optabs.c (expand_binop): Likewise.
10231 (expand_twoval_binop_libfunc): Likewise.
10232 (expand_unop): Likewise.
10233 (prepare_cmp_insn): Likewise.
10234 (prepare_float_lib_cmp): Likewise.
10235 (expand_float): Likewise.
10236 (expand_fix): Likewise.
10237 (expand_fixed_convert): Likewise.
10238 (maybe_emit_sync_lock_test_and_set): Likewise.
10239 (expand_atomic_compare_and_swap): Likewise.
10240 (expand_mem_thread_fence): Likewise.
10241 (expand_atomic_fetch_op): Likewise.
10243 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10245 * doc/generic.texi (OpenACC): Adjust URL.
10246 * doc/invoke.texi (C Dialect Options): Ditto.
10248 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10250 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10251 predicate for operand 1. Add (m,<S>) constraint.
10252 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10253 Prevent memory operand 1 with register operand 2.
10255 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10257 PR rtl-optimization/82024
10258 * combine.c (try_combine): If the combination result is a PARALLEL,
10259 and we only need to retain the SET in there that would be placed
10260 at I2, check that we can place that at I3 instead, before doing so.
10262 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10265 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10267 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10268 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10271 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10272 Jeff Law <law@redhat.com>
10274 * varasm.c (bss_initializer_p): Do not put constants into .bss
10275 (categorize_decl_for_section): Handle bss_initializer_p returning
10276 false when DECL_INITIAL is NULL.
10278 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10281 * config/s390/s390.c (s390_can_inline_p): New function.
10283 2017-09-01 Jeff Law <law@redhat.com>
10285 PR tree-optimization/82052
10286 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10287 Always initialize the returned slot after a hash table miss
10288 when INSERT is true.
10290 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10292 * config/s390/s390.md (mem_signal_fence): Remove.
10293 * doc/md.texi (mem_signal_fence): Remove.
10294 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10296 * target-insns.def (mem_signal_fence): Remove.
10298 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10301 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10304 * asan.c (create_odr_indicator): Strip name encoding from assembler
10305 name before appending it after __odr_asan_.
10307 2017-09-01 Martin Liska <mliska@suse.cz>
10309 PR tree-optimization/82059
10310 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10311 frequency only when an edge is redirected.
10313 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10315 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10316 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10317 (arc_conditional_register_usage): Remove ARC600 lp_count
10319 (arc_file_start): Emit Tag_ARC_CPU_variation.
10320 (arc_can_use_doloop_p): New conditions to use ZOLs.
10321 (hwloop_fail): New function.
10322 (hwloop_optimize): Likewise.
10323 (hwloop_pattern_reg): Likewise.
10324 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10325 (arc_reorg_loops): New function, calls reorg_loops.
10326 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10327 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10329 (arc_loop_hazard): Remove function, functionality moved into
10331 (arc_hazard): Remove arc_loop_hazard call.
10332 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10333 into hwloop_optimize.
10334 (arc_label_align): Remove ZOL handling.
10335 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10336 * config/arc/arc.md (doloop_begin): Remove pattern.
10337 (doloop_begin_i): Likewise.
10338 (doloop_end_i): Likewise.
10339 (doloop_fallback): Likewise.
10340 (doloop_fallback_m): Likewise.
10341 (doloop_end): Reimplement expand.
10342 (arc_lp): New pattern for LP instruction.
10343 (loop_end): New pattern.
10344 (loop_fail): Likewise.
10345 (decrement_and_branch_until_zero): Likewise.
10346 * config/arc/arc.opt (mlpc-width): New option.
10347 * doc/invoke.texi (mlpc-width): Document option.
10349 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10351 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10352 (arc_ccfsm_advance): Fix checking for delay slots.
10353 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10355 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10357 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10359 (movhi_insn): Update store instruction constraint which are saving
10360 6-bit short immediates.
10361 (movsi_insn): Consider also short scaled load operations.
10362 (zero_extendhisi2_i): Use Usd constraint instead of T.
10363 (extendhisi2_i): Add q constraint.
10364 (arc_clzsi2): Add type and length attributes.
10365 (arc_ctzsi2): Likewise.
10366 * config/arc/constraints.md (Usc): Update constraint, the
10367 assembler can parse two relocations for a single instruction.
10369 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10371 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10372 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10374 2017-08-31 Olivier Hainque <hainque@adacore.com>
10376 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10377 match as powerpc-wrs-vxworks*.
10379 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10381 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10382 register constraint for by-element operand.
10383 (aarch64_mls_elt_merge<mode>): Likewise.
10385 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10387 * config/arc/arc.c (arc_can_follow_jump): Check for short
10390 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10392 * config.gcc: Use g.opt for arc.
10393 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10394 functionality moved to ...
10395 (legitimate_scaled_address_p): New function, ...here.
10396 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10397 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10398 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10400 (arc_override_options): Handle G option.
10401 (arc_output_pic_addr_const): Correct function definition.
10402 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10403 (arc_decl_anon_ns_mem_p): Delete.
10404 (arc_in_small_data_p): Overhaul this function to take into
10405 consideration the value given via G option.
10406 (arc_rewrite_small_data_1): Renamed and corrected old
10407 arc_rewrite_small_data function.
10408 (arc_rewrite_small_data): New function.
10409 (small_data_pattern): Don't use pic_offset_table_rtx.
10410 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10411 * config/arc/simdext.md (movmisalignv2hi): Use
10412 prepare_move_operands function.
10414 (movmisalign*): Likewise.
10415 * doc/invoke.texi (ARC options): Document -G option.
10417 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10419 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10421 * config/arc/arc.c (arc_print_operand): Output scalled address for
10422 sdata whenever is possible.
10423 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10424 load/stores are available.
10425 (compact_sda_memory_operand): Check for the alignment required by
10426 code density instructions.
10427 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10429 * config/arc/constraints.md (Usd): Update constraint.
10430 (Us0): New constraint.
10431 (Usc): Update constraint.
10433 2017-08-31 Richard Biener <rguenther@suse.de>
10435 PR middle-end/82054
10436 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10437 function only once.
10439 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10441 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10442 Resize type_signature.
10444 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10445 Alan Hayward <alan.hayward@arm.com>
10446 David Sherwood <david.sherwood@arm.com>
10448 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10449 subregs whose inner modes can be stored in GPRs.
10450 (aarch64_classify_index): Likewise.
10452 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayward <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10457 (V_INT_EQUIV): ...this.
10458 (v_cmp_result): Rename to...
10459 (v_int_equiv): ...this.
10460 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10461 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10462 (copysign<mode>3): Likewise.
10463 (aarch64_simd_bsl<mode>_internal): Likewise.
10464 (aarch64_simd_bsl<mode>): Likewise.
10465 (vec_cmp<mode><mode>): Likewise.
10466 (vcond<mode><mode>): Likewise.
10467 (vcond<v_cmp_mixed><mode>): Likewise.
10468 (vcondu<mode><v_cmp_mixed>): Likewise.
10469 (aarch64_cm<optab><mode>): Likewise.
10470 (aarch64_cmtst<mode>): Likewise.
10471 (aarch64_fac<optab><mode>): Likewise.
10472 (vec_perm_const<mode>): Likewise.
10473 (vcond_mask_<mode><v_cmp_result>): Rename to...
10474 (vcond_mask_<mode><v_int_equiv>): ...this.
10475 (vec_cmp<mode><v_cmp_result>): Rename to...
10476 (vec_cmp<mode><v_int_equiv>): ...this.
10478 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10479 Alan Hayward <alan.hayward@arm.com>
10480 David Sherwood <david.sherwood@arm.com>
10482 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10484 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10485 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10486 (UNSPEC_LD4_DREG): New unspecs.
10487 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10488 (aarch64_ld2<mode>_dreg_be): Replace with...
10489 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10491 (aarch64_ld3<mode>_dreg_le)
10492 (aarch64_ld3<mode>_dreg_be): Replace with...
10493 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10495 (aarch64_ld4<mode>_dreg_le)
10496 (aarch64_ld4<mode>_dreg_be): Replace with...
10497 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10500 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10502 PR tree-optimization/81987
10503 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10504 insert an initializer in a location not dominated by the stride
10507 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10509 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10510 on the entire header of the finally block in the fallthru case.
10512 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10514 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10516 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10518 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10519 rs6000_emit_move_from_cr and call renamed function.
10520 (rs6000_emit_prologue): Call renamed functions.
10521 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10522 movesi_from_cr, remove volatile CRs.
10524 2017-08-30 Jon Beniston <jon@beniston.com>
10525 Richard Biener <rguenther@suse.de>
10527 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10528 of VECTOR_MODE_P check.
10529 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10530 element vector types.
10532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10534 * df.h (df_read_modify_subreg_p): Remove in favor of...
10535 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10536 const_rtx instead of an rtx.
10537 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10538 * df-problems.c (df_word_lr_mark_ref): Likewise.
10539 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10540 (mark_pseudo_reg_dead): Likewise.
10541 (mark_ref_dead): Likewise.
10542 * reginfo.c (init_subregs_of_mode): Likewise.
10543 * sched-deps.c (sched_analyze_1): Likewise.
10544 * df-scan.c (df_def_record_1): Likewise.
10545 (df_uses_record): Likewise.
10546 (df_read_modify_subreg_p): Remove in favor of...
10547 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10548 const_rtx instead of an rtx.
10550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10551 Alan Hayward <alan.hayward@arm.com>
10552 David Sherwood <david.sherwood@arm.com>
10554 * rtl.h (partial_subreg_p): New function.
10555 * caller-save.c (save_call_clobbered_regs): Use it.
10556 * calls.c (expand_call): Likewise.
10557 * combine.c (combinable_i3pat): Likewise.
10558 (simplify_set): Likewise.
10559 (make_extraction): Likewise.
10560 (make_compound_operation_int): Likewise.
10561 (gen_lowpart_or_truncate): Likewise.
10562 (force_to_mode): Likewise.
10563 (make_field_assignment): Likewise.
10564 (reg_truncated_to_mode): Likewise.
10565 (record_truncated_value): Likewise.
10566 (move_deaths): Likewise.
10567 * cse.c (record_jump_cond): Likewise.
10568 (cse_insn): Likewise.
10569 * cselib.c (cselib_lookup_1): Likewise.
10570 * expmed.c (extract_bit_field_using_extv): Likewise.
10571 * function.c (assign_parm_setup_reg): Likewise.
10572 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10573 * ira-build.c (create_insn_allocnos): Likewise.
10574 * lra-coalesce.c (merge_pseudos): Likewise.
10575 * lra-constraints.c (match_reload): Likewise.
10576 (simplify_operand_subreg): Likewise.
10577 (curr_insn_transform): Likewise.
10578 * lra-lives.c (process_bb_lives): Likewise.
10579 * lra.c (new_insn_reg): Likewise.
10580 (lra_substitute_pseudo): Likewise.
10581 * regcprop.c (mode_change_ok): Likewise.
10582 (maybe_mode_change): Likewise.
10583 (copyprop_hardreg_forward_1): Likewise.
10584 * reload.c (push_reload): Likewise.
10585 (find_reloads): Likewise.
10586 (find_reloads_subreg_address): Likewise.
10587 * reload1.c (alter_reg): Likewise.
10588 (eliminate_regs_1): Likewise.
10589 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10591 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10593 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10594 back to if statements, including unpack.
10596 2017-08-30 Martin Liska <mliska@suse.cz>
10598 PR inline-asm/82001
10599 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10601 (func_checker::compare_asm_inputs_outputs): ... this function.
10602 (func_checker::compare_gimple_asm): Use the function to compare
10603 also ASM constrains.
10604 * ipa-icf-gimple.h: Rename the function.
10606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10607 Alan Hayward <alan.hayward@arm.com>
10608 David Sherwood <david.sherwood@arm.com>
10610 * coretypes.h (complex_mode): New type.
10611 * gdbhooks.py (build_pretty_printer): Handle it.
10612 * machmode.h (complex_mode): New class.
10613 (complex_mode::includes_p): New function.
10614 (is_complex_int_mode): Likewise.
10615 (is_complex_float_mode): Likewise.
10616 * genmodes.c (get_mode_class): Handle complex mode classes.
10617 * function.c (expand_function_end): Use is_complex_int_mode.
10619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10620 Alan Hayward <alan.hayward@arm.com>
10621 David Sherwood <david.sherwood@arm.com>
10623 * coretypes.h (scalar_mode_pod): New typedef.
10624 * gdbhooks.py (build_pretty_printer): Handle it.
10625 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10626 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10627 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10628 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10629 as_a <scalar_mode>.
10631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10632 Alan Hayward <alan.hayward@arm.com>
10633 David Sherwood <david.sherwood@arm.com>
10635 * machmode.h (mode_for_vector): Take a scalar_mode instead
10637 * stor-layout.c (mode_for_vector): Likewise.
10638 * explow.c (promote_mode): Use as_a <scalar_mode>.
10639 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10642 Alan Hayward <alan.hayward@arm.com>
10643 David Sherwood <david.sherwood@arm.com>
10645 * target.def (preferred_simd_mode): Take a scalar_mode
10646 instead of a machine_mode.
10647 * targhooks.h (default_preferred_simd_mode): Likewise.
10648 * targhooks.c (default_preferred_simd_mode): Likewise.
10649 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10650 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10651 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10652 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10653 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10654 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10655 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10656 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10658 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10659 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10660 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10661 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10662 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10663 * doc/tm.texi: Regenerate.
10664 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10668 Alan Hayward <alan.hayward@arm.com>
10669 David Sherwood <david.sherwood@arm.com>
10671 * target.def (scalar_mode_supported_p): Take a scalar_mode
10672 instead of a machine_mode.
10673 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10674 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10675 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10676 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10677 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10678 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10679 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10680 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10681 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10682 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10683 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10684 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10685 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10686 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10688 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10689 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10690 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10691 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10692 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10694 * doc/tm.texi: Regenerate.
10696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10697 Alan Hayward <alan.hayward@arm.com>
10698 David Sherwood <david.sherwood@arm.com>
10700 * coretypes.h (opt_scalar_mode): New typedef.
10701 * gdbhooks.py (build_pretty_printers): Handle it.
10702 * machmode.h (mode_iterator::get_2xwider): Add overload for
10704 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10706 * expr.c (convert_mode_scalar): Likewise.
10707 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10708 * optabs.c (expand_float): Likewise.
10709 (expand_fix): Likewise.
10710 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10713 Alan Hayward <alan.hayward@arm.com>
10714 David Sherwood <david.sherwood@arm.com>
10716 * optabs.c (expand_float): Explicitly check for scalars before
10717 using a branching expansion.
10718 (expand_fix): Likewise.
10720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10721 Alan Hayward <alan.hayward@arm.com>
10722 David Sherwood <david.sherwood@arm.com>
10724 * expr.c (convert_mode): Split scalar handling out into...
10725 (convert_mode_scalar): ...this new function. Treat the modes
10728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10729 Alan Hayward <alan.hayward@arm.com>
10730 David Sherwood <david.sherwood@arm.com>
10732 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10734 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10737 Alan Hayward <alan.hayward@arm.com>
10738 David Sherwood <david.sherwood@arm.com>
10740 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10741 rather than a machine_mode.
10742 (fixed_from_string): Likewise.
10743 (fixed_convert): Likewise.
10744 (fixed_convert_from_int): Likewise.
10745 (fixed_convert_from_real): Likewise.
10746 (real_convert_from_fixed): Likewise.
10747 * fixed-value.c (fixed_from_double_int): Likewise.
10748 (fixed_from_string): Likewise.
10749 (fixed_convert): Likewise.
10750 (fixed_convert_from_int): Likewise.
10751 (fixed_convert_from_real): Likewise.
10752 (real_convert_from_fixed): Likewise.
10753 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10760 of separate mode class checks. Do not allow vector modes here.
10761 (immed_wide_int_const): Use as_a <scalar_mode>.
10762 * explow.c (trunc_int_for_mode): Likewise.
10763 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10764 (wi::shwi): Likewise.
10765 (wi::min_value): Likewise.
10766 (wi::max_value): Likewise.
10767 * dwarf2out.c (loc_descriptor): Likewise.
10768 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10769 for CONST_WIDE_INT.
10771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10772 Alan Hayward <alan.hayward@arm.com>
10773 David Sherwood <david.sherwood@arm.com>
10775 * tree.h (SCALAR_TYPE_MODE): New macro.
10776 * expr.c (expand_expr_addr_expr_1): Use it.
10777 (expand_expr_real_2): Likewise.
10778 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10779 (fold_convert_const_fixed_from_int): Likewise.
10780 (fold_convert_const_fixed_from_real): Likewise.
10781 (native_encode_fixed): Likewise
10782 (native_encode_complex): Likewise
10783 (native_encode_vector): Likewise.
10784 (native_interpret_fixed): Likewise.
10785 (native_interpret_real): Likewise.
10786 (native_interpret_complex): Likewise.
10787 (native_interpret_vector): Likewise.
10788 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10789 (simd_clone_adjust_argument_types): Likewise.
10790 (simd_clone_init_simd_arrays): Likewise.
10791 (simd_clone_adjust): Likewise.
10792 * stor-layout.c (layout_type): Likewise.
10793 * tree.c (build_minus_one_cst): Likewise.
10794 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10795 * tree-inline.c (estimate_move_cost): Likewise.
10796 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10797 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10798 (vectorizable_reduction): Likewise.
10799 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10800 (vect_recog_mixed_size_cond_pattern): Likewise.
10801 (check_bool_pattern): Likewise.
10802 (adjust_bool_pattern): Likewise.
10803 (search_type_for_mask_1): Likewise.
10804 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10805 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10806 (vectorizable_load): Likewise.
10807 (vectorizable_store): Likewise.
10808 * ubsan.c (ubsan_encode_value): Likewise.
10809 * varasm.c (output_constant): Likewise.
10811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10812 Alan Hayward <alan.hayward@arm.com>
10813 David Sherwood <david.sherwood@arm.com>
10815 * coretypes.h (scalar_mode): New class.
10816 * machmode.h (scalar_mode): Likewise.
10817 (scalar_mode::includes_p): New function.
10818 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10819 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10820 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10821 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10822 * expmed.c (store_bit_field_1): Likewise.
10823 (extract_bit_field_1): Likewise.
10824 * expr.c (write_complex_part): Likewise.
10825 (read_complex_part): Likewise.
10826 (emit_move_complex_push): Likewise.
10827 (expand_expr_real_2): Likewise.
10828 * function.c (assign_parm_setup_reg): Likewise.
10829 (assign_parms_unsplit_complex): Likewise.
10830 * optabs.c (expand_binop): Likewise.
10831 * rtlanal.c (subreg_get_info): Likewise.
10832 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10833 * varasm.c (output_constant_pool_2): Likewise.
10835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10836 Alan Hayward <alan.hayward@arm.com>
10837 David Sherwood <david.sherwood@arm.com>
10839 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10841 (expmed_mult_highpart_optab): Likewise.
10842 (expmed_mult_highpart): Likewise.
10844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10845 Alan Hayward <alan.hayward@arm.com>
10846 David Sherwood <david.sherwood@arm.com>
10848 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10849 instead of a machine_mode.
10850 (builtin_memset_read_str): Likewise.
10851 * builtins.c (c_readstr): Likewise.
10852 (builtin_memcpy_read_str): Likewise.
10853 (builtin_strncpy_read_str): Likewise.
10854 (builtin_memset_read_str): Likewise.
10855 (builtin_memset_gen_str): Likewise.
10856 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10857 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10858 instead of a machine_mode.
10859 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10861 (make_extraction): Likewise.
10862 (try_widen_shift_mode): Take and return scalar_int_modes instead
10864 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10865 a scalar_int_mode instead of a machine_mode.
10866 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10867 (avr_addr_space_pointer_mode): Likewise.
10868 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10869 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10870 (msp430_unwind_word_mode): Likewise.
10871 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10872 (spu_addr_space_pointer_mode): Likewise.
10873 (spu_addr_space_address_mode): Likewise.
10874 (spu_libgcc_cmp_return_mode): Likewise.
10875 (spu_libgcc_shift_count_mode): Likewise.
10876 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10877 (rl78_addr_space_pointer_mode): Likewise.
10878 (fl78_unwind_word_mode): Likewise.
10879 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10881 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10882 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10883 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10884 (mips_valid_pointer_mode): Likewise.
10885 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10886 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10887 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10889 (ft32_addr_space_address_mode): Likewise.
10890 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10891 scalar_int_mode instead of a machine_mode.
10892 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10894 (m32c_addr_space_address_mode): Likewise.
10895 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10896 (rs6000_eh_return_filter_mode): Likewise.
10897 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10898 (rs6000_eh_return_filter_mode): Likewise.
10899 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10900 (s390_libgcc_shift_count_mode): Likewise.
10901 (s390_unwind_word_mode): Likewise.
10902 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10904 * target.def (mode_rep_extended): Likewise.
10905 (valid_pointer_mode): Likewise.
10906 (addr_space.valid_pointer_mode): Likewise.
10907 (eh_return_filter_mode): Return a scalar_int_mode rather than
10909 (libgcc_cmp_return_mode): Likewise.
10910 (libgcc_shift_count_mode): Likewise.
10911 (unwind_word_mode): Likewise.
10912 (addr_space.pointer_mode): Likewise.
10913 (addr_space.address_mode): Likewise.
10914 * doc/tm.texi: Regenerate.
10915 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10917 (do_jump): Use scalar_int_mode for local variables.
10918 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10919 rather than a machine_mode.
10920 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10921 (scompare_loc_descriptor_wide): Likewise.
10922 (scompare_loc_descriptor_narrow): Likewise.
10923 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10925 * except.c (sjlj_emit_dispatch_table): Likewise.
10926 (expand_builtin_eh_copy_values): Likewise.
10927 * explow.c (convert_memory_address_addr_space_1): Likewise.
10928 Take a scalar_int_mode rather than a machine_mode.
10929 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10930 than a machine_mode.
10931 (memory_address_addr_space): Use scalar_int_mode for local variables.
10932 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10933 rather than a machine_mode.
10934 * expmed.c (mask_rtx): Likewise.
10935 (init_expmed_one_conv): Likewise.
10936 (expand_mult_highpart_adjust): Likewise.
10937 (extract_high_half): Likewise.
10938 (expmed_mult_highpart_optab): Likewise.
10939 (expmed_mult_highpart): Likewise.
10940 (expand_smod_pow2): Likewise.
10941 (expand_sdiv_pow2): Likewise.
10942 (emit_store_flag_int): Likewise.
10943 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10945 (extract_low_bits): Likewise.
10946 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10948 * expr.c (pieces_addr::adjust): Likewise.
10949 (can_store_by_pieces): Likewise.
10950 (store_by_pieces): Likewise.
10951 (clear_by_pieces_1): Likewise.
10952 (expand_expr_addr_expr_1): Likewise.
10953 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10954 (expand_expr_real_1): Likewise.
10955 (try_casesi): Likewise.
10956 * final.c (shorten_branches): Likewise.
10957 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10958 type of "mode" to machine_mode.
10959 * internal-fn.c (expand_arith_overflow_result_store): Take a
10960 scalar_int_mode rather than a machine_mode.
10961 (expand_mul_overflow): Use scalar_int_mode for local variables.
10962 * loop-doloop.c (doloop_modify): Likewise.
10963 (doloop_optimize): Likewise.
10964 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10965 than a machine_mode.
10966 (expand_doubleword_shift_condmove): Likewise.
10967 (expand_doubleword_shift): Likewise.
10968 (expand_doubleword_clz): Likewise.
10969 (expand_doubleword_popcount): Likewise.
10970 (expand_doubleword_parity): Likewise.
10971 (expand_absneg_bit): Use scalar_int_mode for local variables.
10972 (prepare_float_lib_cmp): Likewise.
10973 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10974 rather than a machine_mode.
10975 (convert_memory_address_addr_space): Likewise.
10976 (get_mode_bounds): Likewise.
10977 (get_address_mode): Return a scalar_int_mode rather than a
10979 * rtlanal.c (get_address_mode): Likewise.
10980 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10981 than a machine_mode.
10982 * targhooks.c (default_mode_rep_extended): Likewise.
10983 (default_valid_pointer_mode): Likewise.
10984 (default_addr_space_valid_pointer_mode): Likewise.
10985 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10986 than a machine_mode.
10987 (default_libgcc_cmp_return_mode): Likewise.
10988 (default_libgcc_shift_count_mode): Likewise.
10989 (default_unwind_word_mode): Likewise.
10990 (default_addr_space_pointer_mode): Likewise.
10991 (default_addr_space_address_mode): Likewise.
10992 * targhooks.h (default_eh_return_filter_mode): Likewise.
10993 (default_libgcc_cmp_return_mode): Likewise.
10994 (default_libgcc_shift_count_mode): Likewise.
10995 (default_unwind_word_mode): Likewise.
10996 (default_addr_space_pointer_mode): Likewise.
10997 (default_addr_space_address_mode): Likewise.
10998 (default_mode_rep_extended): Take a scalar_int_mode rather than
11000 (default_valid_pointer_mode): Likewise.
11001 (default_addr_space_valid_pointer_mode): Likewise.
11002 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11004 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11005 rather than a machine_mode.
11006 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11007 for local variables.
11008 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11009 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11010 than a machine_mode.
11012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11013 Alan Hayward <alan.hayward@arm.com>
11014 David Sherwood <david.sherwood@arm.com>
11016 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11017 the mode argument to scalar_int_mode.
11018 (do_jump_by_parts_zero_rtx): Likewise.
11019 (do_jump_by_parts_equality_rtx): Likewise.
11020 (do_jump_by_parts_greater): Take a mode argument.
11021 (do_jump_by_parts_equality): Likewise.
11022 (do_jump_1): Update calls accordingly.
11024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11025 Alan Hayward <alan.hayward@arm.com>
11026 David Sherwood <david.sherwood@arm.com>
11028 * is-a.h (safe_dyn_cast): New function.
11029 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11030 (jump_table_for_label): Likewise.
11031 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11032 instead of an rtx_insn *.
11033 (shorten_branches): Use dyn_cast instead of LABEL_P and
11034 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11035 rtx_jump_table_data::get_data_mode.
11036 (final_scan_insn): Likewise.
11038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11039 Alan Hayward <alan.hayward@arm.com>
11040 David Sherwood <david.sherwood@arm.com>
11042 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11043 trying to combine a full-register integer set with a subreg
11046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11047 Alan Hayward <alan.hayward@arm.com>
11048 David Sherwood <david.sherwood@arm.com>
11050 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11051 that is always either address_mode or pointer_mode.
11053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11054 Alan Hayward <alan.hayward@arm.com>
11055 David Sherwood <david.sherwood@arm.com>
11057 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11058 when the two are known to be equal.
11060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11061 Alan Hayward <alan.hayward@arm.com>
11062 David Sherwood <david.sherwood@arm.com>
11064 * simplify-rtx.c (simplify_const_unary_operation): Use
11065 is_a <scalar_int_mode> instead of checking for a nonzero
11066 precision. Forcibly convert op_mode to a scalar_int_mode
11067 in that case. More clearly differentiate the operand and
11068 result modes and use the former when deciding what the value
11069 of a count-bits operation should be. Use is_int_mode instead
11070 of checking for a MODE_INT. Remove redundant check for whether
11071 this mode has a zero precision.
11073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11074 Alan Hayward <alan.hayward@arm.com>
11075 David Sherwood <david.sherwood@arm.com>
11077 * optabs.c (widen_leading): Change the type of the mode argument
11078 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11079 (widen_bswap): Likewise.
11080 (expand_parity): Likewise.
11081 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11082 (expand_ffs): Likewise.
11083 (epand_unop): Check for scalar integer modes before calling the
11086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11087 Alan Hayward <alan.hayward@arm.com>
11088 David Sherwood <david.sherwood@arm.com>
11090 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11092 (expand_expr_real_1): Update call accordingly.
11094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11095 Alan Hayward <alan.hayward@arm.com>
11096 David Sherwood <david.sherwood@arm.com>
11098 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11099 value_mode arguments. Use scalar_int_mode internally.
11100 (store_bit_field_1): Rename the new integer mode from imode
11101 to op0_mode and use it instead of GET_MODE (op0). Update calls
11102 to store_split_bit_field, store_bit_field_using_insv and
11103 store_fixed_bit_field.
11104 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11105 Use scalar_int_mode internally. Use a bit count rather than a mode
11106 when calculating the largest bit size for get_best_mode.
11107 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11108 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11109 Remove assertion that OP0 has a scalar integer mode.
11110 (store_split_bit_field): Add op0_mode and value_mode arguments.
11111 Update calls to extract_fixed_bit_field.
11112 (extract_bit_field_using_extv): Add an op0_mode argument.
11113 Use scalar_int_mode internally.
11114 (extract_bit_field_1): Rename the new integer mode from imode to
11115 op0_mode and use it instead of GET_MODE (op0). Update calls to
11116 extract_split_bit_field, extract_bit_field_using_extv and
11117 extract_fixed_bit_field.
11118 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11119 to extract_split_bit_field and extract_fixed_bit_field_1.
11120 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11121 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11122 on the target mode.
11123 (extract_split_bit_field): Add an op0_mode argument. Update call
11124 to extract_fixed_bit_field.
11126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11127 Alan Hayward <alan.hayward@arm.com>
11128 David Sherwood <david.sherwood@arm.com>
11130 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11131 * explow.c (hard_function_value): Likewise.
11132 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11133 convert_to_mode call outside the loop.
11134 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11135 for the mode iterator. Require the mode specified by max_pieces
11137 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11139 (copy_blkmode_to_reg): Likewise.
11140 (set_storage_via_setmem): Likewise.
11141 * optabs.c (prepare_cmp_insn): Likewise.
11142 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11143 * stor-layout.c (finish_bitfield_representative): Likewise.
11145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11146 Alan Hayward <alan.hayward@arm.com>
11147 David Sherwood <david.sherwood@arm.com>
11149 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11150 * expr.c (convert_move): Use them.
11151 (convert_modes): Likewise.
11152 (store_expr_with_bounds): Likewise.
11154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11155 Alan Hayward <alan.hayward@arm.com>
11156 David Sherwood <david.sherwood@arm.com>
11158 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11159 parameter for the mode of "x". Remove the "known_x", "known_mode"
11160 and "known_ret" arguments. Change the type of the mode argument
11161 to scalar_int_mode.
11162 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11163 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11164 (reg_num_sign_bit_copies_for_combine): Likewise.
11165 * rtlanal.c (nonzero_bits1): Likewise.
11166 (num_sign_bit_copies1): Likewise.
11167 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11168 (reg_num_sign_bit_copies_general): Likewise.
11169 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11170 (reg_nonzero_bits_general): Likewise.
11172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11173 Alan Hayward <alan.hayward@arm.com>
11174 David Sherwood <david.sherwood@arm.com>
11176 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11177 than in subroutines. Return 1 for non-integer modes.
11178 (cached_num_sign_bit_copies): Change the type of the mode parameter
11179 to scalar_int_mode.
11180 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11181 classes. Handle CONST_INT_P first and then check whether X also
11182 has a scalar integer mode. Check the same thing for inner registers
11183 of a SUBREG and for values that are being extended or truncated.
11185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11190 in subroutines. Return the mode mask for non-integer modes.
11191 (cached_nonzero_bits): Change the type of the mode parameter
11192 to scalar_int_mode.
11193 (nonzero_bits1): Likewise. Remove early exit for other mode
11194 classes. Handle CONST_INT_P first and then check whether X
11195 also has a scalar integer mode.
11197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11198 Alan Hayward <alan.hayward@arm.com>
11199 David Sherwood <david.sherwood@arm.com>
11201 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11202 Return a scalar_int_mode and assert that the size is greater than
11204 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11205 (op_by_pieces_d::op_by_pieces_d): Likewise.
11206 (op_by_pieces_d::run): Likewise.
11207 (can_store_by_pieces): Likewise.
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 * combine.c (extract_left_shift): Add a mode argument and update
11215 (make_compound_operation_int): Change the type of the mode parameter
11216 to scalar_int_mode and update the call to extract_left_shift.
11218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11219 Alan Hayward <alan.hayward@arm.com>
11220 David Sherwood <david.sherwood@arm.com>
11222 * combine.c (simplify_and_const_int): Change the type of the mode
11223 parameter to scalar_int_mode.
11224 (simplify_and_const_int_1): Likewise. Update recursive call.
11226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11227 Alan Hayward <alan.hayward@arm.com>
11228 David Sherwood <david.sherwood@arm.com>
11230 * combine.c (simplify_compare_const): Check that the mode is a
11231 scalar_int_mode (rather than VOIDmode) before testing its
11233 (simplify_comparison): Move COMPARISON_P handling out of the
11234 loop and restrict the latter part of the loop to scalar_int_modes.
11235 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11236 and when considering SUBREG_REGs. Use is_int_mode instead of
11237 checking GET_MODE_CLASS against MODE_INT.
11239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11240 Alan Hayward <alan.hayward@arm.com>
11241 David Sherwood <david.sherwood@arm.com>
11243 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11244 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11245 shift_unit_mode and for modes involved in scalar shifts.
11247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11248 Alan Hayward <alan.hayward@arm.com>
11249 David Sherwood <david.sherwood@arm.com>
11251 * combine.c (force_int_to_mode): New function, split out from...
11252 (force_to_mode): ...here. Keep xmode up-to-date and use it
11253 instead of GET_MODE (x).
11255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11256 Alan Hayward <alan.hayward@arm.com>
11257 David Sherwood <david.sherwood@arm.com>
11259 * optabs-query.h (extraction_insn::struct_mode): Change type to
11260 opt_scalar_int_mode and update comment.
11261 (extraction_insn::field_mode): Change type to scalar_int_mode.
11262 (extraction_insn::pos_mode): Likewise.
11263 * combine.c (make_extraction): Update accordingly.
11264 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11265 (get_optab_extraction_insn): Likewise.
11266 * recog.c (simplify_while_replacing): Likewise.
11267 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11268 parameter to opt_scalar_int_mode.
11270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11271 Alan Hayward <alan.hayward@arm.com>
11272 David Sherwood <david.sherwood@arm.com>
11274 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11275 to a scalar_int_mode instead of a machine_mode.
11276 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11277 (get_best_mode): Return a boolean and use a pointer argument to store
11278 the selected mode. Replace the limit mode parameter with a bit limit.
11279 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11280 for the values returned by bit_field_mode_iterator::next_mode.
11281 (store_bit_field): Update call to get_best_mode.
11282 (store_fixed_bit_field): Likewise.
11283 (extract_fixed_bit_field): Likewise.
11284 * expr.c (optimize_bitfield_assignment_op): Likewise.
11285 * fold-const.c (optimize_bit_field_compare): Likewise.
11286 (fold_truth_andor_1): Likewise.
11287 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11288 Update for new type of m_mode.
11289 (get_best_mode): As above.
11291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11292 Alan Hayward <alan.hayward@arm.com>
11293 David Sherwood <david.sherwood@arm.com>
11295 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11296 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11297 (store_bit_field): Check is_a <scalar_int_mode> before calling
11298 strict_volatile_bitfield_p.
11299 (extract_bit_field): 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 * target.def (cstore_mode): Return a scalar_int_mode.
11306 * doc/tm.texi: Regenerate.
11307 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11308 * targhooks.h (default_cstore_mode): Likewise.
11309 * targhooks.c (default_cstore_mode): Likewise, using a forced
11311 * expmed.c (emit_cstore): Expect the target of the cstore to be
11314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11315 Alan Hayward <alan.hayward@arm.com>
11316 David Sherwood <david.sherwood@arm.com>
11318 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11320 (niter_desc): Likewise mode.
11321 (iv_analyze): Add a mode parameter.
11323 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11324 and change its type to scalar_int_mode.
11325 * loop-iv.c: Update commentary at head of file.
11326 (iv_constant): Pass the mode paraeter before the rtx it describes
11327 and change its type to scalar_int_mode. Remove VOIDmode handling.
11328 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11329 (iv_extend): Likewise.
11330 (shorten_into_mode): Likewise.
11331 (iv_add): Use scalar_int_mode.
11332 (iv_mult): Likewise.
11333 (iv_shift): Likewise.
11334 (canonicalize_iv_subregs): Likewise.
11335 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11336 it describes and change its mode to scalar_int_mode. Also change
11337 the type of the returned inner_mode to scalar_int_mode.
11338 (get_biv_step): Likewise, turning outer_mode from a pointer
11339 into a direct parameter. Update call to get_biv_step_1.
11340 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11341 iv_constant and get_biv_step.
11342 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11343 and change its type to scalar_int_mode. Don't initialise iv->mode
11344 to VOIDmode and remove later checks for its still being VOIDmode.
11345 Update calls to iv_analyze_op and iv_analyze_expr. Check
11346 is_a <scalar_int_mode> when changing the mode under consideration.
11347 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11348 Update call to iv_analyze_expr.
11349 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11350 inner register is not also a scalar_int_mode. Update call to
11352 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11353 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11354 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11355 separate mode class checks. Update calls to iv_analyze. Remove
11356 fix-up of VOIDmodes after iv_analyze_biv.
11357 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11358 don't have a scalar_int_mode. Update call to biv_p.
11360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11361 Alan Hayward <alan.hayward@arm.com>
11362 David Sherwood <david.sherwood@arm.com>
11364 * cfgexpand.c (convert_debug_memory_address): Use
11365 as_a <scalar_int_mode>.
11366 * combine.c (expand_compound_operation): Likewise.
11367 (make_extraction): Likewise.
11368 (change_zero_ext): Likewise.
11369 (simplify_comparison): Likewise.
11370 * cse.c (cse_insn): Likewise.
11371 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11372 (mem_loc_descriptor): Likewise.
11373 (loc_descriptor): Likewise.
11374 * expmed.c (init_expmed_one_mode): Likewise.
11375 (synth_mult): Likewise.
11376 (emit_store_flag_1): Likewise.
11377 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11378 of a comparison with size.
11379 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11380 (reduce_to_bit_field_precision): Likewise.
11381 * function.c (expand_function_end): Likewise.
11382 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11383 * loop-doloop.c (doloop_modify): Likewise.
11384 * optabs.c (expand_binop): Likewise.
11385 (expand_unop): Likewise.
11386 (expand_copysign_absneg): Likewise.
11387 (prepare_cmp_insn): Likewise.
11388 (maybe_legitimize_operand): Likewise.
11389 * recog.c (const_scalar_int_operand): Likewise.
11390 * rtlanal.c (get_address_mode): Likewise.
11391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11392 (simplify_cond_clz_ctz): Likewise.
11393 * tree-nested.c (get_nl_goto_field): Likewise.
11394 * tree.c (build_vector_type_for_mode): Likewise.
11395 * var-tracking.c (use_narrower_mode): Likewise.
11397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11398 Alan Hayward <alan.hayward@arm.com>
11399 David Sherwood <david.sherwood@arm.com>
11401 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11402 * builtins.c (expand_builtin_signbit): Use it.
11403 * cfgexpand.c (expand_debug_expr): Likewise.
11404 * dojump.c (do_jump): Likewise.
11405 (do_compare_and_jump): Likewise.
11406 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11407 * expmed.c (make_tree): Likewise.
11408 * expr.c (expand_expr_real_2): Likewise.
11409 (expand_expr_real_1): Likewise.
11410 (try_casesi): Likewise.
11411 * fold-const-call.c (fold_const_call_ss): Likewise.
11412 * fold-const.c (unextend): Likewise.
11413 (extract_muldiv_1): Likewise.
11414 (fold_single_bit_test): Likewise.
11415 (native_encode_int): Likewise.
11416 (native_encode_string): Likewise.
11417 (native_interpret_int): Likewise.
11418 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11419 * internal-fn.c (expand_addsub_overflow): Likewise.
11420 (expand_neg_overflow): Likewise.
11421 (expand_mul_overflow): Likewise.
11422 (expand_arith_overflow): Likewise.
11423 * match.pd: Likewise.
11424 * stor-layout.c (layout_type): Likewise.
11425 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11426 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11427 * tree-ssanames.c (get_range_info): Likewise.
11428 * tree-switch-conversion.c (array_value_type) Likewise.
11429 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11430 (vect_recog_divmod_pattern): Likewise.
11431 (vect_recog_mixed_size_cond_pattern): Likewise.
11432 * tree-vrp.c (extract_range_basic): Likewise.
11433 (simplify_float_conversion_using_ranges): Likewise.
11434 * tree.c (int_fits_type_p): Likewise.
11435 * ubsan.c (instrument_bool_enum_load): Likewise.
11436 * varasm.c (mergeable_string_section): Likewise.
11437 (narrowing_initializer_constant_valid_p): Likewise.
11438 (output_constant): Likewise.
11440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11441 Alan Hayward <alan.hayward@arm.com>
11442 David Sherwood <david.sherwood@arm.com>
11444 * machmode.h (NARROWEST_INT_MODE): New macro.
11445 * expr.c (alignment_for_piecewise_move): Use it instead of
11446 GET_CLASS_NARROWEST_MODE (MODE_INT).
11447 (push_block): Likewise.
11448 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11450 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11453 Alan Hayward <alan.hayward@arm.com>
11454 David Sherwood <david.sherwood@arm.com>
11456 * postreload.c (move2add_valid_value_p): Change the type of the
11457 mode parameter to scalar_int_mode.
11458 (move2add_use_add2_insn): Add a mode parameter and use it instead
11460 (move2add_use_add3_insn): Likewise.
11461 (reload_cse_move2add): Update accordingly.
11463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11464 Alan Hayward <alan.hayward@arm.com>
11465 David Sherwood <david.sherwood@arm.com>
11467 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11469 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11470 * optabs.c (expand_unop): Likewise.
11472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11473 Alan Hayward <alan.hayward@arm.com>
11474 David Sherwood <david.sherwood@arm.com>
11476 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11477 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11478 (popcount_loc_descriptor): Likewise.
11479 (bswap_loc_descriptor): Likewise.
11480 (rotate_loc_descriptor): Likewise.
11481 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11482 calling the functions above.
11484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11485 Alan Hayward <alan.hayward@arm.com>
11486 David Sherwood <david.sherwood@arm.com>
11488 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11490 (try_combine): Likewise.
11491 (simplify_if_then_else): Likewise.
11492 * cse.c (cse_insn): Likewise.
11493 * dwarf2out.c (mem_loc_descriptor): Likewise.
11494 * emit-rtl.c (gen_lowpart_common): Likewise.
11495 * simplify-rtx.c (simplify_truncation): Likewise.
11496 (simplify_binary_operation_1): Likewise.
11497 (simplify_const_relational_operation): Likewise.
11498 (simplify_ternary_operation): Likewise.
11499 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11502 Alan Hayward <alan.hayward@arm.com>
11503 David Sherwood <david.sherwood@arm.com>
11505 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11506 * reload.c (push_reload): Likewise.
11507 (find_reloads): Likewise.
11509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11510 Alan Hayward <alan.hayward@arm.com>
11511 David Sherwood <david.sherwood@arm.com>
11513 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11514 (make_compound_operation_int): Likewise.
11515 (change_zero_ext): Likewise.
11516 * expr.c (convert_move): Likewise.
11517 (convert_modes): Likewise.
11518 * fwprop.c (forward_propagate_subreg): Likewise.
11519 * loop-iv.c (get_biv_step_1): Likewise.
11520 * optabs.c (widen_operand): Likewise.
11521 * postreload.c (move2add_valid_value_p): Likewise.
11522 * recog.c (simplify_while_replacing): Likewise.
11523 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11524 (simplify_binary_operation_1): Likewise. Remove redundant
11525 mode equality check.
11527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11528 Alan Hayward <alan.hayward@arm.com>
11529 David Sherwood <david.sherwood@arm.com>
11531 * combine.c (combine_simplify_rtx): Add checks for
11532 is_a <scalar_int_mode>.
11533 (simplify_if_then_else): Likewise.
11534 (make_field_assignment): Likewise.
11535 (simplify_comparison): Likewise.
11536 * ifcvt.c (noce_try_bitop): Likewise.
11537 * loop-invariant.c (canonicalize_address_mult): Likewise.
11538 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11541 Alan Hayward <alan.hayward@arm.com>
11542 David Sherwood <david.sherwood@arm.com>
11544 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11545 is_a <scalar_int_mode> instead of != BLKmode.
11547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11548 Alan Hayward <alan.hayward@arm.com>
11549 David Sherwood <david.sherwood@arm.com>
11551 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11552 instead of != VOIDmode.
11553 * combine.c (if_then_else_cond): Likewise.
11554 (change_zero_ext): Likewise.
11555 * dwarf2out.c (mem_loc_descriptor): Likewise.
11556 (loc_descriptor): Likewise.
11557 * rtlanal.c (canonicalize_condition): Likewise.
11558 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11561 Alan Hayward <alan.hayward@arm.com>
11562 David Sherwood <david.sherwood@arm.com>
11564 * simplify-rtx.c (simplify_binary_operation_1): Use
11565 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * wide-int.h (int_traits<unsigned char>) New class.
11572 (int_traits<unsigned short>) Likewise.
11573 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11574 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11576 * combine.c (set_nonzero_bits_and_sign_copies): Use
11577 is_a <scalar_int_mode>.
11578 (find_split_point): Likewise.
11579 (combine_simplify_rtx): Likewise.
11580 (simplify_logical): Likewise.
11581 (expand_compound_operation): Likewise.
11582 (expand_field_assignment): Likewise.
11583 (make_compound_operation): Likewise.
11584 (extended_count): Likewise.
11585 (change_zero_ext): Likewise.
11586 (simplify_comparison): Likewise.
11587 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11588 (ucompare_loc_descriptor): Likewise.
11589 (minmax_loc_descriptor): Likewise.
11590 (mem_loc_descriptor): Likewise.
11591 (loc_descriptor): Likewise.
11592 * expmed.c (init_expmed_one_mode): Likewise.
11593 * lra-constraints.c (lra_constraint_offset): Likewise.
11594 * optabs.c (prepare_libcall_arg): Likewise.
11595 * postreload.c (move2add_note_store): Likewise.
11596 * reload.c (operands_match_p): Likewise.
11597 * rtl.h (load_extend_op): Likewise.
11598 * rtlhooks.c (gen_lowpart_general): Likewise.
11599 * simplify-rtx.c (simplify_truncation): Likewise.
11600 (simplify_unary_operation_1): Likewise.
11601 (simplify_binary_operation_1): Likewise.
11602 (simplify_const_binary_operation): Likewise.
11603 (simplify_const_relational_operation): Likewise.
11604 (simplify_subreg): Likewise.
11605 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11606 * var-tracking.c (adjust_mems): Likewise.
11607 (prepare_call_arguments): Likewise.
11609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11610 Alan Hayward <alan.hayward@arm.com>
11611 David Sherwood <david.sherwood@arm.com>
11613 * machmode.h (is_int_mode): New fuction.
11614 * combine.c (find_split_point): Use it.
11615 (combine_simplify_rtx): Likewise.
11616 (simplify_if_then_else): Likewise.
11617 (simplify_set): Likewise.
11618 (simplify_shift_const_1): Likewise.
11619 (simplify_comparison): Likewise.
11620 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11621 * cse.c (notreg_cost): Likewise.
11622 (cse_insn): Likewise.
11623 * cselib.c (cselib_lookup_1): Likewise.
11624 * dojump.c (do_jump_1): Likewise.
11625 (do_compare_rtx_and_jump): Likewise.
11626 * dse.c (get_call_args): Likewise.
11627 * dwarf2out.c (rtl_for_decl_init): Likewise.
11628 (native_encode_initializer): Likewise.
11629 * expmed.c (emit_store_flag_1): Likewise.
11630 (emit_store_flag): Likewise.
11631 * expr.c (convert_modes): Likewise.
11632 (store_field): Likewise.
11633 (expand_expr_real_1): Likewise.
11634 * fold-const.c (fold_read_from_constant_string): Likewise.
11635 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11636 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11637 * optabs.c (expand_binop): Likewise.
11638 (expand_unop): Likewise.
11639 (expand_abs_nojump): Likewise.
11640 (expand_one_cmpl_abs_nojump): Likewise.
11641 * simplify-rtx.c (mode_signbit_p): Likewise.
11642 (val_signbit_p): Likewise.
11643 (val_signbit_known_set_p): Likewise.
11644 (val_signbit_known_clear_p): Likewise.
11645 (simplify_relational_operation_1): Likewise.
11646 * tree.c (vector_type_mode): Likewise.
11648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11649 Alan Hayward <alan.hayward@arm.com>
11650 David Sherwood <david.sherwood@arm.com>
11652 * machmode.h (smallest_mode_for_size): Fix formatting.
11653 (smallest_int_mode_for_size): New function.
11654 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11655 instead of smallest_mode_for_size.
11656 * combine.c (make_extraction): Likewise.
11657 * config/arc/arc.c (arc_expand_movmem): Likewise.
11658 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11659 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11660 * config/s390/s390.c (s390_expand_insv): Likewise.
11661 * config/sparc/sparc.c (assign_int_registers): Likewise.
11662 * config/spu/spu.c (spu_function_value): Likewise.
11663 (spu_function_arg): Likewise.
11664 * coverage.c (get_gcov_type): Likewise.
11665 (get_gcov_unsigned_t): Likewise.
11666 * dse.c (find_shift_sequence): Likewise.
11667 * expmed.c (store_bit_field_1): Likewise.
11668 * expr.c (convert_move): Likewise.
11669 (store_field): Likewise.
11670 * internal-fn.c (expand_arith_overflow): Likewise.
11671 * optabs-query.c (get_best_extraction_insn): Likewise.
11672 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11673 * stor-layout.c (layout_type): Likewise.
11674 (initialize_sizetypes): Likewise.
11675 * targhooks.c (default_get_mask_mode): Likewise.
11676 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11679 Alan Hayward <alan.hayward@arm.com>
11680 David Sherwood <david.sherwood@arm.com>
11682 * machmode.h (opt_mode::else_blk): New function.
11683 (int_mode_for_mode): Declare.
11684 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11685 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11687 * cfgexpand.c (expand_debug_expr): Likewise.
11688 * combine.c (gen_lowpart_or_truncate): Likewise.
11689 (gen_lowpart_for_combine): Likewise.
11690 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11691 * config/avr/avr.c (avr_to_int_mode): Likewise.
11692 (avr_out_plus_1): Likewise.
11693 (avr_out_plus): Likewise.
11694 (avr_out_round): Likewise.
11695 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11696 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11697 (s390_expand_vcond): Likewise.
11698 * config/spu/spu.c (spu_split_immediate): Likewise.
11699 (spu_expand_mov): Likewise.
11700 * dse.c (get_stored_val): Likewise.
11701 * expmed.c (store_bit_field_1): Likewise.
11702 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11704 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11705 (extract_low_bits): Likewise.
11706 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11707 handling rather than repeating the check.
11708 (emit_group_store): Likewise.
11709 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11710 * optabs.c (expand_absneg_bit): Likewise.
11711 (expand_copysign_absneg): Likewise.
11712 (expand_copysign_bit): Likewise.
11713 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11714 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11715 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11716 * var-tracking.c (prepare_call_arguments): Likewise.
11717 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11718 int_mode_for_mode instead of mode_for_size.
11719 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11722 Alan Hayward <alan.hayward@arm.com>
11723 David Sherwood <david.sherwood@arm.com>
11725 * machmode.h (int_mode_for_size): New function.
11726 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11727 instead of mode_for_size.
11728 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11730 * combine.c (expand_field_assignment): Use int_mode_for_size
11731 instead of mode_for_size.
11732 (make_extraction): Likewise.
11733 (simplify_shift_const_1): Likewise.
11734 (simplify_comparison): Likewise.
11735 * dojump.c (do_jump): Likewise.
11736 * dwarf2out.c (mem_loc_descriptor): Likewise.
11737 * emit-rtl.c (init_derived_machine_modes): Likewise.
11738 * expmed.c (flip_storage_order): Likewise.
11739 (convert_extracted_bit_field): Likewise.
11740 * expr.c (copy_blkmode_from_reg): Likewise.
11741 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11742 * internal-fn.c (expand_mul_overflow): Likewise.
11743 * lower-subreg.c (simple_move): Likewise.
11744 * optabs-libfuncs.c (init_optabs): Likewise.
11745 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11746 * tree.c (vector_type_mode): Likewise.
11747 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11748 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11749 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11750 * tree-vect-stmts.c (vectorizable_load): Likewise.
11751 (vectorizable_store): Likewise.
11753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11754 Alan Hayward <alan.hayward@arm.com>
11755 David Sherwood <david.sherwood@arm.com>
11757 * coretypes.h (pod_mode): New type.
11758 (scalar_int_mode_pod): New typedef.
11759 * machmode.h (pod_mode): New class.
11760 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11761 * genmodes.c (emit_mode_int_n): Update accordingly.
11762 * lower-subreg.h (target_lower_subreg): Change type to
11763 scalar_int_mode_pod.
11764 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11765 scalar_int_mode_pod.
11767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11768 Alan Hayward <alan.hayward@arm.com>
11769 David Sherwood <david.sherwood@arm.com>
11771 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11772 machine_mode to scalar_int_mode.
11773 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11774 (rs6000_option_override_internal): Remove cast to int.
11775 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11776 machine_mode to scalar_int_mode.
11777 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11778 (rs6000_option_override_internal): Remove cast to int.
11779 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11780 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11782 * config/s390/s390.c (s390_expand_builtin): Likewise.
11783 * coretypes.h (scalar_int_mode): New type.
11784 (opt_scalar_int_mode): New typedef.
11785 * machmode.h (scalar_int_mode): New class.
11786 (scalar_int_mode::includes_p): New function.
11787 (byte_mode): Change type to scalar_int_mode.
11788 (word_mode): Likewise.
11789 (ptr_mode): Likewise.
11790 * emit-rtl.c (byte_mode): Likewise.
11791 (word_mode): Likewise.
11792 (ptr_mode): Likewise.
11793 (init_derived_machine_modes): Update accordingly.
11794 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11795 and MODE_PARTIAL_INT.
11796 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11797 opt_scalar_int_mode.
11799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11800 Alan Hayward <alan.hayward@arm.com>
11801 David Sherwood <david.sherwood@arm.com>
11803 * target.def (libgcc_floating_mode_supported_p): Take a
11805 * doc/tm.texi: Regenerate.
11806 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11808 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11809 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11813 Alan Hayward <alan.hayward@arm.com>
11814 David Sherwood <david.sherwood@arm.com>
11816 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11817 * doc/tm.texi: Regenerate.
11818 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11819 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11820 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11821 * targhooks.h (default_floatn_mode): Likewise.
11822 * targhooks.c (default_floatn_mode): Likewise.
11823 * tree.c (build_common_tree_nodes): Update accordingly.
11825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11826 Alan Hayward <alan.hayward@arm.com>
11827 David Sherwood <david.sherwood@arm.com>
11829 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11830 (mode_iterator::iterate_p): Likewise.
11831 (mode_iterator::get_wider): Likewise.
11832 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
11839 * machmode.h (float_mode_for_size): New function.
11840 * emit-rtl.c (double_mode): Delete.
11841 (init_emit_once): Use float_mode_for_size.
11842 * stor-layout.c (layout_type): Likewise.
11843 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11846 Alan Hayward <alan.hayward@arm.com>
11847 David Sherwood <david.sherwood@arm.com>
11849 * output.h (assemble_real): Take a scalar_float_mode.
11850 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11851 * config/arm/arm.md (consttable_4): Likewise.
11852 (consttable_8): Likewise.
11853 (consttable_16): Likewise.
11854 * config/mips/mips.md (consttable_float): Likewise.
11855 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11856 * varasm.c (assemble_real): Take a scalar_float_mode.
11857 (output_constant_pool_2): Update accordingly.
11858 (output_constant): Likewise.
11860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11861 Alan Hayward <alan.hayward@arm.com>
11862 David Sherwood <david.sherwood@arm.com>
11864 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11865 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11866 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11867 (native_encode_real): Likewise.
11868 (native_interpret_real): Likewise.
11869 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11870 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11873 Alan Hayward <alan.hayward@arm.com>
11874 David Sherwood <david.sherwood@arm.com>
11876 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11877 <scalar_float_mode>. Simplify.
11878 (gen_extend_conv_libfunc): Likewise.
11880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11881 Alan Hayward <alan.hayward@arm.com>
11882 David Sherwood <david.sherwood@arm.com>
11884 * coretypes.h (scalar_float_mode): New type.
11885 * machmode.h (mode_traits::from_int): Use machine_mode if
11886 USE_ENUM_MODES is defined.
11887 (is_a): New function.
11889 (dyn_cast): Likewise.
11890 (scalar_float_mode): New class.
11891 (scalar_float_mode::includes_p): New function.
11892 (is_float_mode): Likewise.
11893 * gdbhooks.py (MachineModePrinter): New class.
11894 (build_pretty_printer): Use it for scalar_float_mode.
11895 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11896 (format_helper::format_helper): Turn into a template.
11897 * genmodes.c (get_mode_class): New function.
11898 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11899 or machine_mode if none.
11900 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11901 as_a <scalar_float_mode>.
11902 * dwarf2out.c (mem_loc_descriptor): Likewise.
11903 (insert_float): Likewise.
11904 (add_const_value_attribute): Likewise.
11905 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11906 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11907 (expand_unop): Update accordingly.
11908 (expand_abs_nojump): Likewise.
11909 (expand_copysign_absneg): Take a scalar_float_mode.
11910 (expand_copysign_bit): Likewise.
11911 (expand_copysign): Update accordingly.
11913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11914 Alan Hayward <alan.hayward@arm.com>
11915 David Sherwood <david.sherwood@arm.com>
11917 * coretypes.h (opt_mode): New class.
11918 * machmode.h (opt_mode): Likewise.
11919 (opt_mode::else_void): New function.
11920 (opt_mode::require): Likewise.
11921 (opt_mode::exists): Likewise.
11922 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11923 (GET_MODE_2XWIDER_MODE): Likewise.
11924 (mode_iterator::get_wider): Update accordingly.
11925 (mode_iterator::get_2xwider): Likewise.
11926 (mode_iterator::get_known_wider): Likewise, turning into a template.
11927 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11928 forcing a wider mode to exist.
11929 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11930 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11931 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11932 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11933 * lower-subreg.c (init_lower_subreg): Likewise.
11934 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11935 on the final iteration.
11936 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11937 a wider mode exists before asking for a move pattern.
11938 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11939 forcing a wider mode to exist.
11940 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11941 returning false if no such mode exists.
11942 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11943 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11944 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11945 Avoid checking for a MODE_INT if we already know the mode is not a
11947 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11948 forcing a wider mode to exist.
11949 (expmed_mult_highpart_optab): Likewise.
11950 (expmed_mult_highpart): Likewise.
11951 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11953 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11954 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11955 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11956 * internal-fn.c (expand_mul_overflow): Update use of
11957 GET_MODE_2XWIDER_MODE.
11958 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11959 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11960 GET_MODE_WIDER_MODE.
11961 (convert_plusminus_to_widen): Likewise.
11962 * tree-switch-conversion.c (array_value_type): Likewise.
11963 * var-tracking.c (emit_note_insn_var_location): Likewise.
11964 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11965 Return false inside rather than outside the loop if no wider mode
11967 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11968 and GET_MODE_2XWIDER_MODE
11969 (can_compare_p): Use else_void.
11970 * gdbhooks.py (OptMachineModePrinter): New class.
11971 (build_pretty_printer): Use it for opt_mode.
11973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11974 Alan Hayward <alan.hayward@arm.com>
11975 David Sherwood <david.sherwood@arm.com>
11977 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11978 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11981 Alan Hayward <alan.hayward@arm.com>
11982 David Sherwood <david.sherwood@arm.com>
11984 * machmode.h (mode_traits): New structure.
11985 (get_narrowest_mode): New function.
11986 (mode_iterator::start): Likewise.
11987 (mode_iterator::iterate_p): Likewise.
11988 (mode_iterator::get_wider): Likewise.
11989 (mode_iterator::get_known_wider): Likewise.
11990 (mode_iterator::get_2xwider): Likewise.
11991 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11992 (FOR_EACH_MODE): Likewise.
11993 (FOR_EACH_MODE_FROM): Likewise.
11994 (FOR_EACH_MODE_UNTIL): Likewise.
11995 (FOR_EACH_WIDER_MODE): Likewise.
11996 (FOR_EACH_2XWIDER_MODE): Likewise.
11997 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11998 * combine.c (simplify_comparison): Likewise
11999 * config/i386/i386.c (type_natural_mode): Likewise.
12000 * cse.c (cse_insn): Likewise.
12001 * dse.c (find_shift_sequence): Likewise.
12002 * emit-rtl.c (init_derived_machine_modes): Likewise.
12003 (init_emit_once): Likewise.
12004 * explow.c (hard_function_value): Likewise.
12005 * expmed.c (extract_fixed_bit_field_1): Likewise.
12006 (extract_bit_field_1): Likewise.
12007 (expand_divmod): Likewise.
12008 (emit_store_flag_1): Likewise.
12009 * expr.c (init_expr_target): Likewise.
12010 (convert_move): Likewise.
12011 (alignment_for_piecewise_move): Likewise.
12012 (widest_int_mode_for_size): Likewise.
12013 (emit_block_move_via_movmem): Likewise.
12014 (copy_blkmode_to_reg): Likewise.
12015 (set_storage_via_setmem): Likewise.
12016 (compress_float_constant): Likewise.
12017 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12018 * optabs-query.c (get_best_extraction_insn): Likewise.
12019 * optabs.c (expand_binop): Likewise.
12020 (expand_twoval_unop): Likewise.
12021 (expand_twoval_binop): Likewise.
12022 (widen_leading): Likewise.
12023 (widen_bswap): Likewise.
12024 (expand_parity): Likewise.
12025 (expand_unop): Likewise.
12026 (prepare_cmp_insn): Likewise.
12027 (prepare_float_lib_cmp): Likewise.
12028 (expand_float): Likewise.
12029 (expand_fix): Likewise.
12030 (expand_sfix_optab): Likewise.
12031 * postreload.c (move2add_use_add2_insn): Likewise.
12032 * reg-stack.c (reg_to_stack): Likewise.
12033 * reginfo.c (choose_hard_reg_mode): Likewise.
12034 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12035 * stor-layout.c (mode_for_size): Likewise.
12036 (smallest_mode_for_size): Likewise.
12037 (mode_for_vector): Likewise.
12038 (finish_bitfield_representative): Likewise.
12039 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12040 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12041 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12042 * var-tracking.c (prepare_call_arguments): Likewise.
12044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12045 Alan Hayward <alan.hayward@arm.com>
12046 David Sherwood <david.sherwood@arm.com>
12048 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12049 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12050 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12051 * machmode.h (mode_size): Move earlier in file.
12052 (mode_precision): Likewise.
12053 (mode_inner): Likewise.
12054 (mode_nunits): Likewise.
12055 (mode_unit_size): Likewise.
12056 (unit_unit_precision): Likewise.
12057 (mode_wider): Likewise.
12058 (mode_2xwider): Likewise.
12059 (machine_mode): New class.
12060 (mode_to_bytes): New function.
12061 (mode_to_bits): Likewise.
12062 (mode_to_precision): Likewise.
12063 (mode_to_inner): Likewise.
12064 (mode_to_unit_size): Likewise.
12065 (mode_to_unit_precision): Likewise.
12066 (mode_to_nunits): Likewise.
12067 (GET_MODE_SIZE): Use mode_to_bytes.
12068 (GET_MODE_BITSIZE): Use mode_to_bits.
12069 (GET_MODE_PRECISION): Use mode_to_precision.
12070 (GET_MODE_INNER): Use mode_to_inner.
12071 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12072 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12073 (GET_MODE_NUNITS): Use mode_to_nunits.
12074 * system.h (ALWAYS_INLINE): New macro.
12075 * config/powerpcspe/powerpcspe-c.c
12076 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12077 int for arg1_mode and arg2_mode.
12079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12080 Alan Hayward <alan.hayward@arm.com>
12081 David Sherwood <david.sherwood@arm.com>
12083 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12084 Prefix mode names with E_ in case statements.
12085 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12086 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12087 (aarch64_split_simd_move): Likewise.
12088 (aarch64_gen_storewb_pair): Likewise.
12089 (aarch64_gen_loadwb_pair): Likewise.
12090 (aarch64_gen_store_pair): Likewise.
12091 (aarch64_gen_load_pair): Likewise.
12092 (aarch64_get_condition_code_1): Likewise.
12093 (aarch64_constant_pool_reload_icode): Likewise.
12094 (get_rsqrte_type): Likewise.
12095 (get_rsqrts_type): Likewise.
12096 (get_recpe_type): Likewise.
12097 (get_recps_type): Likewise.
12098 (aarch64_gimplify_va_arg_expr): Likewise.
12099 (aarch64_simd_container_mode): Likewise.
12100 (aarch64_emit_load_exclusive): Likewise.
12101 (aarch64_emit_store_exclusive): Likewise.
12102 (aarch64_expand_compare_and_swap): Likewise.
12103 (aarch64_gen_atomic_cas): Likewise.
12104 (aarch64_emit_bic): Likewise.
12105 (aarch64_emit_atomic_swap): Likewise.
12106 (aarch64_emit_atomic_load_op): Likewise.
12107 (aarch64_evpc_trn): Likewise.
12108 (aarch64_evpc_uzp): Likewise.
12109 (aarch64_evpc_zip): Likewise.
12110 (aarch64_evpc_ext): Likewise.
12111 (aarch64_evpc_rev): Likewise.
12112 (aarch64_evpc_dup): Likewise.
12113 (aarch64_gen_ccmp_first): Likewise.
12114 (aarch64_gen_ccmp_next): Likewise.
12115 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12116 (alpha_emit_xfloating_libcall): Likewise.
12117 (emit_insxl): Likewise.
12118 (alpha_arg_type): Likewise.
12119 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12120 (arc_preferred_simd_mode): Likewise.
12121 (arc_secondary_reload): Likewise.
12122 (get_arc_condition_code): Likewise.
12123 (arc_print_operand): Likewise.
12124 (arc_legitimate_constant_p): Likewise.
12125 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12126 * config/arc/arc.md (casesi_load): Likewise.
12127 (casesi_compact_jump): Likewise.
12128 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12129 (cc_use_register): Likewise.
12130 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12131 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12132 (arm_init_iwmmxt_builtins): Likewise.
12133 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12134 (neon_expand_vector_init): Likewise.
12135 (arm_attr_length_move_neon): Likewise.
12136 (maybe_get_arm_condition_code): Likewise.
12137 (arm_emit_vector_const): Likewise.
12138 (arm_preferred_simd_mode): Likewise.
12139 (arm_output_iwmmxt_tinsr): Likewise.
12140 (thumb1_output_casesi): Likewise.
12141 (thumb2_output_casesi): Likewise.
12142 (arm_emit_load_exclusive): Likewise.
12143 (arm_emit_store_exclusive): Likewise.
12144 (arm_expand_compare_and_swap): Likewise.
12145 (arm_evpc_neon_vuzp): Likewise.
12146 (arm_evpc_neon_vzip): Likewise.
12147 (arm_evpc_neon_vrev): Likewise.
12148 (arm_evpc_neon_vtrn): Likewise.
12149 (arm_evpc_neon_vext): Likewise.
12150 (arm_validize_comparison): Likewise.
12151 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12152 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12153 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12154 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12155 (c6x_preferred_simd_mode): Likewise.
12156 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12157 (epiphany_rtx_costs): Likewise.
12158 * config/epiphany/predicates.md (proper_comparison_operator):
12160 * config/frv/frv.c (condexec_memory_operand): Likewise.
12161 (frv_emit_move): Likewise.
12162 (output_move_single): Likewise.
12163 (output_condmove_single): Likewise.
12164 (frv_hard_regno_mode_ok): Likewise.
12165 (frv_matching_accg_mode): Likewise.
12166 * config/h8300/h8300.c (split_adds_subs): Likewise.
12167 (h8300_rtx_costs): Likewise.
12168 (h8300_print_operand): Likewise.
12169 (compute_mov_length): Likewise.
12170 (output_logical_op): Likewise.
12171 (compute_logical_op_length): Likewise.
12172 (compute_logical_op_cc): Likewise.
12173 (h8300_shift_needs_scratch_p): Likewise.
12174 (output_a_shift): Likewise.
12175 (compute_a_shift_length): Likewise.
12176 (compute_a_shift_cc): Likewise.
12177 (expand_a_rotate): Likewise.
12178 (output_a_rotate): Likewise.
12179 * config/i386/i386.c (classify_argument): Likewise.
12180 (function_arg_advance_32): Likewise.
12181 (function_arg_32): Likewise.
12182 (function_arg_64): Likewise.
12183 (function_value_64): Likewise.
12184 (ix86_gimplify_va_arg): Likewise.
12185 (ix86_legitimate_constant_p): Likewise.
12186 (put_condition_code): Likewise.
12187 (split_double_mode): Likewise.
12188 (ix86_avx256_split_vector_move_misalign): Likewise.
12189 (ix86_expand_vector_logical_operator): Likewise.
12190 (ix86_split_idivmod): Likewise.
12191 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12192 (ix86_build_const_vector): Likewise.
12193 (ix86_build_signbit_mask): Likewise.
12194 (ix86_match_ccmode): Likewise.
12195 (ix86_cc_modes_compatible): Likewise.
12196 (ix86_expand_branch): Likewise.
12197 (ix86_expand_sse_cmp): Likewise.
12198 (ix86_expand_sse_movcc): Likewise.
12199 (ix86_expand_int_sse_cmp): Likewise.
12200 (ix86_expand_vec_perm_vpermi2): Likewise.
12201 (ix86_expand_vec_perm): Likewise.
12202 (ix86_expand_sse_unpack): Likewise.
12203 (ix86_expand_int_addcc): Likewise.
12204 (ix86_split_to_parts): Likewise.
12205 (ix86_vectorize_builtin_gather): Likewise.
12206 (ix86_vectorize_builtin_scatter): Likewise.
12207 (avx_vpermilp_parallel): Likewise.
12208 (inline_memory_move_cost): Likewise.
12209 (ix86_tieable_integer_mode_p): Likewise.
12210 (x86_maybe_negate_const_int): Likewise.
12211 (ix86_expand_vector_init_duplicate): Likewise.
12212 (ix86_expand_vector_init_one_nonzero): Likewise.
12213 (ix86_expand_vector_init_one_var): Likewise.
12214 (ix86_expand_vector_init_concat): Likewise.
12215 (ix86_expand_vector_init_interleave): Likewise.
12216 (ix86_expand_vector_init_general): Likewise.
12217 (ix86_expand_vector_set): Likewise.
12218 (ix86_expand_vector_extract): Likewise.
12219 (emit_reduc_half): Likewise.
12220 (ix86_emit_i387_round): Likewise.
12221 (ix86_mangle_type): Likewise.
12222 (ix86_expand_round_sse4): Likewise.
12223 (expand_vec_perm_blend): Likewise.
12224 (canonicalize_vector_int_perm): Likewise.
12225 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12226 (expand_vec_perm_1): Likewise.
12227 (expand_vec_perm_interleave3): Likewise.
12228 (expand_vec_perm_even_odd_pack): Likewise.
12229 (expand_vec_perm_even_odd_1): Likewise.
12230 (expand_vec_perm_broadcast_1): Likewise.
12231 (ix86_vectorize_vec_perm_const_ok): Likewise.
12232 (ix86_expand_vecop_qihi): Likewise.
12233 (ix86_expand_mul_widen_hilo): Likewise.
12234 (ix86_expand_sse2_abs): Likewise.
12235 (ix86_expand_pextr): Likewise.
12236 (ix86_expand_pinsr): Likewise.
12237 (ix86_preferred_simd_mode): Likewise.
12238 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12239 * config/i386/sse.md (*andnot<mode>3): Likewise.
12240 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12241 (*<code><mode>3): Likewise.
12242 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12243 (ia64_expand_atomic_op): Likewise.
12244 (ia64_arg_type): Likewise.
12245 (ia64_mode_to_int): Likewise.
12246 (ia64_scalar_mode_supported_p): Likewise.
12247 (ia64_vector_mode_supported_p): Likewise.
12248 (expand_vec_perm_broadcast): Likewise.
12249 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12250 (iq2000_function_arg_advance): Likewise.
12251 (iq2000_function_arg): Likewise.
12252 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12253 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12254 (m68k_libcall_value): Likewise.
12255 (m68k_function_value): Likewise.
12256 (sched_attr_op_type): Likewise.
12257 * config/mcore/mcore.c (mcore_output_move): Likewise.
12258 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12260 (microblaze_function_arg): Likewise.
12261 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12262 (mips_print_operand): Likewise.
12263 (mips_mode_ok_for_mov_fmt_p): Likewise.
12264 (mips_vector_mode_supported_p): Likewise.
12265 (mips_preferred_simd_mode): Likewise.
12266 (mips_expand_vpc_loongson_even_odd): Likewise.
12267 (mips_expand_vec_unpack): Likewise.
12268 (mips_expand_vi_broadcast): Likewise.
12269 (mips_expand_vector_init): Likewise.
12270 (mips_expand_vec_reduc): Likewise.
12271 (mips_expand_msa_cmp): Likewise.
12272 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12273 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12274 (cc_flags_for_mode): Likewise.
12275 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12276 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12277 (nds32_output_casesi_pc_relative): Likewise.
12278 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12279 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12280 (nvptx_gen_unpack): Likewise.
12281 (nvptx_gen_pack): Likewise.
12282 (nvptx_gen_shuffle): Likewise.
12283 (nvptx_gen_wcast): Likewise.
12284 (nvptx_preferred_simd_mode): Likewise.
12285 * config/pa/pa.c (pa_secondary_reload): Likewise.
12286 * config/pa/predicates.md (base14_operand): Likewise.
12287 * config/powerpcspe/powerpcspe-c.c
12288 (altivec_resolve_overloaded_builtin): Likewise.
12289 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12291 (rs6000_preferred_simd_mode): Likewise.
12292 (output_vec_const_move): Likewise.
12293 (rs6000_expand_vector_extract): Likewise.
12294 (rs6000_split_vec_extract_var): Likewise.
12295 (reg_offset_addressing_ok_p): Likewise.
12296 (rs6000_legitimate_offset_address_p): Likewise.
12297 (rs6000_legitimize_address): Likewise.
12298 (rs6000_emit_set_const): Likewise.
12299 (rs6000_const_vec): Likewise.
12300 (rs6000_emit_move): Likewise.
12301 (spe_build_register_parallel): Likewise.
12302 (rs6000_darwin64_record_arg_recurse): Likewise.
12303 (swap_selector_for_mode): Likewise.
12304 (spe_init_builtins): Likewise.
12305 (paired_init_builtins): Likewise.
12306 (altivec_init_builtins): Likewise.
12307 (do_load_for_compare): Likewise.
12308 (rs6000_generate_compare): Likewise.
12309 (rs6000_expand_float128_convert): Likewise.
12310 (emit_load_locked): Likewise.
12311 (emit_store_conditional): Likewise.
12312 (rs6000_output_function_epilogue): Likewise.
12313 (rs6000_handle_altivec_attribute): Likewise.
12314 (rs6000_function_value): Likewise.
12315 (emit_fusion_gpr_load): Likewise.
12316 (emit_fusion_p9_load): Likewise.
12317 (emit_fusion_p9_store): Likewise.
12318 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12319 (fusion_gpr_mem_load): Likewise.
12320 (fusion_addis_mem_combo_load): Likewise.
12321 (fusion_addis_mem_combo_store): Likewise.
12322 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12323 (fusion_gpr_mem_load): Likewise.
12324 (fusion_addis_mem_combo_load): Likewise.
12325 (fusion_addis_mem_combo_store): Likewise.
12326 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12328 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12329 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12330 (rs6000_preferred_simd_mode): Likewise.
12331 (output_vec_const_move): Likewise.
12332 (rs6000_expand_vector_extract): Likewise.
12333 (rs6000_split_vec_extract_var): Likewise.
12334 (reg_offset_addressing_ok_p): Likewise.
12335 (rs6000_legitimate_offset_address_p): Likewise.
12336 (rs6000_legitimize_address): Likewise.
12337 (rs6000_emit_set_const): Likewise.
12338 (rs6000_const_vec): Likewise.
12339 (rs6000_emit_move): Likewise.
12340 (rs6000_darwin64_record_arg_recurse): Likewise.
12341 (swap_selector_for_mode): Likewise.
12342 (paired_init_builtins): Likewise.
12343 (altivec_init_builtins): Likewise.
12344 (rs6000_expand_float128_convert): Likewise.
12345 (emit_load_locked): Likewise.
12346 (emit_store_conditional): Likewise.
12347 (rs6000_output_function_epilogue): Likewise.
12348 (rs6000_handle_altivec_attribute): Likewise.
12349 (rs6000_function_value): Likewise.
12350 (emit_fusion_gpr_load): Likewise.
12351 (emit_fusion_p9_load): Likewise.
12352 (emit_fusion_p9_store): Likewise.
12353 * config/rx/rx.c (rx_gen_move_template): Likewise.
12354 (flags_from_mode): Likewise.
12355 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12356 (s390_slb_comparison): Likewise.
12357 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12358 (s390_vector_mode_supported_p): Likewise.
12359 (s390_cc_modes_compatible): Likewise.
12360 (s390_match_ccmode_set): Likewise.
12361 (s390_canonicalize_comparison): Likewise.
12362 (s390_emit_compare_and_swap): Likewise.
12363 (s390_branch_condition_mask): Likewise.
12364 (s390_rtx_costs): Likewise.
12365 (s390_secondary_reload): Likewise.
12366 (__SECONDARY_RELOAD_CASE): Likewise.
12367 (s390_expand_cs): Likewise.
12368 (s390_preferred_simd_mode): Likewise.
12369 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12370 * config/sh/sh.c (sh_print_operand): Likewise.
12371 (dump_table): Likewise.
12372 (sh_secondary_reload): Likewise.
12373 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12374 * config/sh/sh.md (casesi_worker_1): Likewise.
12375 (casesi_worker_2): Likewise.
12376 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12377 (fcc_comparison_operator): Likewise.
12378 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12379 (emit_soft_tfmode_cvt): Likewise.
12380 (sparc_preferred_simd_mode): Likewise.
12381 (output_cbranch): Likewise.
12382 (sparc_print_operand): Likewise.
12383 (sparc_expand_vec_perm_bmask): Likewise.
12384 (vector_init_bshuffle): Likewise.
12385 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12386 (spu_vector_mode_supported_p): Likewise.
12387 (spu_expand_insv): Likewise.
12388 (spu_emit_branch_or_set): Likewise.
12389 (spu_handle_vector_attribute): Likewise.
12390 (spu_builtin_splats): Likewise.
12391 (spu_builtin_extract): Likewise.
12392 (spu_builtin_promote): Likewise.
12393 (spu_expand_sign_extend): Likewise.
12394 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12395 (tilegx_simd_int): Likewise.
12396 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12397 (tilepro_simd_int): Likewise.
12398 * config/v850/v850.c (const_double_split): Likewise.
12399 (v850_print_operand): Likewise.
12400 (ep_memory_offset): Likewise.
12401 * config/vax/vax.c (vax_rtx_costs): Likewise.
12402 (vax_output_int_move): Likewise.
12403 (vax_output_int_add): Likewise.
12404 (vax_output_int_subtract): Likewise.
12405 * config/visium/predicates.md (visium_branch_operator): Likewise.
12406 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12407 (visium_print_operand_address): Likewise.
12408 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12409 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12410 (xtensa_expand_conditional_branch): Likewise.
12411 (xtensa_copy_incoming_a7): Likewise.
12412 (xtensa_output_literal): Likewise.
12413 * dfp.c (decimal_real_maxval): Likewise.
12414 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12417 Alan Hayward <alan.hayward@arm.com>
12418 David Sherwood <david.sherwood@arm.com>
12420 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12421 (mode_nunits_inline): Likewise.
12422 (mode_inner_inline): Likewise.
12423 (mode_unit_size_inline): Likewise.
12424 (mode_unit_precision_inline): Likewise.
12425 (emit_insn_modes_h): Likewise. Also emit a #define of the
12427 (emit_mode_wider): Add an E_ prefix to mode names.
12428 (emit_mode_complex): Likewise.
12429 (emit_mode_inner): Likewise.
12430 (emit_mode_adjustments): Likewise.
12431 (emit_mode_int_n): Likewise.
12432 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12433 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12434 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12435 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12436 (CRC32_BUILTIN, ENTRY): Likewise.
12437 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12438 (aarch64_pop_regs): Likewise.
12439 (aarch64_process_components): Likewise.
12440 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12441 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12442 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12443 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12444 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12445 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12447 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12448 (pcmpistr): Likewise.
12449 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12450 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12451 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12453 * config/rl78/rl78.c (mduc_regs): Likewise.
12454 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12455 (htm_expand_builtin): Likewise.
12456 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12457 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12458 * config/xtensa/xtensa.c (print_operand): Likewise.
12459 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12460 (NUM_MODE_VECTOR_INT): Likewise.
12461 * genoutput.c (null_operand): Likewise.
12462 (output_operand_data): Likewise.
12463 * genrecog.c (print_parameter_value): Likewise.
12464 * lra.c (debug_operand_data): Likewise.
12466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12468 * dwarf2out.c (scompare_loc_descriptor_wide)
12469 (scompare_loc_descriptor_narrow): New functions, split out from...
12470 (scompare_loc_descriptor): ...here.
12471 * expmed.c (emit_store_flag_int): New function, split out from...
12472 (emit_store_flag): ...here.
12474 2017-08-30 Richard Biener <rguenther@suse.de>
12476 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12477 (dwarf2out_early_finish): Move setting of AT_pubnames from
12478 early debug output to early finish.
12480 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12482 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12483 and -mdata-region to the assembler.
12485 2017-08-30 Richard Biener <rguenther@suse.de>
12487 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12489 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12490 (dwarf2out_early_global_decl): Only generate a DIE for the
12491 abstract origin if it doesn't already exist or is a declaration DIE.
12492 (resolve_addr): Do not add the linkage name twice when
12493 generating a stub DIE for the DW_TAG_GNU_call_site target.
12495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12497 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12498 Use machine_mode rather than int for arg1_mode.
12500 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12503 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12504 that the second argument of the built-in functions to unpack
12505 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12506 switch statement instead a lot of if statements.
12507 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12508 Allow 64-bit values to be in Altivec registers as well as
12509 traditional floating point registers.
12510 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12512 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12514 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12515 MAX_REGS_PER_ADDRESS == 1.
12517 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12519 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12520 * config/i386/i386.c (ix86_option_override_internal): Simplify
12521 setting of opts->x_flag_entry.
12523 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12524 Jakub Jelinek <jakub@redhat.com>
12525 Richard Biener <rguenther@suse.de>
12527 PR tree-optimization/81503
12528 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12529 folded constant fits in the target type; reorder tests for clarity.
12531 2017-08-29 Martin Liska <mliska@suse.cz>
12533 * passes.def: Include pass_lower_switch.
12534 * stmt.c (dump_case_nodes): Remove and move to
12535 tree-switch-conversion.
12536 (case_values_threshold): Likewise.
12537 (expand_switch_as_decision_tree_p): Likewise.
12538 (emit_case_decision_tree): Likewise.
12539 (expand_case): Likewise.
12540 (balance_case_nodes): Likewise.
12541 (node_has_low_bound): Likewise.
12542 (node_has_high_bound): Likewise.
12543 (node_is_bounded): Likewise.
12544 (emit_case_nodes): Likewise.
12545 (struct simple_case_node): New struct.
12546 (add_case_node): Remove.
12547 (emit_case_dispatch_table): Use vector instead of case_list.
12548 (reset_out_edges_aux): Remove.
12549 (compute_cases_per_edge): Likewise.
12550 (expand_case): Build list of simple_case_node.
12551 (expand_sjlj_dispatch_table): Use it.
12552 * tree-switch-conversion.c (struct case_node): Moved from
12553 stmt.c and adjusted.
12554 (emit_case_nodes): Likewise.
12555 (node_has_low_bound): Likewise.
12556 (node_has_high_bound): Likewise.
12557 (node_is_bounded): Likewise.
12558 (case_values_threshold): Likewise.
12559 (reset_out_edges_aux): Likewise.
12560 (compute_cases_per_edge): Likewise.
12561 (add_case_node): Likewise.
12562 (dump_case_nodes): Likewise.
12563 (balance_case_nodes): Likewise.
12564 (expand_switch_as_decision_tree_p): Likewise.
12565 (emit_jump): Likewise.
12566 (emit_case_decision_tree): Likewise.
12567 (try_switch_expansion): Likewise.
12568 (do_jump_if_equal): Likewise.
12569 (emit_cmp_and_jump_insns): Likewise.
12570 (fix_phi_operands_for_edge): New function.
12571 (record_phi_operand_mapping): Likewise.
12572 (class pass_lower_switch): New pass.
12573 (pass_lower_switch::execute): New function.
12574 (make_pass_lower_switch): Likewise.
12575 (conditional_probability):
12576 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12577 * tree-pass.h: Add make_pass_lower_switch.
12579 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12582 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12585 2017-08-29 Richard Biener <rguenther@suse.de>
12587 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12588 we do not add a DW_AT_inline attribute twice.
12589 (gen_subprogram_die): Remove code setting DW_AT_inline on
12590 DECL_ABSTRACT_P nodes.
12592 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12594 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12595 calls to internal functions.
12596 (gimplify_modify_expr): Likewise.
12597 * tree-call-cdce.c (use_internal_fn): Likewise.
12598 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12599 (convert_to_divmod): Set the nothrow flag.
12600 * tree-if-conv.c (predicate_mem_writes): Likewise.
12601 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12602 (vectorizable_call): Likewise.
12603 (vectorizable_store): Likewise.
12604 (vectorizable_load): Likewise.
12605 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12606 (vect_recog_mask_conversion_pattern): Likewise.
12608 2017-08-29 Martin Liska <mliska@suse.cz>
12611 * gcc.c (driver_handle_option): Add new argument.
12612 * opts-common.c (handle_option): Pass
12613 target_option_override_hook.
12614 * opts-global.c (lang_handle_option): Add new option.
12615 (set_default_handlers): Add new argument.
12616 (decode_options): Likewise.
12617 * opts.c (target_handle_option): Likewise.
12618 (common_handle_option): Call target_option_override_hook.
12619 * opts.h (struct cl_option_handler_func): Add hook for
12620 target option override.
12621 (struct cl_option_handlers): Likewise.
12622 (set_default_handlers): Add new argument.
12623 (decode_options): Likewise.
12624 (common_handle_option): Likewise.
12625 (target_handle_option): Likewise.
12626 * toplev.c (toplev::main): Pass targetm.target_option.override
12629 2017-08-29 Richard Biener <rguenther@suse.de>
12630 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12632 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12633 life to the active subtree.
12635 2017-08-28 Jeff Law <law@redhat.com>
12637 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12638 derive_equivalences.
12639 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12640 Code moved into....
12641 (edge_info::derive_equivalences): New private member function
12643 * tree-ssa-dom.c (class edge_info): Changed from a struct
12644 to a class. Add ctor/dtor, methods and data members.
12645 (edge_info::edge_info): Renamed from allocate_edge_info.
12646 Initialize additional members.
12647 (edge_info::~edge_info): New.
12648 (free_dom_edge_info): Delete the edge info.
12649 (record_edge_info): Use new class & associated member functions.
12650 Tighten forms for testing for edge equivalences.
12651 (record_temporary_equivalences): Iterate over the simple
12652 equivalences rather than assuming there's only one per edge.
12653 (cprop_into_successor_phis): Iterate over the simple
12654 equivalences rather than assuming there's only one per edge.
12655 (optimize_stmt): Use operand_equal_p rather than pointer
12656 equality for mini-DSE code.
12658 2017-08-28 Nathan Sidwell <nathan@acm.org>
12660 * gcc.c (execute): Fold SIGPIPE handling into switch
12661 statement. Adjust internal error message.
12663 2017-08-28 Richard Biener <rguenther@suse.de>
12666 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12667 Do nothing for removed DIEs.
12669 2017-08-28 Richard Biener <rguenther@suse.de>
12671 PR tree-optimization/81977
12672 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12675 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12678 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12679 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12680 using targetm.gen_mem_thread_fence.
12682 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12685 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12686 predicate to register_operand. Reorder operands.
12687 (*btr<mode>): Ditto.
12688 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12689 (*btr<mode>_mask): Ditto.
12691 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12693 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12694 * config/rs6000/xmmintrin.h: New file.
12695 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12697 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12700 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12701 parameter and_insn and return it.
12702 (recombine_lvx_pattern): Insert a copy to ensure availability of
12703 the base register of the copied masking operation at the point of
12704 the instruction replacement.
12705 (recombine_stvx_pattern): Likewise.
12707 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12709 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12710 undocumented switches.
12711 (-mpower9-dform-vector): Likewise.
12712 (-mpower9-dform): Likewise.
12713 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12714 comments to delete references to -mpower9-dform* switches.
12715 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12716 Delete reference to -mpower9-dform* switches, test for
12717 -mpower9-vector instead.
12718 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12719 (OTHER_P9_VECTOR_MASKS): Likewise.
12720 (POWERPC_MASKS): Likewise.
12721 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12722 tests against -mpower9-dform* to -mpower9-vector. Delete code
12723 that checked for -mpower9-dform* consistancy with other options.
12724 Add test for -mpower9-misc to enable other power9 switches.
12725 (rs6000_init_hard_regno_mode_ok): Likewise.
12726 (rs6000_option_override_internal): Likewise.
12727 (rs6000_emit_prologue): Likewise.
12728 (rs6000_emit_epilogue): Likewise.
12729 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12730 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12731 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12733 (emit_fusion_p9_store): Likewise.
12734 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12735 resetting these macros if the assembler does not support ISA 3.0
12737 (TARGET_P9_DFORM_VECTOR): Likewise.
12738 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12739 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12741 2017-08-25 Alan Modra <amodra@gmail.com>
12743 PR rtl-optimization/81747
12744 * cse.c (cse_extended_basic_block): Don't attempt to record
12745 equivalences for degenerate conditional branches that branch
12746 to their fall-through.
12748 2017-08-24 Martin Sebor <msebor@redhat.com>
12750 PR middle-end/81908
12751 * gimple-fold.c (size_must_be_zero_p): New function.
12752 (gimple_fold_builtin_memory_op): Call it.
12754 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12756 * config/rs6000/mm_malloc.h: New file.
12758 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12760 PR tree-optimization/81913
12761 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12762 analysis when either IVs in condition can wrap.
12764 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12766 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12767 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12769 2017-08-24 Richard Biener <rguenther@suse.de>
12772 * targhooks.c (default_target_can_inline_p): Properly
12773 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12774 is present and always compare.
12775 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12776 infer -mfpmath=sse from TARGET_SSE_P.
12777 (ix86_can_inline_p): Properly use target_option_default_node when
12778 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12780 2017-08-24 Richard Biener <rguenther@suse.de>
12783 * dwarf2out.c (output_die): Handle flag_generate_offload like
12785 (output_comp_unit): Likewise.
12786 (gen_array_type_die): Likewise.
12787 (dwarf2out_early_finish): Likewise.
12788 (note_variable_value_in_expr): Likewise.
12789 (dwarf2out_finish): Likewise. Adjust assert.
12790 * cgraphunit.c (symbol_table::compile): Move setting of
12791 flag_generate_offload earlier ...
12792 (symbol_table::finalize_compilation_unit): ... here, before
12793 early debug finalization.
12795 2017-08-24 Richard Biener <rguenther@suse.de>
12797 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12798 and ipa-fnsummary.h.
12799 (ix86_can_inline_p): When ix86_fpmath flags do not match
12800 check whether the callee uses FP math at all.
12802 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12804 PR middle-end/81931
12805 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12806 instead of TYPE_PRECISION.
12808 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12809 Alan Hayward <alan.hayward@arm.com>
12810 David Sherwood <david.sherwood@arm.com>
12812 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12814 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12816 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12818 (ix86_frame::stack_realign_allocate): New field.
12819 (struct machine_frame_state): Modify comments.
12820 (machine_frame_state::sp_realigned_fp_end): New field.
12821 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12822 layout calculation.
12823 (sp_valid_at): Add assertion to assure no attempt to access invalid
12824 offset of a realigned stack.
12825 (fp_valid_at): Likewise.
12826 (choose_baseaddr): Modify comments.
12827 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12828 ix86_expand_prologue.
12829 (ix86_expand_prologue): Modify stack realignment and allocation.
12830 (ix86_expand_epilogue): Modify comments.
12831 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12832 avx2_runtime, avx512f, and avx512f_runtime.
12834 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12836 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12837 (mstackrealign): Do not init to -1.
12838 * config/i386/i386.c (ix86_option_override_internal):
12839 Check opts_set, not opts when setting default value of
12840 opts->x_ix86_force_align_arg_pointer.
12842 2017-08-23 Richard Biener <rguenther@suse.de>
12844 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12845 lang_hooks.decl_printable_name.
12846 * print-rtl-function.c (print_rtx_function): Likewise.
12847 * tree-pretty-print.c (dump_function_header): Likewise.
12849 2017-08-23 Richard Biener <rguenther@suse.de>
12852 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12853 -g0 at compile-time.
12855 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12857 PR middle-end/19706
12858 * doc/sourcebuild.texi (Other hardware attributes):
12861 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12863 PR middle-end/19706
12864 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12865 Fix single-use check.
12867 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12869 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12871 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12873 * doc/install.texi: Modify to add more details on running selected
12876 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12878 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12879 is combined with -mabi=ms.
12880 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12883 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12885 PR tree-optimization/81488
12886 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12887 and cached_basis fields.
12888 (MAX_SPREAD): New constant.
12889 (alloc_cand_and_find_basis): Initialize new fields.
12890 (clear_visited): New function.
12891 (create_phi_basis_1): Rename from create_phi_basis, set visited
12892 and cached_basis fields.
12893 (create_phi_basis): New wrapper function.
12894 (phi_add_costs_1): Rename from phi_add_costs, add spread
12895 parameter, set visited field, short-circuit when limits reached.
12896 (phi_add_costs): New wrapper function.
12897 (record_phi_increments_1): Rename from record_phi_increments, set
12899 (record_phi_increments): New wrapper function.
12900 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12901 (phi_incr_cost): New wrapper function.
12902 (all_phi_incrs_profitable_1): Rename from
12903 all_phi_incrs_profitable, set visited field.
12904 (all_phi_incrs_profitable): New wrapper function.
12906 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12907 Alan Hayward <alan.hayward@arm.com>
12908 David Sherwood <david.sherwood@arm.com>
12910 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12911 that takes the outer and inner modes.
12912 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12913 comparison as the canonical test for a paradoxical subreg.
12914 * combine.c (simplify_set): Use paradoxical_subreg_p.
12915 (make_extraction): Likewise.
12916 (force_to_mode): Likewise.
12917 (rtx_equal_for_field_assignment_p): Likewise.
12918 (gen_lowpart_for_combine): Likewise.
12919 (simplify_comparison): Likewise.
12920 * cse.c (equiv_constant): Likewise.
12921 * expmed.c (store_bit_field_1): Likewise.
12922 * final.c (alter_subreg): Likewise.
12923 * fwprop.c (propagate_rtx): Likewise.
12924 (forward_propagate_subreg): Likewise.
12925 * ira-conflicts.c (ira_build_conflicts): Likewise.
12926 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12927 * lra-constraints.c (curr_insn_transform): Likewise.
12928 (split_reg): Likewise.
12929 * lra-eliminations.c (move_plus_up): Likewise.
12930 (lra_eliminate_regs_1): Likewise.
12931 * recog.c (general_operand): Likewise.
12932 * ree.c (combine_reaching_defs): Likewise.
12933 * reload.c (push_reload): Likewise.
12934 (find_reloads): Likewise.
12935 * reload1.c (elimination_effects): Likewise.
12936 (compute_reload_subreg_offset): Likewise.
12937 (choose_reload_regs): Likewise.
12938 * rtlanal.c (subreg_lsb_1): Likewise.
12939 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12940 (simplify_subreg): Likewise.
12941 * var-tracking.c (track_loc_p): Likewise.
12942 * emit-rtl.c (byte_lowpart_offset): Likewise.
12943 (paradoxical_subreg_p): Delete out-of-line definition.
12945 2017-08-22 Jeff Law <law@redhat.com>
12947 PR tree-optimization/81741
12948 PR tree-optimization/71947
12949 * tree-ssa-dom.c: Include tree-inline.h.
12950 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12951 equivalences if one is more expensive to compute than the other.
12952 * tree-ssa-scopedtables.h (class const_or_copies): Make
12953 record_const_or_copy_raw method private.
12954 (class avail_exprs_stack): New method simplify_binary_operation.
12955 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12956 avail_exprs_stack::simplify_binary_operation as needed.
12957 (avail_exprs_stack::simplify_binary_operation): New function.
12959 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12961 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12962 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12963 (DOT_SYMBOLS): Likewise.
12964 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12965 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12966 (RS6000_ABI_NAME): Likewise.
12967 (TARGET_CMODEL): Likewise.
12968 (TOC_SECTION_ASM_OP): Likewise.
12969 (SET_CMODEL): New macro.
12970 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12972 2017-08-22 Richard Biener <rguenther@suse.de>
12974 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12975 to allow for free-lang-data replacements similar to verify_type_variant.
12977 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12978 Alan Hayward <alan.hayward@arm.com>
12979 David Sherwood <david.sherwood@arm.com>
12981 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12982 VOIDmode for the LABEL_REF.
12984 2017-08-22 Richard Biener <rguenther@suse.de>
12986 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12988 2017-08-22 Richard Biener <rguenther@suse.de>
12990 * common.opt (feliminate-dwarf2-dups): Ignore.
12991 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12992 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12993 same_die_p_wrap, compute_section_prefix,
12994 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12995 (comdat_symbol_id, comdat_symbol_number): Likewise.
12996 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12998 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12999 (output_die): Mark unreachable path unreachable.
13000 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13001 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13002 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13003 (dwarf2out_early_finish): Likewise.
13005 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13007 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13009 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13012 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13013 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13014 (avr_attribute_table) <io, io_low, address>: Initialize
13015 .decl_required with true.
13017 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13019 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13020 undocumented debugging options.
13021 (-mvsx-scalar-double): Likewise.
13022 (-mallow-df-permute): Likewise.
13023 (-mvectorize-builtins): Likewise.
13024 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13025 (rs6000_builtin_vectorized_function): Likewise.
13026 (rs6000_builtin_md_vectorized_function): Likewise.
13027 (rs6000_opt_vars): Likewise.
13029 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13032 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13033 (*btrq_imm): Rename from *btrq.
13034 (*btcq_imm): Rename from *btcq.
13035 (btsc): New code attribute.
13036 (*<btsc><mode>): New insn pattern.
13037 (*btr<mode>): Ditto.
13038 (*<btsc><mode>_mask): New insn_and_split pattern.
13039 (*btr<mode>_mask): Ditto.
13041 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13042 Alan Hayward <alan.hayward@arm.com>
13043 David Sherwood <david.sherwood@arm.com>
13045 * function.c (pad_below): Simplify padding calculation.
13047 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13048 Alan Hayward <alan.hayward@arm.com>
13049 David Sherwood <david.sherwood@arm.com>
13051 * target.def (function_prologue): Remove frame size argument.
13052 (function_epilogue): Likewise.
13053 * doc/tm.texi: Regenerate.
13054 * final.c (final_start_function): Update call to function_prologue.
13055 (final_end_function): Update call to function_epilogue.
13056 (default_function_pro_epilogue): Remove frame size argument.
13057 * output.h (default_function_pro_epilogue): Likewise.
13058 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13059 (arm_output_function_prologue): Likewise.
13060 * config/frv/frv.c (frv_function_prologue): Likewise.
13061 (frv_function_epilogue): Likewise.
13062 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13063 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13064 (ia64_output_function_epilogue): Likewise.
13065 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13066 (m32r_output_function_epilogue): Likewise.
13067 * config/microblaze/microblaze.c (microblaze_function_prologue)
13068 (microblaze_function_epilogue): Likewise.
13069 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13070 (mips_output_function_epilogue): Likewise.
13071 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13072 (mmix_target_asm_function_epilogue): Likewise.
13073 * config/msp430/msp430.c (msp430_start_function): Likewise.
13074 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13075 (nds32_asm_function_epilogue): Likewise.
13076 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13077 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13078 (pa_output_function_epilogue): Likewise.
13079 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13080 (rs6000_output_function_epilogue): Likewise.
13081 * config/rl78/rl78.c (rl78_start_function): Likewise.
13082 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13083 (rs6000_output_function_epilogue): Likewise.
13084 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13085 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13086 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13087 (sparc_asm_function_epilogue): Likewise.
13089 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13091 * tree.h (type_has_mode_precision_p): New function.
13092 * convert.c (convert_to_integer_1): Use it.
13093 * expr.c (expand_expr_real_2): Likewise.
13094 (expand_expr_real_1): Likewise.
13095 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13096 * match.pd: Likewise.
13097 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13098 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13099 * tree-tailcall.c (process_assignment): Likewise.
13100 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13101 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13102 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13103 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13104 (vectorizable_assignment): Likewise.
13105 (vectorizable_shift): Likewise.
13106 (vectorizable_operation): Likewise.
13107 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13109 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13111 * match.pd: Add pow (C, x) simplification.
13113 2017-08-21 Richard Biener <rguenther@suse.de>
13115 PR tree-optimization/81900
13116 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13117 for blocks with abnormal predecessors.
13118 (compute_antic): Do not set visited flag prematurely.
13120 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13123 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13125 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13127 * stor-layout.h (vector_type_mode): Move to...
13128 * tree.h (vector_type_mode): ...here.
13129 * stor-layout.c (vector_type_mode): Move to...
13130 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13132 2017-08-21 Richard Biener <rguenther@suse.de>
13134 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13135 register_external_die hooks.
13136 (debug_false_tree_charstarstar_uhwistar): Declare.
13137 (debug_nothing_tree_charstar_uhwi): Likewise.
13138 * debug.c (do_nothing_debug_hooks): Adjust.
13139 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13140 (debug_nothing_tree_charstar_uhwi): Likewise.
13141 * dbxout.c (dbx_debug_hooks): Adjust.
13142 (xcoff_debug_hooks): Likewise.
13143 * sdbout.c (sdb_debug_hooks): Likewise.
13144 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13145 * dwarf2out.c (macinfo_label_base): New global.
13146 (dwarf2out_register_external_die): New function for the
13147 register_external_die hook.
13148 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13149 (dwarf2_debug_hooks): Use them.
13150 (dwarf2_lineno_debug_hooks): Adjust.
13151 (struct die_struct): Add with_offset flag.
13152 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13153 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13154 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13155 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13156 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13157 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13158 defining section names for the early LTO debug variants.
13159 (reset_indirect_string): New helper.
13160 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13161 (print_dw_val): Add support for offsetted symbol references.
13162 (get_ultimate_context): Split out from is_cxx.
13163 (is_cxx): Use get_ultimate_context.
13164 (is_fortran): Add decl overload.
13165 (compute_comp_unit_symbol): Split out worker from
13166 compute_section_prefix.
13167 (compute_section_prefix): Call compute_comp_unit_symbol and
13168 set comdat_type_p here.
13169 (output_die): Skip DIE symbol output for the LTO added one.
13170 Handle DIE symbol references with offset.
13171 (output_comp_unit): Guard section name mangling properly.
13172 For LTO debug sections emit a symbol at the section beginning
13173 which we use to refer to its DIEs.
13174 (add_abstract_origin_attribute): For DIEs registered via
13175 dwarf2out_register_external_die directly refer to the early
13176 DIE rather than indirectly through the shadow one we created.
13177 Remove obsolete call to dwarf2out_abstract_function for
13178 non-function/block origins.
13179 (gen_array_type_die): When generating early LTO debug do
13180 not emit DW_AT_string_length.
13181 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13182 late when in LTO. As suggested place a gcc_unreachable for
13183 the DECL_ABSTRACT_P case.
13184 (gen_subprogram_die): Avoid another specification DIE
13185 for early built declarations/definitions for the late LTO case.
13186 (gen_variable_die): Add type references for late duplicated VLA dies
13188 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13189 we have the abstract instance already.
13190 (process_scope_var): Adjust decl DIE contexts in LTO which
13191 first puts them in limbo.
13192 (gen_decl_die): Do not generate type DIEs late apart from
13193 types for VLAs or for decls we do not yet have a DIE. Do not
13194 call dwarf2out_abstract_function late.
13195 (dwarf2out_early_global_decl): Make sure to create DIEs
13196 for abstract instances of a decl first.
13197 (dwarf2out_late_global_decl): Adjust comment.
13198 (output_macinfo_op): With multiple macro sections use
13199 macinfo_label_base to distinguish labels.
13200 (output_macinfo): Likewise. Update macinfo_label_base.
13201 Pass in the line info label.
13202 (note_variable_value_in_expr): When generating LTO resolve
13203 all variable values here by generating DIEs as needed.
13204 (init_sections_and_labels): Add early LTO debug flag parameter
13205 and generate different sections and names if set. Add generation
13206 counter for the labels so we can have multiple of them.
13207 (reset_dies): Helper to allow DIEs to be output multiple times.
13208 (dwarf2out_finish): When outputting DIEs to the fat part of an
13209 LTO object first reset DIEs.
13210 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13211 (modified_type_die): Check for decl_ultimate_origin being self
13213 (gen_type_die_with_usage): Likewise.
13214 (gen_typedef_die): Allow decl_ultimate_origin being self.
13215 (set_decl_abstract_flags): Remove.
13216 (set_block_abstract_flags): Likewise.
13217 (dwarf2out_abstract_function): Treat the early generated DIEs
13218 as the abstract copy and only add DW_AT_inline and
13219 DW_AT_artificial here and call set_decl_origin_self.
13220 If the DIE has an abstract origin don't do anything.
13221 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13222 if we have none yet (Go fails to build one, PR78628).
13223 (variably_modified_type_p): Prevent endless recursion for Ada
13224 cyclic pointer types.
13225 * lto-streamer-in.c: Include debug.h.
13226 (dref_queue): New global.
13227 (lto_read_tree_1): Stream in DIE references.
13228 (lto_input_tree): Register DIE references.
13229 (input_function): Stream DECL_DEBUG_ARGS.
13230 * lto-streamer-out.c: Include debug.h.
13231 (lto_write_tree_1): Output DIE references.
13232 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13233 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13234 (output_function): Stream DECL_DEBUG_ARGS.
13235 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13236 Stream DECL_ABSTRACT_ORIGIN.
13237 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13238 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13239 DECL_CONTEXT for file-scope decls.
13240 * lto-streamer.h (struct dref_entry): Declare.
13241 (dref_queue): Likewise.
13242 * cfgexpand.c (pass_expand::execute): Do not call the
13243 outlining_inline_function hook here.
13244 * lto-wrapper.c (debug_obj): New global.
13245 (tool_cleanup): Unlink it if required.
13246 (debug_objcopy): New function.
13247 (run_gcc): Handle early debug sections in the IL files by
13248 extracting them to separate files, partially linkin them and
13249 feeding the result back as result to the linker.
13250 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13251 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13252 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13253 sections into a separate segment.
13254 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13256 (darwin_asm_dwarf_section): Likewise.
13257 (darwin_asm_output_dwarf_offset): Likewise.
13258 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13260 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13261 Alan Hayward <alan.hayward@arm.com>
13262 David Sherwood <david.sherwood@arm.com>
13264 * read-md.h (md_reader::record_potential_iterator_use): Replace
13265 pointer argument with an rtx and an index.
13266 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13267 (apply_mode_iterator): Likewise.
13268 (apply_code_iterator): Likewise.
13269 (apply_int_iterator): Likewise.
13270 (apply_subst_iterator): Likewise.
13271 (record_iterator_use): Likewise.
13272 (record_attribute_use): Likewise.
13273 (md_reader::record_potential_iterator_use): Likewise. Update calls
13274 to record_iterator_use and apply_iterator.
13275 (iterator_use): Replace ptr with x and index.
13276 (attribute_use): Likewise.
13277 (apply_attribute_uses): Update calls to apply_iterator.
13278 (apply_iterators): Likewise. Update initialization of iterator_use.
13279 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13280 and record_potential_iterator_use.
13281 (rtx_reader::read_rtx_operand): Likewise.
13283 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13284 Alan Hayward <alan.hayward@arm.com>
13285 David Sherwood <david.sherwood@arm.com>
13287 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13290 2017-08-21 Richard Biener <rguenther@suse.de>
13292 PR middle-end/81884
13293 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13294 at struct end conservatively when comparing common bases.
13296 2017-08-21 Richard Biener <rguenther@suse.de>
13298 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13299 (mem_ref_in_stmt): Remove.
13300 (determine_max_movement): Use ref index to get at the reference.
13301 (invariantness_dom_walker::before_dom_children): Deal with
13302 lim data already initialized.
13303 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13305 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13307 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13308 (TARGET_ISA_ROUND): Ditto.
13309 (TARGET_ROUND): Ditto.
13310 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13311 * config/i386/i386.md: Ditto.
13312 * config/i386/sse.md: Ditto.
13313 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13314 with OPTION_MASK_ISA_SSE4_1.
13316 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13319 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13320 __builtin_ia32_lzcnt_u16.
13322 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13325 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13326 (rs6000_set_current_function): Rewrite function to use it.
13328 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13331 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13332 and TYPE_WARN_IF_NOT_ALIGN.
13333 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13334 (handle_warn_if_not_align): New.
13335 (place_union_field): Call handle_warn_if_not_align.
13336 (place_field): Call handle_warn_if_not_align.
13337 Copy TYPE_WARN_IF_NOT_ALIGN.
13338 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13339 (layout_type): Likewise.
13340 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13342 (tree_decl_common): Add warn_if_not_align.
13343 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13344 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13345 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13346 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13347 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13348 * doc/extend.texi: Document warn_if_not_aligned attribute.
13349 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13351 2017-08-17 Martin Liska <mliska@suse.cz>
13354 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13355 (get_data_dependence): Use it as pointer type.
13356 (distribute_loop): Likewise.
13358 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13360 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13361 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13362 (p8_vmrgow_<mode>_direct): New define_insn.
13363 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13364 handle endianness for vmrgew and vmrgow permute patterns.
13366 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13368 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13369 * config/rs6000/rs6000-cpus.def: Remove comment.
13370 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13371 (POWERPC_MASKS): Likewise.
13372 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13373 use of TARGET_VSX_TIMODE.
13374 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13375 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13376 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13377 (rs6000_option_override_internal): Remove dead code.
13378 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13379 (rs6000_legitimize_reload_address): Likewise.
13380 (rs6000_legitimate_address_p): Likewise.
13381 (rs6000_opt_masks): Delete "vsx-timode".
13382 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13383 from function comment.
13384 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13385 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13386 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13388 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13389 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13390 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13391 (VSX_TI): Likewise.
13393 (define_peephole2): Likewise.
13395 2017-08-17 Martin Sebor <msebor@redhat.com>
13398 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13399 past the end of an array.
13400 (test_pp_format): Add test cases.
13402 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13404 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13405 missing ECF_NOTHROW flags.
13407 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13410 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13411 operands residing in integer registers.
13412 (*vsx_le_perm_load_<mode>): Likewise.
13413 (*vsx_le_perm_store_<mode>): Likewise.
13414 (define_peephole2): Add peepholes to optimize the above.
13416 2017-08-17 Marek Polacek <polacek@redhat.com>
13418 PR middle-end/81814
13419 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13420 to mimic what shorten_compare did. Change the return type to bool.
13421 (fold_cond_expr_with_comparison): Update call to
13422 operand_equal_for_comparison_p.
13423 (fold_ternary_loc): Likewise.
13425 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13427 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13429 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13431 2017-08-17 Richard Biener <rguenther@suse.de>
13433 * tree-ssa-structalias.c (solve_graph): When propagating
13434 to successors update the graphs succ edges and avoid duplicate work.
13436 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13439 * config/i386/i386.c (ix86_option_override_internal): Save target
13440 specific options after ix86_stack_protector_guard_reg was changed.
13442 2017-08-17 Richard Biener <rguenther@suse.de>
13444 PR tree-optimization/81827
13445 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13446 (new_var_info): Initialize it conservatively.
13447 (get_call_vi): Mark register vars.
13448 (new_scalar_tmp_constraint_exp): Likewise.
13449 (handle_rhs_call): Likewise.
13450 (handle_const_call): Likewise.
13451 (create_function_info_for): Likewise.
13452 (solve_constraints): Sort varinfos to separate register from
13453 non-register vars to pack points-to solution bitmaps during
13456 2017-08-17 Marek Polacek <polacek@redhat.com>
13458 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13460 2017-08-17 Richard Biener <rguenther@suse.de>
13462 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13463 to true when overflow is undefined and we saturated the result.
13465 2017-08-17 Alan Modra <amodra@gmail.com>
13468 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13469 Don't use store multiple if only one reg needs saving.
13470 (interesting_frame_related_regno): New function.
13471 (rs6000_frame_related): Don't emit frame info for regs that
13473 (rs6000_emit_epilogue): Likewise.
13475 2017-08-16 Nathan Sidwell <nathan@acm.org>
13477 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13478 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13479 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13480 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13481 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13482 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13483 (verify_type): Adjust for TYPE_BINFO move.
13484 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13485 process TYPE_BINFO directly.
13486 (hash_tree): Likewise.
13487 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13489 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13492 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13494 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13496 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13499 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13500 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13501 Add ix86_binary_operator_ok to insn constraint.
13502 (*iordi_1_bts): Ditto.
13503 (*xordi_1_btc): Ditto.
13504 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13505 Update corresponding peephole2 pattern.
13509 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13511 PR tree-optimization/81832
13512 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13513 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13515 2017-08-16 Marek Polacek <polacek@redhat.com>
13518 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13519 regarding size_zero_node.
13521 2017-08-16 Martin Liska <mliska@suse.cz>
13524 * config.gcc: Respect previously set extra_objs in case
13527 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13529 PR tree-optimization/81835
13530 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13531 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13532 not depend on the phi.
13534 2017-08-16 Alan Modra <amodra@gmail.com>
13536 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13539 2017-08-16 Alan Modra <amodra@gmail.com>
13541 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13542 (save_reg_p): ..into this. Update all callers.
13543 (first_reg_to_save): Simplify.
13545 2017-08-16 Alan Modra <amodra@gmail.com>
13547 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13550 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13554 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13555 constant count if that count is less than 32.
13557 2017-08-15 Nathan Sidwell <nathan@acm.org>
13559 * gcc.c (execute): Emit friendlier message if inferior is killed
13560 by an external cause.
13562 2017-08-15 Richard Biener <rguenther@suse.de>
13564 PR tree-optimization/81790
13565 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13566 CONSTRUCTORs from simplifying and VN.
13568 2017-08-14 Martin Sebor <msebor@redhat.com>
13570 * builtin-attrs.def: Add comments.
13572 2017-08-14 Martin Sebor <msebor@redhat.com>
13575 * doc/extend.texi (attribute nonstring): Document new attribute.
13577 2017-08-14 Martin Sebor <msebor@redhat.com>
13580 * tree-diagnostic.c (default_tree_printer): Handle %G.
13581 * gimple-pretty-print.h (percent_G_format): Declare new function.
13582 * gimple-pretty-print.c (percent_G_format): Define.
13583 * tree-pretty-print.c (percent_K_format): Add argument.
13585 2017-08-14 Martin Sebor <msebor@redhat.com>
13587 PR translation/79998
13588 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13589 Remove a stray space.
13591 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13594 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13595 (*iordi_1_bts): Ditto.
13596 (*xordi_1_btc): Ditto.
13598 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13601 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13604 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13605 quoted strings, and make more translator-friendly.
13606 (darwin_rs6000_override_options): Likewise.
13607 (rs6000_option_override_internal): Likewise.
13608 (rs6000_return_in_memory): Fix overlong line.
13609 (init_cmulative_args): Use quoted strings, and make more
13610 translator-friendly.
13611 (rs6000_pass_by_reference): Fix overlong line.
13612 (def_builtin): Use quoted strings.
13613 (altivec_expand_predicate_builtin): Use quoted strings, and make
13614 more translator-friendly.
13615 (htm_expand_builtin): Use quoted strings.
13616 (cpu_expand_builtin): Use quoted strings, and make more
13617 translator-friendly.
13618 (altivec_expand_builtin): Likewise.
13619 (paired_expand_predicate_builtin): Likewise.
13620 (rs6000_invalid_builtin): Likewise.
13621 (builtin_function_type): Use quoted strings.
13622 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13623 more translator-friendly.
13624 (rs6000_trampoline_init): Likewise.
13625 (rs6000_handle_altivec_attribute): Likewise.
13626 (rs6000_inner_target_options): Use quoted strings.
13627 (rs6000_disable_incompatible_switches): Likewise.
13628 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13629 strings, and make more translator-friendly.
13630 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13632 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13634 PR tree-optimization/81799
13635 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13636 cond_expr to simple gimple operand.
13638 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13640 PR middle-end/46932
13641 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13643 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13647 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13648 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13649 TARGET_GASISR_PROLOGUES.
13650 * config/avr/avr.c (avr_option_override): Same.
13651 (avr_pass_pre_proep::execute): Same.
13653 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13656 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13657 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13659 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13661 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13662 define_insn_and_split. Split to a memory load from 0 in
13663 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13664 using PTR mode iterator.
13665 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13666 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13667 (*add_tp_<mode>): Redefine as define_insn_and_split.
13668 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13669 address space. Merge with *add_tp_x32 using PTR mode iterator.
13670 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13671 Split to an add with a memory load from 0 in
13672 DEFAULT_TLS_SEG_REG address space.
13674 2017-08-12 Andrew Pinski <apinski@cavium.com>
13676 * config/aarch64/aarch64-option-extensions.def (rdma):
13677 Fix feature string to what Linux prints out in /proc/cpuinfo.
13679 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13682 * dwarf2out.c (modified_type_die): For C typedef types that have
13683 an ultimate origin, process the ultimate origin instead of the
13685 (gen_typedef_die): Assert that input DECLs have no ultimate
13687 (gen_type_die_with_usage): For typedef variants that have an
13688 ultimate origin, just call gen_decl_die on the original DECL.
13689 (process_scope_var): Avoid creating DIEs for local typedefs and
13690 concrete static variables.
13692 2017-08-12 Alan Modra <amodra@gmail.com>
13696 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13697 match gnu-user.h startfile.
13698 (ENDFILE_LINUX_SPEC): Similarly.
13700 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13703 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13705 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13707 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13708 * config/aarch64/aarch64.md (mov<mode>): Change.
13709 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13710 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13711 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13713 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13715 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13716 for storage order barriers.
13718 2017-08-11 Martin Liska <mliska@suse.cz>
13721 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13722 variables of void type.
13724 2017-08-11 Martin Liska <mliska@suse.cz>
13726 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13727 TARGET_SUPPORTS_ALIASES.
13728 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13729 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13730 (optimize_weakref): Likewise.
13731 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13732 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13733 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13735 2017-08-11 Martin Liska <mliska@suse.cz>
13738 * config/i386/i386.c (make_resolver_func): Do complete
13739 refactoring of the function.
13741 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13744 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13745 * config/i386/i386.c (ix86_stack_protect_guard): Use
13746 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13747 * doc/invoke.texi (x86 Options): Document
13748 -mstack-protector-guard-symbol= option.
13750 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13752 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13753 * config/i386/i386.c (ix86_split_stack_guard): New function.
13754 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13755 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13756 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13757 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13758 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13759 (split_stack_space_check): Call ix86_split_stack_guard.
13761 2017-08-10 Martin Sebor <msebor@redhat.com>
13763 * print-tree.c (print_node): Print location using the established
13765 Replace spaces with colons.
13766 (debug_raw, debug): Ditto.
13768 2017-08-10 Martin Sebor <msebor@redhat.com>
13771 * pretty-print.c (pp_format): Correct the handling of %s precision.
13773 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13776 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13778 (ix86_finalize_stack_frame_flags): This. Also clear
13779 frame_pointer_needed if -fno-omit-frame-pointer is used without
13781 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13782 with ix86_finalize_stack_frame_flags.
13783 (ix86_expand_epilogue): Likewise.
13784 (ix86_expand_split_stack_prologue): Likewise.
13785 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13787 2017-08-10 Martin Liska <mliska@suse.cz>
13790 * c-attribs.c (handle_target_attribute):
13791 Report warning for an empty string argument of target attribute.
13793 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13796 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13798 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13799 or DECL_NONLOCAL labels.
13800 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13801 or DECL_NONLOCAL labels here.
13803 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13806 to indicate when early gimple folding has been disabled.
13807 (rs6000_gimple_fold_builtin): Add debug content.
13808 (rs6000_invalid_builtin): Fix whitespace.
13809 (rs6000_expand_builtin): Fix whitespace.
13810 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13812 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13815 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13816 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13818 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13820 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13821 (qdf24xx): Likewise.
13822 * config/aarch64/aarch64.md: Include falkor.md.
13823 * config/aarch64/falkor.md: New.
13825 2017-08-09 Marek Polacek <polacek@redhat.com>
13828 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13829 * diagnostic.c (emit_diagnostic): Add a comment.
13830 (emit_diagnostic_valist): New function.
13832 2017-08-09 Marek Polacek <polacek@redhat.com>
13835 * input.c (make_location): New overload.
13836 * input.h (make_location): Declare.
13838 2017-08-08 Alan Modra <amodra@gmail.com>
13839 H.J. Lu <hongjiu.lu@intel.com>
13842 * gcc.c (NO_PIE_SPEC): Delete.
13843 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13845 (LINK_PIE_SPEC): ..to here.
13846 (LINK_COMMAND_SPEC): Support -no-pie.
13847 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13848 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13850 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13851 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13852 (ENDFILE_CRTEND_SPEC): Similarly.
13854 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13857 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13858 (mstack-protector-guard-offset=): Ditto.
13859 * config/i386/i386.c (ix86_option_override): Handle
13860 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13862 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13863 ix86_stack_protect_guard_offset variables.
13864 (TARGET_STACK_PROTECT_GUARD): Always define.
13865 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13866 and -mstack-protector-guard-offset= options.
13868 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13870 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13871 boundary case for the last candidate.
13873 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13875 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13876 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13878 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13880 PR middle-end/19706
13881 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13882 * config/aarch64/aarch64-builtins.c
13883 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13884 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13885 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13887 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13888 Andrew Pinski <pinskia@gmail.com>
13890 PR middle-end/19706
13891 * internal-fn.def (XORSIGN): New.
13892 * optabs.def (xorsign_optab): New.
13893 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13894 (convert_expand_mult_copysign): New.
13895 (pass_optimize_widening_mul::execute): Call
13896 convert_expand_mult_copysign.
13898 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13900 PR tree-optimization/81354
13901 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13902 Insert on edges rather than explicitly creating landing pads.
13903 (analyze_candidates_and_replace): Commit edge inserts.
13905 2017-08-08 Richard Biener <rguenther@suse.de>
13907 PR middle-end/81719
13908 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13909 (expand_simple_operations): Also look through ADDR_EXPRs with
13910 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13912 2017-08-08 Richard Biener <rguenther@suse.de>
13914 PR tree-optimization/81723
13915 * tree-vect-slp.c (struct bst_traits): New hash traits.
13916 (bst_fail): New global.
13917 (vect_build_slp_tree_2): New worker, split out from ...
13918 (vect_build_slp_tree): ... this now wrapping it with using
13919 bst_fail set to cache SLP tree build fails. Properly handle
13921 (vect_analyze_slp_instance): Allocate and free bst_fail.
13923 2017-08-08 Martin Liska <mliska@suse.cz>
13926 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13927 LABEL_DECLs that can be from a different function.
13929 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13931 PR tree-optimization/81744
13932 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13933 loop's number of iterations.
13935 2017-08-08 Martin Liska <mliska@suse.cz>
13937 * asan.c: Include header files.
13938 * attribs.c (build_decl_attribute_variant): New function moved
13940 (build_type_attribute_qual_variant): Likewise.
13941 (cmp_attrib_identifiers): Likewise.
13942 (simple_cst_list_equal): Likewise.
13943 (omp_declare_simd_clauses_equal): Likewise.
13944 (attribute_value_equal): Likewise.
13945 (comp_type_attributes): Likewise.
13946 (build_type_attribute_variant): Likewise.
13947 (lookup_ident_attribute): Likewise.
13948 (remove_attribute): Likewise.
13949 (merge_attributes): Likewise.
13950 (merge_type_attributes): Likewise.
13951 (merge_decl_attributes): Likewise.
13952 (merge_dllimport_decl_attributes): Likewise.
13953 (handle_dll_attribute): Likewise.
13954 (attribute_list_equal): Likewise.
13955 (attribute_list_contained): Likewise.
13956 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13957 (lookup_attribute_by_prefix): Likewise.
13958 * bb-reorder.c: Include header files.
13959 * builtins.c: Likewise.
13960 * calls.c: Likewise.
13961 * cfgexpand.c: Likewise.
13962 * cgraph.c: Likewise.
13963 * cgraphunit.c: Likewise.
13964 * convert.c: Likewise.
13965 * dwarf2out.c: Likewise.
13966 * final.c: Likewise.
13967 * fold-const.c: Likewise.
13968 * function.c: Likewise.
13969 * gimple-expr.c: Likewise.
13970 * gimple-fold.c: Likewise.
13971 * gimple-pretty-print.c: Likewise.
13972 * gimple.c: Likewise.
13973 * gimplify.c: Likewise.
13974 * hsa-common.c: Likewise.
13975 * hsa-gen.c: Likewise.
13976 * internal-fn.c: Likewise.
13977 * ipa-chkp.c: Likewise.
13978 * ipa-cp.c: Likewise.
13979 * ipa-devirt.c: Likewise.
13980 * ipa-fnsummary.c: Likewise.
13981 * ipa-inline.c: Likewise.
13982 * ipa-visibility.c: Likewise.
13984 * lto-cgraph.c: Likewise.
13985 * omp-expand.c: Likewise.
13986 * omp-general.c: Likewise.
13987 * omp-low.c: Likewise.
13988 * omp-offload.c: Likewise.
13989 * omp-simd-clone.c: Likewise.
13990 * opts-global.c: Likewise.
13991 * passes.c: Likewise.
13992 * predict.c: Likewise.
13993 * sancov.c: Likewise.
13994 * sanopt.c: Likewise.
13995 * symtab.c: Likewise.
13996 * toplev.c: Likewise.
13997 * trans-mem.c: Likewise.
13998 * tree-chkp.c: Likewise.
13999 * tree-eh.c: Likewise.
14000 * tree-into-ssa.c: Likewise.
14001 * tree-object-size.c: Likewise.
14002 * tree-parloops.c: Likewise.
14003 * tree-profile.c: Likewise.
14004 * tree-ssa-ccp.c: Likewise.
14005 * tree-ssa-live.c: Likewise.
14006 * tree-ssa-loop.c: Likewise.
14007 * tree-ssa-sccvn.c: Likewise.
14008 * tree-ssa-structalias.c: Likewise.
14009 * tree-ssa.c: Likewise.
14010 * tree-streamer-in.c: Likewise.
14011 * tree-vectorizer.c: Likewise.
14012 * tree-vrp.c: Likewise.
14013 * tsan.c: Likewise.
14014 * ubsan.c: Likewise.
14015 * varasm.c: Likewise.
14016 * varpool.c: Likewise.
14017 * tree.c: Remove functions moved to attribs.[ch].
14018 * tree.h: Likewise.
14019 * config/aarch64/aarch64.c: Add attrs.h header file.
14020 * config/alpha/alpha.c: Likewise.
14021 * config/arc/arc.c: Likewise.
14022 * config/arm/arm.c: Likewise.
14023 * config/avr/avr.c: Likewise.
14024 * config/bfin/bfin.c: Likewise.
14025 * config/c6x/c6x.c: Likewise.
14026 * config/cr16/cr16.c: Likewise.
14027 * config/cris/cris.c: Likewise.
14028 * config/darwin.c: Likewise.
14029 * config/epiphany/epiphany.c: Likewise.
14030 * config/fr30/fr30.c: Likewise.
14031 * config/frv/frv.c: Likewise.
14032 * config/ft32/ft32.c: Likewise.
14033 * config/h8300/h8300.c: Likewise.
14034 * config/i386/winnt.c: Likewise.
14035 * config/ia64/ia64.c: Likewise.
14036 * config/iq2000/iq2000.c: Likewise.
14037 * config/lm32/lm32.c: Likewise.
14038 * config/m32c/m32c.c: Likewise.
14039 * config/m32r/m32r.c: Likewise.
14040 * config/m68k/m68k.c: Likewise.
14041 * config/mcore/mcore.c: Likewise.
14042 * config/microblaze/microblaze.c: Likewise.
14043 * config/mips/mips.c: Likewise.
14044 * config/mmix/mmix.c: Likewise.
14045 * config/mn10300/mn10300.c: Likewise.
14046 * config/moxie/moxie.c: Likewise.
14047 * config/msp430/msp430.c: Likewise.
14048 * config/nds32/nds32-isr.c: Likewise.
14049 * config/nds32/nds32.c: Likewise.
14050 * config/nios2/nios2.c: Likewise.
14051 * config/nvptx/nvptx.c: Likewise.
14052 * config/pa/pa.c: Likewise.
14053 * config/pdp11/pdp11.c: Likewise.
14054 * config/powerpcspe/powerpcspe.c: Likewise.
14055 * config/riscv/riscv.c: Likewise.
14056 * config/rl78/rl78.c: Likewise.
14057 * config/rx/rx.c: Likewise.
14058 * config/s390/s390.c: Likewise.
14059 * config/sh/sh.c: Likewise.
14060 * config/sol2.c: Likewise.
14061 * config/sparc/sparc.c: Likewise.
14062 * config/spu/spu.c: Likewise.
14063 * config/stormy16/stormy16.c: Likewise.
14064 * config/tilegx/tilegx.c: Likewise.
14065 * config/tilepro/tilepro.c: Likewise.
14066 * config/v850/v850.c: Likewise.
14067 * config/vax/vax.c: Likewise.
14068 * config/visium/visium.c: Likewise.
14069 * config/xtensa/xtensa.c: Likewise.
14071 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14074 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14075 constraints since the -mupper-regs-* switches have been
14077 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14078 into a vector from a double word element that was extracted from
14079 another vector, and eliminate extra XXPERMDI instructions.
14080 (vsx_concat_<mode>_2): Likewise.
14081 (vsx_concat_<mode>_3): Likewise.
14082 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14083 concat to allow optimizing inserts from previous extracts.
14085 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14087 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14088 memory reference to a SSP offset in TLS address space.
14089 (ix86_print_operand) <case '@'>: Remove.
14090 (ix86_print_operand_punct_valid_p): Remove '@' code.
14091 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14092 UNSPEC_SP_TLS_TEST.
14093 (stack_tls_protect_set_<mode>): Remove.
14094 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14095 (stack_tls_protect_test_<mode>): Remove.
14096 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14098 2017-08-07 Olivier Hainque <hainque@adacore.com>
14101 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14103 2017-08-07 Douglas Rupp <rupp@adacore.com>
14105 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14106 variable was referenced as multidir in command.
14108 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14111 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14114 2017-08-07 Martin Liska <mliska@suse.cz>
14116 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14117 * config/rl78/rl78.c: Add include of attribs.h.
14118 * config/sh/sh.c: Likewise.
14119 * config/v850/v850.c: Likewise.
14121 2017-08-07 Tom de Vries <tom@codesourcery.com>
14123 PR middle-end/78266
14124 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14126 2017-08-07 Martin Liska <mliska@suse.cz>
14128 * config/mips/mips.c: Include attribs.h.
14130 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14133 * doc/invoke.texi: Document change in behvaior for -Ofast for
14136 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14138 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14140 (aarch64_pop_regs): Likewise.
14141 (aarch64_gen_load_pair): Likewise.
14142 (aarch64_save_callee_saves): Likewise.
14143 (aarch64_restore_callee_saves): Likewise.
14145 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14147 * config/i386/i386.c: Revert the last change.
14149 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14152 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14154 (ix86_finalize_stack_frame_flags): This. Also clear
14155 frame_pointer_needed if -fno-omit-frame-pointer is used without
14157 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14158 with ix86_finalize_stack_frame_flags.
14159 (ix86_expand_epilogue): Likewise.
14160 (ix86_expand_split_stack_prologue): Likewise.
14162 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14165 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14166 to P_AES for Westmere.
14168 2017-08-07 Jonathan Yong <10walls@gmail.com>
14170 * config/i386/mingw.opt (fset-stack-executable): Removed.
14171 * config/i386/cygming.opt (fset-stack-executable): Moved
14173 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14175 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14177 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14179 2017-08-07 Marek Polacek <polacek@redhat.com>
14181 PR middle-end/81737
14182 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14184 2017-08-07 Martin Liska <mliska@suse.cz>
14186 * attribs.h (canonicalize_attr_name): New function.
14187 (cmp_attribs): Move from c-format.c and adjusted.
14188 (is_attribute_p): Moved from tree.h.
14189 * tree-inline.c: Add new includes.
14190 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14191 (private_is_attribute_p): Remove.
14192 (private_lookup_attribute): Likewise.
14193 (private_lookup_attribute_by_prefix): Simplify.
14194 (remove_attribute): Use is_attribute_p.
14195 * tree.h: Remove removed declarations.
14197 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14199 PR middle-end/81698
14200 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14201 instead of computing it in the function. Formatting fix.
14202 (expand_case): Don't rely on default_edge being the first edge,
14203 clear it if removing it, pass default_edge to
14204 emit_case_dispatch_table.
14205 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14208 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14210 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14211 insn in the function, emit NOP after the insn.
14213 2017-08-06 Tom de Vries <tom@codesourcery.com>
14215 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14218 2017-08-06 Tom de Vries <tom@codesourcery.com>
14220 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14223 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14225 PR tree-optimization/57371
14226 * match.pd: New pattern.
14228 2017-08-04 Marek Polacek <polacek@redhat.com>
14230 PR middle-end/81695
14231 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14232 perform the computation in offset_int.
14234 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14236 PR tree-optimization/81136
14237 * tree-vectorizer.h: Include tree-hash-traits.h.
14238 (vec_base_alignments): New typedef.
14239 (vec_info): Add a base_alignments field.
14240 (vect_record_base_alignments): Declare.
14241 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14243 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14244 (create_data_ref): Add an is_conditional_in_stmt argument.
14245 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14246 the is_conditional_in_stmt field.
14247 (data_ref_loc): Add an is_conditional_in_stmt field.
14248 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14249 (find_data_references_in_stmt): Update call to create_data_ref.
14250 (graphite_find_data_references_in_stmt): Likewise.
14251 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14252 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14253 (vect_record_base_alignment): New function.
14254 (vect_record_base_alignments): Likewise.
14255 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14256 for nested statements even if we fail to compute a misalignment.
14257 Use pooled base alignments for unconditional references.
14258 (vect_find_same_alignment_drs): Compare base addresses instead
14260 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14261 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14263 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14265 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14266 Add an explicit name for the enum. Use auto_vec for slp_instances
14267 and grouped_stores.
14268 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14270 (_bb_vec_info): Add a constructor and destructor.
14271 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14272 (destroy_loop_vec_info): Delete.
14273 (vect_destroy_datarefs): Likewise.
14274 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14275 (vec_info::vec_info): New function.
14276 (vec_info::~vec_info): Likewise.
14277 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14278 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14279 destroy_loop_vec_info.
14280 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14281 (_loop_vec_info::_loop_vec_info): ...this.
14282 (destroy_loop_vec_info): Replace with...
14283 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14284 the stmt_vec_infos. Leave handling of vec_info information to its
14285 destructor. Remove explicit vector releases.
14286 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14287 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14288 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14289 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14290 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14291 (destroy_bb_vec_info): Replace with...
14292 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14293 information to its destructor.
14294 (vect_slp_analyze_bb_1): Use new and delete instead of
14295 new_bb_vec_info and destroy_bb_vec_info.
14296 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14299 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14301 * tree-data-ref.h (subscript): Add access_fn field.
14302 (data_dependence_relation): Add could_be_independent_p.
14303 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14304 (same_access_functions): Move to tree-data-ref.c.
14305 * tree-data-ref.c (ref_contains_union_access_p): New function.
14306 (access_fn_component_p): Likewise.
14307 (access_fn_components_comparable_p): Likewise.
14308 (dr_analyze_indices): Add a reference to access_fn_component_p.
14309 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14311 (constant_access_functions): Likewise.
14312 (add_other_self_distances): Likewise.
14313 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14314 (initialize_data_dependence_relation): Use XCNEW and remove
14315 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14316 of access functions that have the same type. Allow the
14317 subsequence to end with different bases in some circumstances.
14318 Record the chosen access functions in SUB_ACCESS_FN.
14319 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14320 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14321 (subscript_dependence_tester_1): Likewise dra and drb.
14322 (build_classic_dist_vector): Update calls accordingly.
14323 (subscript_dependence_tester): Likewise.
14324 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14325 DDR_COULD_BE_INDEPENDENT_P.
14326 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14327 comp_alias_ddrs instead of may_alias_ddrs.
14328 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14330 (vect_analyze_data_ref_dependence): Use it if
14331 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14332 distance vectors if that fails.
14333 (dependence_distance_ge_vf): New function.
14334 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14335 LOOP_VINFO_MAY_ALIAS_DDRS.
14337 2017-08-04 Richard Biener <rguenther@suse.de>
14339 PR middle-end/81705
14340 * fold-const.c (fold_binary_loc): Properly restrict
14341 minus_var0 && minus_var1 case when associating undefined overflow
14344 2017-08-04 Tom de Vries <tom@codesourcery.com>
14346 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14348 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14350 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14351 Don't start diagnostic messages with a capital letter.
14352 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14354 (rs6000_invalid_builtin): Likewise.
14355 (rs6000_trampoline_init): Likewise.
14357 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14360 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14361 after setting changeable df flags.
14363 2017-08-03 Richard Biener <rguenther@suse.de>
14365 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14366 up if the use is in USE - X.
14368 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14370 * toplev.c (dumpfile.h): New include.
14371 (internal_error_reentered): New static function. Use it...
14372 (internal_error_function): ...here to handle reentered internal_error.
14374 2017-08-03 Richard Biener <rguenther@suse.de>
14376 PR middle-end/81148
14377 * fold-const.c (split_tree): Add minus_var and minus_con
14378 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14379 here but always use minus_*.
14380 (associate_trees): Assert we never associate with MINUS_EXPR
14381 and NULL first operand. Do not recurse for PLUS_EXPR operands
14382 when associating as MINUS_EXPR either.
14383 (fold_binary_loc): Track minus_var and minus_con.
14385 2017-08-03 Tom de Vries <tom@codesourcery.com>
14388 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14389 ACCEL_COMPILER, apply finish_options on
14390 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14392 2017-08-03 Tom de Vries <tom@codesourcery.com>
14395 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14396 function_entry_patch_area_size > 0.
14398 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14401 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14402 instead of 10??LU, perform unit multiplication in wide_int,
14403 don't change alloc_object_size_limit if the limit is larger
14406 PR tree-optimization/81655
14407 PR tree-optimization/81588
14408 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14409 the case when ranges[i].low and high are 1 for unsigned type with
14412 PR middle-end/81052
14413 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14414 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14416 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14418 * tree-vrp.h: Add include guard.
14420 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14423 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14424 (ud2): New insn pattern.
14425 * config/i386/i386.c (ix86_expand_epilogue):
14426 For naked functions, generate ud2 instead of trap insn.
14428 2017-08-02 Marek Polacek <polacek@redhat.com>
14431 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14433 2017-08-02 Tom de Vries <tom@codesourcery.com>
14434 Cesar Philippidis <cesar@codesourcery.com>
14436 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14437 Add missing edge probabilities.
14439 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14441 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14442 Correct endianness.
14444 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14446 PR middle-end/79499
14447 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14448 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14449 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14450 prologue_seq sequences - if any.
14452 2017-08-02 Richard Biener <rguenther@suse.de>
14454 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14455 via vectors if supported, integer extracts via punning if supported
14456 or otherwise vector extracts.
14458 2017-08-02 Richard Biener <rguenther@suse.de>
14460 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14462 (bitmap_insert_into_set): ... this.
14464 2017-08-02 Richard Biener <rguenther@suse.de>
14466 PR tree-optimization/81633
14468 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14470 PR tree-optimization/71752
14471 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14473 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14475 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14476 (machine_function::call_ms2sysv_pad_out): Remove field.
14477 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14478 (ix86_compute_frame_layout): Likewise.
14480 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14483 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14484 attribute with interrupt attribute.
14486 2017-08-01 Andrew Pinski <apinski@cavium.com>
14488 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14489 BIT_INSERT_EXPR's operand 1
14490 to see if the types precision matches.
14492 2017-08-01 Martin Liska <mliska@suse.cz>
14494 PR middle-end/70140
14495 * builtins.c (expand_builtin_memcpy_args): Remove.
14496 (expand_builtin_memcpy): Call newly added function
14497 expand_builtin_memory_copy_args.
14498 (expand_builtin_memcpy_with_bounds): Likewise.
14499 (expand_builtin_mempcpy): Remove last argument.
14500 (expand_builtin_mempcpy_with_bounds): Likewise.
14501 (expand_builtin_memory_copy_args): New function created from
14502 expand_builtin_mempcpy_args with small modifications.
14503 (expand_builtin_mempcpy_args): Remove.
14504 (expand_builtin_stpcpy): Remove unused argument.
14505 (expand_builtin): Likewise.
14506 (expand_builtin_with_bounds): Likewise.
14508 2017-08-01 Martin Liska <mliska@suse.cz>
14511 Make mempcpy more optimal (PR middle-end/70140).
14513 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14516 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14517 __builtin_vec_cmpne verify both arguments are compatible vectors
14518 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14519 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14520 move computation of aligned to after checking the argument types.
14524 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14525 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14528 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14529 Doug Rupp <rupp@adacore.com>
14530 Olivier Hainque <hainque@adacore.com>
14532 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14533 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14535 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14536 for TARGET_OS_CPP_BUILTIN.
14537 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14538 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14540 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14541 passing required abi options to the assembler for EABI configurations.
14542 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14543 of .text.hot and .text.unlikely sections for kernel modules when
14544 using ARM style exceptions.
14545 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14546 options. Add EXTRA_CC1_SPEC.
14547 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14549 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14551 (ARM_TARGET2_DWARF_FORMAT): Define.
14552 * config/arm/t-vxworks: Adjust multilib control to removal of the
14553 Diab command line options.
14555 2017-08-01 Martin Liska <mliska@suse.cz>
14557 PR gcov-profile/81561
14558 * gcov.c (unblock): Make unblocking safe as we need to preserve
14559 index correspondence of blocks and block_lists.
14561 2017-08-01 Richard Biener <rguenther@suse.de>
14563 PR tree-optimization/81181
14564 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14565 (compute_antic): ... end of iteration here.
14567 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14569 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14570 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14571 (ftree-slp-vectorize): Likewise.
14572 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14573 can no longer be set independent of flag_tree_loop_vectorize.
14574 * omp-general.c (emp_max_vf): Likewise.
14575 * opts.c (enable_fdo_optimizations): Remove references to
14576 flag_tree_vectorize, these are now implicit.
14577 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14578 and leave it for the options machinery.
14580 2017-08-01 Martin Liska <mliska@suse.cz>
14582 PR middle-end/70140
14583 * builtins.c (expand_builtin_memcpy_args): Remove.
14584 (expand_builtin_memcpy): Call newly added function
14585 expand_builtin_memory_copy_args.
14586 (expand_builtin_memcpy_with_bounds): Likewise.
14587 (expand_builtin_mempcpy): Remove last argument.
14588 (expand_builtin_mempcpy_with_bounds): Likewise.
14589 (expand_builtin_memory_copy_args): New function created from
14590 expand_builtin_mempcpy_args with small modifications.
14591 (expand_builtin_mempcpy_args): Remove.
14592 (expand_builtin_stpcpy): Remove unused argument.
14593 (expand_builtin): Likewise.
14594 (expand_builtin_with_bounds): Likewise.
14596 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14599 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14600 print "ds:" only for immediates in generic address space.
14602 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14605 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14606 (ix86_function_ok_for_sibcall): Return false for naked functions.
14608 2017-08-01 Richard Biener <rguenther@suse.de>
14610 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14611 (compute_antic): Seed worklist with exit block predecessors.
14612 * cfganal.c (dfs_find_deadend): For a cycle return the source
14613 of the edge closing it.
14615 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14617 * config/aarch64/aarch64.c
14618 (aarch64_can_const_movi_rtx_p): Move 0 check.
14620 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14622 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14623 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14625 * match.pd: Ditto in address comparison pattern.
14627 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14629 PR tree-optimization/81627
14630 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14631 closed ssa form for store-store chain.
14633 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14635 PR tree-optimization/81620
14636 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14637 for store-store chain.
14639 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14641 PR tree-optimization/81588
14642 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14643 ranges[i].in_p, invert comparison code ccode. For >/>=,
14644 swap rhs1 and rhs2 and comparison code unconditionally,
14645 for </<= don't do that. Don't swap rhs1/rhs2 again if
14646 ranges[i].in_p, instead invert comparison code ccode if
14647 opcode or oe->rank is BIT_IOR_EXPR.
14650 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14651 a direct optab to conversion optab.
14652 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14653 with GET_MODE_INNER as last argument instead of optab_handler.
14654 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14655 vector extraction if possible and optab is available.
14656 * expr.c (store_constructor): Use convert_optab_handler instead
14657 of optab_handler. Use vector initialization from smaller
14658 vectors if possible and optab is available.
14659 * tree-vect-stmts.c (vectorizable_load): Likewise.
14660 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14661 now have two modes.
14662 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14663 of vec_init from half-sized vectors with the same element mode.
14664 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14665 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14666 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14667 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14668 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14669 after mode in gen_vec_extract* calls.
14670 (vec_extract<mode>): Renamed to ...
14671 (vec_extract<mode><ssescalarmodelower>): ... this.
14672 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14673 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14674 element mode after mode in gen_vec_init* calls.
14675 (VEC_INIT_HALF_MODE): New mode iterator.
14676 (vec_init<mode>): Renamed to ...
14677 (vec_init<mode><ssescalarmodelower>): ... this.
14678 (vec_init<mode><ssehalfvecmodelower>): New expander.
14679 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14680 (vec_extractv2sfsf): ... this.
14681 (vec_initv2sf): Renamed to ...
14682 (vec_initv2sfsf): ... this.
14683 (vec_extractv2si): Renamed to ...
14684 (vec_extractv2sisi): ... this.
14685 (vec_initv2si): Renamed to ...
14686 (vec_initv2sisi): ... this.
14687 (vec_extractv4hi): Renamed to ...
14688 (vec_extractv4hihi): ... this.
14689 (vec_initv4hi): Renamed to ...
14690 (vec_initv4hihi): ... this.
14691 (vec_extractv8qi): Renamed to ...
14692 (vec_extractv8qiqi): ... this.
14693 (vec_initv8qi): Renamed to ...
14694 (vec_initv8qiqi): ... this.
14695 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14696 (vec_init<mode>): Renamed to ...
14697 (vec_init<mode><VEC_base_l>): ... this.
14698 (vec_extract<mode>): Renamed to ...
14699 (vec_extract<mode><VEC_base_l>): ... this.
14700 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14701 (vec_initv2sfsf): ... this.
14702 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14703 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14704 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14705 element mode after mode in gen_vec_init* calls.
14706 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14707 (vec_init<mode><Vel>): ... this.
14708 (vec_extract<mode>): Renamed to ...
14709 (vec_extract<mode><Vel>): ... this.
14710 * config/aarch64/iterators.md (Vel): New mode attribute.
14711 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14712 Add element mode after mode in gen_vec_extract* calls.
14713 * config/s390/vector.md (non_vec_l): New mode attribute.
14714 (vec_extract<mode>): Renamed to ...
14715 (vec_extract<mode><non_vec_l>): ... this.
14716 (vec_init<mode>): Renamed to ...
14717 (vec_init<mode><non_vec_l>): ... this.
14718 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14719 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14721 * config/arm/iterators.md (V_elem_l): New mode attribute.
14722 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14723 (vec_extract<mode><V_elem_l>): ... this.
14724 (vec_extractv2di): Renamed to ...
14725 (vec_extractv2didi): ... this.
14726 (vec_init<mode>): Renamed to ...
14727 (vec_init<mode><V_elem_l>): ... this.
14728 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14729 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14730 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14731 Add element mode after gen_vec_extract* calls.
14732 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14733 (vec_init<mode><unitmode>): ... this.
14734 (vec_extract<mode>): Renamed to ...
14735 (vec_extract<mode><unitmode>): ... this.
14736 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14737 (vec_init<mode><unitmode>): ... this.
14738 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14739 (vec_initv2sfsf): ... this.
14740 (vec_extractv2sf): Renamed to ...
14741 (vec_extractv2sfsf): ... this.
14742 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14743 Add element mode after gen_vec_extract* calls.
14744 * config/mips/mips.md (unitmode): New mode iterator.
14745 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14746 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14747 * config/spu/spu.md (inner_l): New mode attribute.
14748 (vec_init<mode>): Renamed to ...
14749 (vec_init<mode><inner_l>): ... this.
14750 (vec_extract<mode>): Renamed to ...
14751 (vec_extract<mode><inner_l>): ... this.
14752 * config/sparc/sparc.md (veltmode): New mode iterator.
14753 (vec_init<VMALL:mode>): Renamed to ...
14754 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14755 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14756 (vec_initv2sisi): ... this.
14757 (vec_initv2sf): Renamed to ...
14758 (vec_initv2sfsf): ... this.
14759 (vec_extractv2sf): Renamed to ...
14760 (vec_extractv2sfsf): ... this.
14761 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14762 (vec_init<mode>): Renamed to ...
14763 (vec_init<mode><VEC_base_l>): ... this.
14764 (vec_extract<mode>): Renamed to ...
14765 (vec_extract<mode><VEC_base_l>): ... this.
14766 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14767 (vec_initv2sfsf): ... this.
14768 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14769 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14770 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14771 gen_vec_init* calls.
14773 2017-08-01 Richard Biener <rguenther@suse.de>
14775 PR tree-optimization/81297
14776 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14777 TREE_OVERFLOW from INTEGER_CSTs.
14779 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14781 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14783 2017-07-31 Carl Love <cel@us.ibm.com>
14785 * config/rs6000/rs6000-c: Add support for built-in functions
14786 vector signed char vec_xl_be (signed long long, signed char *);
14787 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14788 vector signed int vec_xl_be (signed long long, signed int *);
14789 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14790 vector signed long long vec_xl_be (signed long long, signed long long *);
14791 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14792 vector signed short vec_xl_be (signed long long, signed short *);
14793 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14794 vector double vec_xl_be (signed long long, double *);
14795 vector float vec_xl_be (signed long long, float *);
14796 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14797 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14798 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14800 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14801 (altivec_expand_builtin): Add switch statement to call
14802 altivec_expand_xl_be for each builtin.
14803 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14804 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14805 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14806 __builtin_vsx_le_be_v16qi.
14807 * doc/extend.texi: Update the built-in documentation file for the
14808 new built-in functions.
14810 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14813 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14815 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14817 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14819 * config.gcc: Add z14.
14820 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14821 CPU model numbers for z13s and z14.
14822 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14824 * config/s390/s390-opts.h (enum processor_type): Rename
14825 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14826 * config/s390/s390.c (processor_table): Add field for CPU name to
14827 be passed to Binutils.
14828 (s390_asm_output_machine_for_arch): Use the new field in
14829 processor_table for Binutils.
14830 (s390_expand_builtin): Replace arch12 with z14.
14831 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14832 (s390_get_sched_attrmask): Likewise.
14833 (s390_get_unit_mask): Likewise.
14834 * config/s390/s390.opt: Add z14 to processor_type enum.
14836 2017-07-31 Martin Jambor <mjambor@suse.cz>
14839 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14840 regardless of optimization level.
14842 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14843 Martin Liska <mliska@suse.cz>
14845 * predict.def: Remove old comment and adjust probability.
14846 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14847 PREDICT statements.
14849 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14852 * config/i386/i386.c (ix86_function_naked): New function.
14853 (ix86_can_use_return_insn_p): Return false for naked functions.
14854 (ix86_expand_prologue): Skip prologue for naked functions.
14855 (ix86_expand_epilogue): Skip epilogue for naked functions
14856 and emit trap instruction.
14857 (ix86_warn_func_return): New function.
14858 (ix86_attribute_table): Add "naked" attribute specification.
14859 (TARGET_WARN_FUNC_RETURN): Define.
14860 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14862 2017-07-31 Martin Liska <mliska@suse.cz>
14864 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14865 (dump_gimple_bb_header): Always dump BB info.
14866 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14868 2017-07-31 Martin Liska <mliska@suse.cz>
14871 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14872 also with current_function_decl non-null equality.
14874 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14877 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14878 change type to the element type, instead add eltype variable and
14879 use it where we are interested in the element type.
14881 PR tree-optimization/81603
14882 * ipa-polymorphic-call.c
14883 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14884 offset arithmetic in offset_int, bail out if the resulting bit offset
14885 doesn't fit into shwi.
14887 2017-07-31 Martin Liska <mliska@suse.cz>
14889 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14890 (gimplify_save_expr): Fix comment.
14892 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14895 * config/i386/i386.c (ix86_function_arg): Update arguments for
14897 (ix86_compute_frame_layout): Set the initial stack offset to
14898 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14899 INCOMING_FRAME_SP_OFFSET.
14900 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14901 stack before exception handler returns.
14902 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14903 the 'ERROR_CODE' for exception handler.
14905 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14907 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14908 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14909 (ASM_OUTPUT_REG_POP): Ditto.
14910 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14911 instead of asm_fprintf to output pure string.
14913 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14915 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14916 to imported_module_or_decl hook.
14917 (debug_nothing_tree_tree_tree_bool): Remove.
14918 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14919 * debug.c (do_nothing_debug_hooks): Use
14920 debug_nothing_tree_tree_tree_bool_bool instead of
14921 debug_nothing_tree_tree_tree_bool.
14922 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14923 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14924 * sdbout.c (sdb_debug_hooks): Likewise.
14925 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14926 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14928 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14929 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14930 attribute, don't add anything.
14932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14934 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14935 (fold_build2_stat_loc): Likewise.
14936 (fold_build3_stat_loc): Likewise.
14937 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14938 (fold_build1_loc): Remove macro.
14939 (fold_build2_loc): Likewise.
14940 (fold_build3_loc): Likewise.
14942 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14944 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14945 (gimple_build_debug_bind_source_stat): Likewise.
14946 * gimple.h (gimple_build_debug_bind): Remove macro.
14947 (gimple_build_debug_bind_source): Likewise.
14949 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14951 * bitmap.c (bitmap_alloc): Adjust.
14952 (bitmap_gc_alloc): Likewise.
14953 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14957 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14958 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14959 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14960 (bitmap_gc_alloc_stat): Likewise.
14961 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14963 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14965 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14966 * rtl.h (shallow_copy_rtx): Remove macro.
14968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14970 * emit-rtl.c (gen_raw_REG): Adjust.
14971 * gengenrtl.c (gendef): Likewise.
14972 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14973 * rtl.h (rtx_alloc): Remove macro.
14975 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14977 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14978 (build_tree_list_stat): Likewise.
14979 * tree.h (build_tree_list): Remove macro.
14980 (build_tree_list_vec): Likewise.
14982 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14984 * tree.c (make_vector_stat): Remove _stat from name.
14985 (build_vector_stat): Likewise.
14986 * tree.h (make_vector_stat): Remove macro.
14987 (build_vector_stat): Likewise.
14989 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14991 * tree.h (build_var_debug_value): Remove prototype.
14993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14995 * tree.c (tree_cons_stat): Remove _stat from name.
14996 * tree.h (tree_cons): Remove macro.
14998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15000 * tree.c (build_vl_exp_stat): Remove _stat from name.
15001 * tree.h (build_vl_exp): Remove macro.
15003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15005 * tree.c (build_decl_stat): Remove _stat from name.
15006 * tree.h (build_decl): Remove macro.
15008 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15010 * gimple.c (gimple_build_with_ops_stat): Adjust.
15011 (gimple_alloc_stat): Remove _stat from name.
15012 * gimple.h (gimple_alloc): Remove macro.
15014 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15016 * tree.c (make_tree_vec_stat): Remove _stat from name.
15017 (grow_tree_vec_stat): Likewise.
15018 * tree.h (make_tree_vec_stat): Adjust prototype.
15019 (grow_tree_vec_stat): Likewise.
15020 (make_tree_vec): Remove macro.
15021 (grow_tree_vec): Likewise.
15023 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15025 * fold-const.c (fold_build1_stat_loc): Adjust.
15026 (fold_build2_stat_loc): Likewise.
15027 (fold_build3_stat_loc): Likewise.
15028 * tree.c (build0_stat): Remove _stat from name.
15029 (build1_stat): Likewise.
15030 (build2_stat): Likewise.
15031 (build3_stat): Likewise.
15032 (build4_stat): Likewise.
15033 (build5_stat): Likewise.
15034 * tree.h (build1_loc): Remove macro, and rename _stat function
15036 (build2_loc): Likewise.
15037 (build3_loc): Likewise.
15038 (build4_loc): Likewise.
15039 (build5_loc): Likewise.
15041 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15043 * tree.c (make_int_cst_stat): Remove _stat from name.
15044 * tree.h (make_int_cst_stat): Adjust prototype.
15045 (make_int_cst): Remove macro.
15047 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15049 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15050 * tree.h (make_tree_binfo_stat): Adjust prototype.
15051 (make_tree_binfo): Remove.
15053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15055 * tree.c (copy_node_stat): Rename to copy_node.
15056 (build_distinct_type_copy): Adjust.
15057 * tree.h (copy_node_stat): Adjust prototype.
15058 (copy_node): Remove macro.
15060 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15062 * tree.c (make_node_stat): rename to make_node.
15063 (build_tree_list_stat): Adjust.
15064 (build0_stat): Likewise.
15065 (build2_stat): Likewise.
15066 (build3_stat): Likewise.
15067 (build4_stat): Likewise.
15068 (build5_stat): Likewise.
15069 (build_decl_stat): Likewise.
15070 * tree.h (make_node_stat): Adjust prototype.
15071 (make_node): remove macro.
15073 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15075 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15076 (PPC_FEATURE2_SCV): Likewise.
15077 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15079 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15081 * config/aarch64/aarch64.c
15082 (aarch64_internal_mov_immediate): Add new special pattern.
15083 * config/aarch64/aarch64.md (*movdi_aarch64):
15084 Add reg/32bit const mov case.
15086 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15087 Richard Sandiford <richard.sandiford@linaro.org>
15089 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15090 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15091 Add integer and movi cases.
15092 (movi-split-hf-df-sf split, fp16): New.
15093 (enabled): Added TARGET_FP_F16INST.
15094 * config/aarch64/iterators.md (GPF_HF): New.
15095 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15097 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15099 * config/aarch64/aarch64.c
15100 (aarch64_simd_container_mode): Add prototype.
15101 (aarch64_expand_mov_immediate): Add HI support.
15102 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15103 (aarch64_can_const_movi_rtx_p): New.
15104 (aarch64_preferred_reload_class):
15105 Remove restrictions of using FP registers for certain SIMD operations.
15106 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15107 (aarch64_valid_floating_const): Add integer move validation.
15108 (aarch64_simd_imm_scalar_p): Remove.
15109 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15110 (aarch64_legitimate_constant_p): Expand list of supported cases.
15111 * config/aarch64/aarch64-protos.h
15112 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15113 (aarch64_reinterpret_float_as_int): New.
15114 (aarch64_simd_imm_scalar_p): Remove.
15115 * config/aarch64/constraints.md (Uvi): New.
15116 (Dd): Split into Ds and new Dd.
15117 * config/aarch64/aarch64.md (*movsi_aarch64):
15119 (*movdi_aarch64): Add SIMD mov case.
15121 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15123 * tree-predcom.c: (struct chain): Handle store-store chain in which
15124 stores for elimination only store loop invariant values.
15125 (execute_pred_commoning_chain): Ditto.
15126 (prepare_initializers_chain_store_elim): Ditto.
15127 (prepare_finalizers): Ditto.
15128 (is_inv_store_elimination_chain): New function.
15129 (initialize_root_vars_store_elim_1): New function.
15131 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15133 * tree-predcom.c: Revise general description of the pass.
15134 (enum chain_type): New enum type for store elimination.
15135 (struct chain): New field supporting store elimination.
15136 (struct component): Ditto.
15137 (dump_chain): Dump store-stores chain.
15138 (release_chain): Release resources.
15139 (split_data_refs_to_components): Compute and create component
15140 contains only stores for elimination.
15141 (get_chain_last_ref_at): New function.
15142 (make_invariant_chain): Initialization.
15143 (make_rooted_chain): Specify chain type in parameter and record it.
15144 (add_looparound_copies): Skip for store-stores chain.
15145 (determine_roots_comp): Compute type of chain and pass it to
15147 (initialize_root_vars_store_elim_2): New function.
15148 (finalize_eliminated_stores): New function.
15149 (remove_stmt): Handle store for elimination.
15150 (execute_pred_commoning_chain): Execute predictive commoning on
15151 store-store chains.
15152 (determine_unroll_factor): Skip unroll for store-stores chain.
15153 (prepare_initializers_chain_store_elim): New function.
15154 (prepare_initializers_chain): Hanlde store-store chain.
15155 (prepare_finalizers_chain, prepare_finalizers): New function.
15156 (tree_predictive_commoning_loop): Return integer value indicating
15157 if loop is unrolled or lcssa form is corrupted.
15158 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15160 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15162 * tree-predcom.c (initialize_root): Delete.
15163 (execute_pred_commoning_chain): Initialize root vars and replace
15164 reference of non-combined chain directly, rather than call above
15167 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15169 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15170 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15172 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15174 * tree-predcom.c (struct chain): New field init_seq.
15175 (release_chain): Release init_seq.
15176 (prepare_initializers_chain): Record intialization stmts in above
15178 (insert_init_seqs): New function.
15179 (tree_predictive_commoning_loop): Call insert_init_seqs.
15181 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15183 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15185 2017-07-28 Richard Biener <rguenther@suse.de>
15187 * match.pd: Remove superfluous :c.
15188 * genmatch.c (simplify::id): Add member.
15189 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15191 (current_id): New global.
15192 (dt_node::parent): Move from ...
15193 (dt_operand::parent): ... here. Add for_id member.
15194 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15195 (decision_tree::find_node): Relax order requirement when
15196 merging DT_TRUE nodes to ones inbetween the current simplify
15197 and the one we try to merge with. Add diagnostic whenever
15198 we need to enforce pattern order by not merging.
15199 (decision_tree::insert): Set current_id.
15200 (decision_tree::print_node): Dump parent node and for_id.
15201 (parser::last_id): Add member.
15202 (parser::push_simplify): Assign unique id.
15203 (parser::parser): Initialize last_id.
15205 2017-07-28 Martin Liska <mliska@suse.cz>
15208 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15209 gimple_build_debug_bind.
15211 2017-07-28 Richard Biener <rguenther@suse.de>
15213 PR tree-optimization/81502
15214 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15216 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15218 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15219 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15220 for BIT_FIELD_REF args.
15221 * fold-const.c (make_bit_field_ref): Likewise.
15222 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15224 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15227 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15228 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15229 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15230 Or it into SANITIZER_UNDEFINED.
15231 * ubsan.c: Include gimple-fold.h and varasm.h.
15232 (ubsan_expand_ptr_ifn): New function.
15233 (instrument_pointer_overflow): New function.
15234 (maybe_instrument_pointer_overflow): New function.
15235 (instrument_object_size): Formatting fix.
15236 (pass_ubsan::execute): Call instrument_pointer_overflow
15237 and maybe_instrument_pointer_overflow.
15238 * internal-fn.c (expand_UBSAN_PTR): New function.
15239 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15240 * sanitizer.def (__ubsan_handle_pointer_overflow,
15241 __ubsan_handle_pointer_overflow_abort): New builtins.
15242 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15243 * internal-fn.def (UBSAN_PTR): New internal function.
15244 * opts.c (sanitizer_opts): Add pointer-overflow.
15245 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15246 * fold-const.c (build_range_check): Compute pointer range check in
15247 integral type if pointer arithmetics would be needed. Formatting
15250 2017-07-28 Martin Liska <mliska@suse.cz>
15253 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15254 parameters that are of a variable-length.
15256 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15258 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15260 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15261 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15262 (CRT_CALL_STATIC_FUNCTION): Likewise.
15263 (ASM_DEFAULT_SPEC): New define.
15264 (ASM_SPEC32): Likewise.
15265 (ASM_SPEC64): Likewise.
15266 (ASM_SPEC_COMMON): Likewise.
15267 (ASM_SPEC): Likewise.
15268 (INVALID_64BIT): Likewise.
15269 (LINK_OS_DEFAULT_SPEC): Likewise.
15270 (LINK_OS_SPEC32): Likewise.
15271 (LINK_OS_SPEC64): Likewise.
15272 (POWERPC_LINUX): Likewise.
15273 (PTRDIFF_TYPE): Likewise.
15274 (RESTORE_FP_PREFIX): Likewise.
15275 (RESTORE_FP_SUFFIX): Likewise.
15276 (SAVE_FP_PREFIX): Likewise.
15277 (SAVE_FP_SUFFIX): Likewise.
15278 (SIZE_TYPE): Likewise.
15279 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15280 (TARGET_64BIT): Likewise.
15281 (TARGET_64BIT): Likewise.
15282 (TARGET_AIX): Likewise.
15283 (WCHAR_TYPE_SIZE): Likewise.
15284 (WCHAR_TYPE): Undefine.
15285 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15286 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15287 (CPP_OS_RTEMS_SPEC): Delete.
15288 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15289 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15291 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15293 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15295 PR tree-optimization/81578
15296 * tree-parloops.c (build_new_reduction): Bail out if
15297 reduction_code isn't one of the standard OpenMP reductions.
15298 Move the details printing after that decision.
15300 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15302 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15303 related to reload_in_progress.
15304 (splat_input_operand): Likewise.
15305 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15307 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15309 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15310 (TARGET_INSTANTIATE_DECLS): Likewise.
15311 (legitimate_indexed_address_p): Delete reload_in_progress code.
15312 (rs6000_debug_legitimate_address_p): Likewise.
15313 (rs6000_eliminate_indexed_memrefs): Likewise.
15314 (rs6000_emit_le_vsx_store): Likewise.
15315 (rs6000_emit_move_si_sf_subreg): Likewise.
15316 (rs6000_emit_move): Likewise.
15317 (register_to_reg_type): Likewise.
15318 (rs6000_pre_atomic_barrier): Likewise.
15319 (rs6000_machopic_legitimize_pic_address): Likewise.
15320 (rs6000_allocate_stack_temp): Likewise.
15321 (rs6000_address_for_fpconvert): Likewise.
15322 (rs6000_address_for_altivec): Likewise.
15323 (rs6000_secondary_memory_needed_rtx): Delete function.
15324 (rs6000_check_sdmode): Likewise.
15325 (rs6000_alloc_sdmode_stack_slot): Likewise.
15326 (rs6000_instantiate_decls): Likewise.
15327 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15328 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15329 Delete reload_in_progress.
15330 (*vec_reload_and_plus_<mptrsize>): Likewise.
15331 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15332 (vsx_div_v2di): Likewise.
15333 (vsx_udiv_v2di): Likewise.
15335 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15337 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15338 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15339 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15340 (rs6000_debug_reg_global): Delete print of LRA status.
15341 (rs6000_option_override_internal): Delete dead LRA related code.
15342 (rs6000_lra_p): Delete function.
15343 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15345 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15347 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15348 * config/riscv/rtems.h: New file.
15350 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15351 Sudakshina Das <sudi.das@arm.com>
15353 * config/aarch64/aarch64.md
15354 (define_split for and<mode>3nr_compare): Move
15355 non aarch64_logical_operand to a register.
15356 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15357 register immediate operand to a register.
15358 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15360 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15362 PR middle-end/81564
15363 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15365 2017-07-27 Richard Biener <rguenther@suse.de>
15367 PR tree-optimization/81573
15368 PR tree-optimization/81494
15369 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15370 multi defuse cycle case.
15372 2017-07-27 Richard Biener <rguenther@suse.de>
15374 PR tree-optimization/81571
15375 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15378 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15380 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15381 earlier and only if MASK_FPU is set. Adjust formatting.
15383 2017-07-27 Martin Liska <mliska@suse.cz>
15385 * opt-functions.awk: Add validation of value of Init.
15386 * optc-gen.awk: Pass new argument.
15388 2017-07-27 Martin Liska <mliska@suse.cz>
15390 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15391 Fix wrong condition.
15393 2017-07-27 Martin Liska <mliska@suse.cz>
15395 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15396 BBs and edges seen by autoFDO.
15398 2017-07-27 Richard Biener <rguenther@suse.de>
15400 PR tree-optimization/81502
15401 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15402 with incompatible but same sized type.
15403 (execute_update_addresses_taken): Likewise.
15405 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15407 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15408 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15410 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15413 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15414 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15415 Change s_operand to memory_operand.
15417 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15419 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15420 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15421 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15422 Emit instructions rather than returning an expression. Handle TFmode
15423 and KFmode by casting to TImode.
15424 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15425 (rs6000_emit_le_vsx_store): Likewise.
15426 * config/rs6000/vsx.md (VSX_TI): New iterator.
15427 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15428 (*vsx_le_undo_permute_<mode>): Likewise.
15429 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15430 emit the split sequence.
15431 (*vsx_le_perm_store_<mode>): Likewise.
15433 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15435 PR tree-optimization/81555
15436 PR tree-optimization/81556
15437 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15438 if true, force CHANGED for the recursive invocation.
15439 (reassociate_bb): Remember original length of ops array, pass
15440 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15442 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15443 attributes for noipa attribute. For naked attribute use
15444 lookup_attribute first before lookup_attribute_spec.
15445 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15447 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15448 for functions with noipa attribute.
15449 (cgraph_externally_visible_p): Return true for functions with noipa
15451 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15452 for functions with noipa attribute.
15453 * doc/extend.texi: Document noipa function attribute.
15454 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15455 also for functions with noipa attribute.
15456 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15458 2017-07-26 Andrew Pinski <apinski@cavium.com>
15460 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15461 vec_unalign_load_cost and vec_unalign_store_cost.
15463 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15465 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15466 -mvsx-small-integer option.
15467 (ISA_3_0_MASKS_IEEE): Likewise.
15468 (OTHER_VSX_VECTOR_MASKS): Likewise.
15469 (POWERPC_MASKS): Likewise.
15470 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15471 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15472 code, only testing for DImode being allowed in non-VSX floating
15474 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15475 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15477 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15478 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15479 TARGET_P8_VECTOR test.
15480 (rs6000_secondary_reload_simple_move): Likewise.
15481 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15482 since TARGET_P9_VECTOR was already tested.
15483 (rs6000_opt_masks): Remove -mvsx-small-integer.
15484 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15485 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15487 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15488 test for TARGET_VEXTRACTUB was used, and that uses
15490 (p9 extract splitter): Likewise.
15491 (vsx_extract_<mode>_di_p9): Likewise.
15492 (vsx_extract_<mode>_store_p9): Likewise.
15493 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15494 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15495 the elimination of TARGET_VSX_SMALL_INTEGER.
15496 (vsx_extract_<mode>_p8): Likewise.
15497 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15498 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15499 (vsx_set_<mode>_p9): Likewise.
15500 (vsx_set_v4sf_p9): Likewise.
15501 (vsx_set_v4sf_p9_zero): Likewise.
15502 (vsx_insert_extract_v4sf_p9): Likewise.
15503 (vsx_insert_extract_v4sf_p9_2): Likewise.
15504 * config/rs6000/rs6000.md (sign extend splitter): Change
15505 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15506 (floatsi<mode>2_lfiwax_mem): Likewise.
15507 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15508 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15509 since a test for TARGET_P9_VECTOR was used.
15510 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15511 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15512 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15513 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15514 TARGET_P8_VECTOR test.
15515 (fix_trunc<mode>si2_stfiwx): Likewise.
15516 (fix_trunc<mode>si2_internal): Likewise.
15517 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15518 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15520 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15521 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15522 TARGET_P8_VECTOR test.
15523 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15524 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15525 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15527 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15528 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15529 since a test for TARGET_P9_VECTOR was used.
15530 (splitter for loading small constants): Likewise.
15532 2017-07-26 Andrew Pinski <apinski@cavium.com>
15534 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15537 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15540 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15541 (fp_valid_at): Likewise.
15543 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15545 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15546 (qdf24xx_addrcost_table): Likewise.
15547 (cortexa57_tunings): Update to use generic_branch_cost.
15548 (cortexa72_tunings): Likewise.
15549 (cortexa73_tunings): Likewise.
15550 (qdf24xx_tunings): Likewise.
15552 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15554 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15555 (thunderx2t99_branch_cost): Likewise.
15556 (cortexa35_tunings): Update to use generic_branch_cost.
15557 (cortexa53_tunings): Likewise.
15558 (cortexa57_tunings): Likewise.
15559 (cortexa72_tunings): Likewise.
15560 (cortexa73_tunings): Likewise.
15561 (thunderx2t99_tunings): Likewise.
15563 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15565 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15566 (sparc_option_override): Honour MASK_FSMULD.
15567 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15568 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15569 * config/sparc/sparc.opt (mfsmuld): New option.
15570 * doc/invoke.texi (mfsmuld): Document option.
15572 2017-07-26 Marek Polacek <polacek@redhat.com>
15574 PR middle-end/70992
15575 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15577 2017-07-26 Richard Biener <rguenther@suse.de>
15579 * gimple-match-head.c (do_valueize): Return OP if valueize
15581 (get_def): New helper to get at the def stmt of a SSA name
15582 if valueize allows.
15583 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15584 do_valueize to get at the def stmt.
15585 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15587 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15589 PR middle-end/46932
15590 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15592 2017-07-26 Martin Liska <mliska@suse.cz>
15595 * function.c (expand_function_start): Make expansion of
15596 nonlocal_goto_save_area after parm_birth_insn.
15598 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15600 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15601 from all CPU target flags enable members.
15603 2017-07-26 Richard Biener <rguenther@suse.de>
15605 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15606 (decision_tree::gen): Make 'type' const.
15607 (write_predicate): Likewise.
15609 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15611 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15612 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15613 (rs6000_option_override_internal): Likewise.
15614 (rs6000_expand_vector_set): Likewise.
15615 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15616 (TARGET_UPPER_REGS_SF): Likewise.
15617 (TARGET_UPPER_REGS_DI): Likewise.
15618 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15619 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15620 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15621 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15622 (Splitters for DI constants in Altivec registers): Likewise.
15623 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15624 (vsx_set_v4sf_p9): Likewise.
15625 (vsx_set_v4sf_p9_zero): Likewise.
15626 (vsx_insert_extract_v4sf_p9): Likewise.
15627 (vsx_insert_extract_v4sf_p9_2): Likewise.
15629 2017-07-25 Carl Love <cel@us.ibm.com>
15631 * doc/extend.texi: Update the built-in documentation file for the
15632 existing built-in functions
15633 vector signed char vec_cnttz (vector signed char);
15634 vector unsigned char vec_cnttz (vector unsigned char);
15635 vector signed short vec_cnttz (vector signed short);
15636 vector unsigned short vec_cnttz (vector unsigned short);
15637 vector signed int vec_cnttz (vector signed int);
15638 vector unsigned int vec_cnttz (vector unsigned int);
15639 vector signed long long vec_cnttz (vector signed long long);
15640 vector unsigned long long vec_cnttz (vector unsigned long long);
15642 2017-07-25 Andrew Pinski <apinski@cavium.com>
15644 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15645 accesses where the use is for the first operand of a BIT_INSERT.
15647 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15650 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15651 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15653 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15655 * config/i386/gstabs.h: Delete.
15656 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15658 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15660 * config/i386/i386.c (ix86_decompose_address): Do not check for
15661 register RTX when looking at index_reg or base_reg.
15662 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15664 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15666 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15667 to update EH info here.
15669 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15671 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15673 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15675 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15677 2017-07-25 Torsten Duwe <duwe@suse.de>
15679 * common.opt: Introduce -fpatchable-function-entry
15680 command line option, and its variables function_entry_patch_area_size
15681 and function_entry_patch_area_start.
15682 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15683 including a two-value parser.
15684 * target.def (print_patchable_function_entry): New target hook.
15685 * targhooks.h (default_print_patchable_function_entry): New function.
15686 * targhooks.c (default_print_patchable_function_entry): Likewise.
15687 * toplev.c (process_options): Switch off IPA-RA if
15688 patchable function entries are being generated.
15689 * varasm.c (assemble_start_function): Look at the
15690 patchable-function-entry command line switch and current
15691 function attributes and maybe generate NOP instructions by
15692 calling the print_patchable_function_entry hook.
15693 * doc/extend.texi: Document patchable_function_entry attribute.
15694 * doc/invoke.texi: Document -fpatchable_function_entry
15695 command line option.
15696 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15698 * doc/tm.texi: Re-generate.
15700 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15703 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15704 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15706 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15708 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15710 2017-07-25 Richard Biener <rguenther@suse.de>
15712 PR tree-optimization/81455
15713 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15714 not walk in cycles when looking for guards.
15716 2017-07-25 Richard Biener <rguenther@suse.de>
15718 PR tree-optimization/81529
15719 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15720 when optimizing backedge uses.
15722 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15724 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15726 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15727 to dl_section_ref. On AIX, append an expression to subtract
15728 the size of the section length to dl_section_ref.
15730 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15732 * configure.ac: If any of the config.* scripts fail, exit 1.
15733 * configure: Regenerate.
15735 2017-07-25 Richard Biener <rguenther@suse.de>
15737 PR middle-end/81546
15738 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15739 of immediate uses, be more verbose on errors.
15741 2017-07-25 Richard Biener <rguenther@suse.de>
15743 PR tree-optimization/81510
15744 * tree-vect-loop.c (vect_is_simple_reduction): When the
15745 reduction stmt is not inside the loop bail out.
15747 2017-07-25 Richard Biener <rguenther@suse.de>
15749 PR tree-optimization/81303
15750 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15751 profitability check against LOOP_VINFO_NITERSM1.
15753 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15755 * domwalk.c (cmp_bb_postorder): Simplify.
15756 (sort_bbs_postorder): New function. Use it...
15757 (dom_walker::walk): ...here to optimize common cases.
15759 2017-07-25 Martin Liska <mliska@suse.cz>
15762 * ipa-visibility.c (function_and_variable_visibility): Make the
15763 redirection just on target that supports aliasing.
15764 Fix GNU coding style.
15766 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15769 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15771 * config/arm/rtems.h: Likewise.
15772 * config/bfin/rtems.h: Likewise.
15773 * config/i386/rtemself.h: Likewise.
15774 * config/lm32/rtems.h: Likewise.
15775 * config/m32c/rtems.h: Likewise.
15776 * config/m68k/rtemself.h: Likewise.
15777 * config/microblaze/rtems.h: Likewise.
15778 * config/mips/rtems.h: Likewise.
15779 * config/moxie/rtems.h: Likewise.
15780 * config/nios2/rtems.h: Likewise.
15781 * config/powerpcspe/rtems.h: Likewise.
15782 * config/rs6000/rtems.h: Likewise.
15783 * config/rtems.h: Likewise.
15784 * config/sh/rtems.h: Likewise.
15785 * config/sh/rtemself.h: Likewise.
15786 * config/sparc/rtemself.h: Likewise.
15788 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15791 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15792 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15793 * tree-ssa-structalias.c (alias_get_name): Same.
15795 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15798 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15799 instructions if no du chain is found.
15801 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15803 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15805 2017-07-25 Richard Biener <rguenther@suse.de>
15807 PR middle-end/81505
15808 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15811 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15813 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15814 upper-regs options.
15815 (ISA_2_7_MASKS_SERVER): Likewise.
15816 (ISA_3_0_MASKS_IEEE): Likewise.
15817 (OTHER_P8_VECTOR_MASKS): Likewise.
15818 (OTHER_VSX_VECTOR_MASKS): Likewise.
15819 (POWERPC_MASKS): Likewise.
15820 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15821 duplicate list of options.
15822 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15823 explicit -mupper-regs options.
15824 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15825 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15826 alias for -mupper-regs-df.
15827 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15828 (rs6000_init_hard_regno_mode_ok): Likewise.
15829 (rs6000_option_override_internal): Likewise.
15830 (rs6000_opt_masks): Likewise.
15831 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15832 options in terms of whether -mvsx or -mpower8-vector was used.
15833 (TARGET_UPPER_REGS_DI): Likewise.
15834 (TARGET_UPPER_REGS_SF): Likewise.
15835 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15836 -mupper-regs-* options.
15838 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15840 * passes.c (emergency_dump_function): Print some empty lines and a
15841 header before the RTL dump.
15843 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15845 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15847 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15850 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15851 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15853 2017-07-24 Carl Love <cel@us.ibm.com>
15855 * config/rs6000/rs6000-c.c: Add support for built-in functions
15856 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15857 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15858 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15859 vec_extract_fp_from_shortl): Add defines for the two builtins.
15860 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15861 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15863 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15864 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15865 * doc/extend.texi: Update the built-in documentation file for the
15866 new built-in function.
15868 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15871 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15873 * doc/generic.texi: Likewise.
15874 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15875 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15877 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15879 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15880 (aarch64_mls_elt_merge<mode>): Likewise.
15882 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15884 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15885 having __cxa_atexit.
15887 2017-07-23 Michael Collison <michael.collison@arm.com>
15889 * config/arm/arm.c (arm_option_override): Deprecate
15890 use of -mstructure-size-boundary.
15891 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15892 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15894 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15897 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15898 Reduce cost estimate for direct moves.
15900 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15903 * config/i386/i386.c (ix86_option_override_internal): Disable
15904 BMI, BMI2 and TBM instructions for -m16.
15906 2017-07-21 Carl Love <cel@us.ibm.com>
15908 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15909 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15910 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15911 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15912 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15913 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15914 VMULOSW): New enum "unspec" values.
15915 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15916 altivec_vmulosw): New patterns.
15917 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15918 VMULOSW): Add definitions.
15920 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15922 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15923 (qdf24xx): Likewise.
15924 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15925 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15926 (AARCH64_FL_V8_1): Renumber.
15927 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15928 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15929 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15930 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15931 rdma to feature modifiers list.
15933 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15935 PR middle-end/56727
15936 * ipa-visibility (function_and_variable_visibility): Convert
15937 recursive PLT call to direct call if appropriate.
15939 2017-07-21 Andrew Pinski <apinski@cavium.com>
15941 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15942 operand 1 to see if the types precision matches.
15943 * fold-const.c (operand_equal_p): Likewise.
15945 2017-07-21 Richard Biener <rguenther@suse.de>
15947 PR tree-optimization/81303
15948 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15949 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15950 (vect_peeling_hash_get_lowest_cost): Adjust.
15951 (vect_enhance_data_refs_alignment): Likewise. Use
15952 vect_get_peeling_costs_all_drs to compute the penalty for no
15953 peeling to match up costs.
15955 2017-07-21 Richard Biener <rguenther@suse.de>
15957 PR tree-optimization/81500
15958 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15959 we didn't identify a reduction path.
15961 2017-07-21 Tom de Vries <tom@codesourcery.com>
15962 Cesar Philippidis <cesar@codesourcery.com>
15964 PR gcov-profile/81442
15965 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15968 2017-07-21 Tom de Vries <tom@codesourcery.com>
15971 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15973 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15974 nvptx_override_options_after_change.
15976 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15978 * dwarf2out.c (output_file_names): Avoid double testing for
15979 dwarf_version >= 5.
15981 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15983 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15985 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15987 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15989 (try_crossjump_to_edge): Do not punt on partitioned functions.
15991 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15993 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15994 Put all BBs reachable only via paths crossing cold region to cold
15996 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15998 2016-07-21 Richard Biener <rguenther@suse.de>
16000 PR tree-optimization/81303
16001 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16002 into account prologue and epilogue iterations when raising
16003 min_profitable_iters to sth at least covering one vector iteration.
16005 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16007 * config/arm/arm.c (arm_test_cpu_arch_dat):
16010 2017-07-20 Nathan Sidwell <nathan@acm.org>
16012 Remove TYPE_METHODS.
16013 * tree.h (TYPE_METHODS): Delete.
16014 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16015 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16016 (dbxout_type_methods): Scan TYPE_FIELDS.
16017 (dbxout_type): Don't check TYPE_METHODS here.
16018 * function.c (use_register_for_decl): Always ignore register for
16019 class types when not optimizing.
16020 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16021 * tree.c (free_lang_data_in_type): Stitch out member functions and
16022 templates from TYPE_FIELDS.
16023 (build_distinct_type_copy, verify_type_variant,
16024 verify_type): Member fns are on TYPE_FIELDS.
16025 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16026 * tree-pretty-print.c (dump_generic_node): Likewise.
16028 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16031 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16032 V2TImode and V4TImode.
16033 (ix86_expand_vector_extract): Likewise.
16034 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16035 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16036 (ssescalarmode): Handle V4TImode and V2TImode.
16037 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16038 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16039 (VEXTRACTI128_MODE): New mode iterator.
16040 (splitter for *vec_extractv?ti first element): New.
16041 (VEC_INIT_MODE): New mode iterator.
16042 (vec_init<mode>): Consolidate 3 expanders into one using
16043 VEC_INIT_MODE mode iterator.
16045 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16047 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16048 non_spilled_static_chain_regno_p.
16050 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16052 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16054 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16056 * bb-reorder.c (connect_traces): Allow copying of blocks within
16059 2017-07-20 Richard Biener <rguenther@suse.de>
16061 * gimple.h (gimple_phi_result): Add gphi * overload.
16062 (gimple_phi_result_ptr): Likewise.
16063 (gimple_phi_arg): Likewise. Adjust index assert to only
16064 allow actual argument accesses rather than all slots available
16066 (gimple_phi_arg_def): Add gphi * overload.
16067 * tree-phinodes.c (make_phi_node): Initialize only actual
16069 (resize_phi_node): Clear memory not covered by old node,
16070 do not initialize excess argument slots.
16071 (reserve_phi_args_for_new_edge): Initialize new argument slot
16074 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16076 PR tree-optimization/81388
16078 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16080 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16081 by removing computation of may_be_zero.
16083 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16084 Tom de Vries <tom@codesourcery.com>
16086 PR middle-end/81030
16087 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16088 when gimple level profile disagrees with what RTL expander did.
16090 2017-07-20 Richard Biener <rguenther@suse.de>
16092 PR tree-optimization/61171
16093 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16094 (vect_analyze_stmt): Add slp instance parameter.
16095 (vectorizable_reduction): Likewise.
16096 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16097 (vect_is_simple_reduction): Deal with chains not detected
16098 as SLP reduction chain, specifically not properly associated
16099 chains containing a mix of plus/minus.
16100 (get_reduction_op): Remove.
16101 (get_initial_defs_for_reduction): Simplify, pass in whether
16102 this is a reduction chain, pass in the SLP node for the PHIs.
16103 (vect_create_epilog_for_reduction): Get the SLP instance as
16105 (vectorizable_reduction): Get the SLP instance as arg.
16106 During analysis remember the SLP node with the PHIs in the
16107 instance. Simplify getting at the vectorized reduction PHIs.
16108 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16109 through SLP instance.
16110 (vect_slp_analyze_operations): Likewise.
16111 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16112 (vect_transform_stmt): Likewise.
16114 2017-07-20 Tom de Vries <tom@codesourcery.com>
16116 PR tree-optimization/81489
16117 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16118 read of phi arg location to before loop that modifies phi.
16120 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16122 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16125 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16127 PR middle-end/81331
16128 * except.c (execute): Fix ordering issue.
16130 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16132 PR rtl-optimization/81423
16133 * combine.c (make_compound_operation_int): Don't try to optimize
16134 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16136 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16138 PR rtl-optimization/81423
16139 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16140 with a constant that is -1 in the truncated to mode.
16142 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16144 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16145 (determine_unlikely_bbs): ... here.
16146 * predict.h (propagate_unlikely_bbs_forward): Declare.
16147 * cfgexpand.c (pass_expand::execute): Use it.
16148 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16150 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16151 propagate_unlikely_bbs_forward.
16153 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16155 PR middle-end/81331
16156 * except.c (maybe_add_nop_after_section_switch): New function.
16159 2017-07-19 Tom de Vries <tom@codesourcery.com>
16161 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16163 2017-07-19 Tom de Vries <tom@codesourcery.com>
16165 * gimple.h (gimple_phi_arg): Make assert more strict.
16167 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16169 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16170 * config/rs6000/mmintrin.h: New file.
16171 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16173 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16175 PR tree-optimization/81346
16176 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16178 2017-07-19 Tom de Vries <tom@codesourcery.com>
16180 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16182 2017-07-19 Tom de Vries <tom@codesourcery.com>
16184 * config/nvptx/nvptx-modes.def: Add V2DImode.
16185 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16186 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16187 (nvptx_output_mov_insn): Handle lack of mov.b128.
16188 (nvptx_print_operand): Handle 'H' and 'L' codes.
16189 (nvptx_vector_mode_supported): Allow V2DImode.
16190 (nvptx_preferred_simd_mode): New function.
16191 (nvptx_data_alignment): New function.
16192 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16193 nvptx_preferred_simd_mode.
16194 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16196 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16198 2017-07-19 Tom de Vries <tom@codesourcery.com>
16200 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16201 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16202 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16203 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16204 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16205 (mov<VECIM>_insn): New define_insn.
16206 (define_expand "mov<VECIM>): New define_expand.
16208 2017-07-19 Tom de Vries <tom@codesourcery.com>
16210 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16212 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16214 PR tree-optimization/81346
16215 * fold-const.h (fold_div_compare, range_check_type): Declare.
16216 * fold-const.c (range_check_type): New function.
16217 (build_range_check): Use range_check_type.
16218 (fold_div_compare): No longer static, rewritten into
16219 a match.pd helper function.
16220 (fold_comparison): Don't call fold_div_compare here.
16221 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16222 as helper function.
16224 2017-07-19 Nathan Sidwell <nathan@acm.org>
16226 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16227 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16228 * tree.c (find_decls_types_r, verify_type): Use
16229 TYPE_{MIN,MAX}_VALUE_RAW.
16230 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16231 (hash_tree): Likewise.
16232 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16234 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16237 2017-07-18 Tom de Vries <tom@codesourcery.com>
16239 PR middle-end/81464
16240 * omp-expand.c (expand_omp_for_static_chunk): Handle
16241 equal-argument loop exit phi.
16243 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16246 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16247 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16248 operand 2 predicate.
16249 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16250 operand 2 predicate.
16251 (ror,rol -> rorx splitters): Use const_int_operand as
16252 operand 2 predicate.
16254 2017-06-18 Richard Biener <rguenther@suse.de>
16256 PR tree-optimization/81410
16257 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16258 the gap in the ! slp_perm SLP case after each group.
16260 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16262 PR middle-end/81463
16263 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16266 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16268 PR middle-end/81462
16269 * predict.c (set_even_probabilities): Cleanup; do not affect
16270 probabilities that are already known.
16271 (combine_predictions_for_bb): Call even when count is set.
16273 2017-07-18 Nathan Sidwell <nathan@acm.org>
16275 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16278 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16281 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16282 optimization for loop niter analysis.
16284 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16287 * config/avr/avr.c (avr_optimize_casesi): Don't use
16288 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16290 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16292 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16293 body_cost_vec from _vect_peel_extended_info.
16294 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16295 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16298 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16300 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16302 2017-07-18 Richard Biener <rguenther@suse.de>
16304 PR tree-optimization/80620
16305 PR tree-optimization/81403
16306 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16307 info when re-using a VN table entry.
16309 2017-07-18 Richard Biener <rguenther@suse.de>
16311 PR tree-optimization/81418
16312 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16313 vectype_in. Verify that with lane-reducing reduction operations
16314 we have a single def-use cycle.
16316 2017-07-17 Carl Love <cel@us.ibm.com>
16318 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16320 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16321 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16322 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16323 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16324 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16325 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16326 VMULOSW): New enum "unspec" values.
16327 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16328 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16329 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16330 altivec_vmulosw): New patterns.
16331 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16332 VMULOSW): Add definitions.
16333 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16335 * config/alpha/alpha.c: Include predict.h.
16337 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16339 * tree-vrp.c (compare_assert_loc): Fix comparison function
16340 to return predictable results.
16342 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16344 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16346 (subdi3): Likewise.
16347 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16348 * doc/invoke.texi (mexpand-adddi): Update text.
16350 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16352 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16353 that also clobbers the CC register. The old expand code is moved
16355 (*arc_clzsi2): ... here.
16356 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16357 the CC register. The old expand code is moved to ...
16358 (arc_ctzsi2): ... here.
16360 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16362 * config/arc/arc.opt (mindexed-loads): Use initial value
16363 TARGET_INDEXED_LOADS_DEFAULT.
16364 (mauto-modify-reg): Use initial value
16365 TARGET_AUTO_MODIFY_REG_DEFAULT.
16366 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16367 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16368 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16369 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16371 2017-07-17 Martin Liska <mliska@suse.cz>
16374 * opts.c (finish_options): Do not allow -fgnu-tm
16375 w/ -fsanitize={kernel-,}address. Say sorry.
16377 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16380 * tree-loop-distribution.c (classify_partition): Only assert on
16381 numer of iterations.
16382 (merge_dep_scc_partitions): Delete prameter. Update function call.
16383 (distribute_loop): Remove code handling loop with unknown niters.
16384 (pass_loop_distribution::execute): Skip loop with unknown niters.
16386 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16389 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16390 function sort_partitions_by_post_order.
16392 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16394 PR tree-optimization/81374
16395 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16396 the max index of basic blocks, rather than number of basic blocks.
16398 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16400 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16402 (arc_legitimate_pic_operand_p): Likewise.
16403 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16405 (arc_needs_pcl_p): Likewise.
16406 (arc_legitimate_pc_offset_p): Likewise.
16407 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16408 function is also used in constrains.md.
16409 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16410 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16411 PLUS. Only return true/false in known cases, otherwise assert.
16412 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16413 is already called in arc_legitimate_constant_p.
16414 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16416 (LEGITIMATE_PIC_OPERAND_P): Use
16417 arc_raw_symbolic_reference_mentioned_p function.
16418 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16423 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16424 Andrew Burgess <andrew.burgess@embecosm.com>
16426 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16427 (arc_return_address_register): New function.
16428 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16429 (arc_handle_fndecl_attribute): Add naked attribute.
16430 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16431 (TARGET_WARN_FUNC_RETURN): Likewise.
16432 (arc_allocate_stack_slots_for_args): New function.
16433 (arc_warn_func_return): Likewise.
16434 (machine_function): Change type fn_type.
16435 (arc_compute_function_type): Consider new naked function type,
16436 change function return type.
16437 (arc_must_save_register): Adapt to handle new
16438 arc_compute_function_type's return type.
16439 (arc_expand_prologue): Likewise.
16440 (arc_expand_epilogue): Likewise.
16441 (arc_return_address_regs): Delete.
16442 (arc_return_address_register): New function.
16443 (arc_epilogue_uses): Use above function.
16444 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16445 (arc_function_type): Change encoding, add naked type.
16446 (ARC_INTERRUPT_P): Change to handle the new encoding.
16447 (ARC_FAST_INTERRUPT_P): Likewise.
16448 (ARC_NORMAL_P): Define.
16449 (ARC_NAKED_P): Likewise.
16450 (arc_compute_function_type): Delete prototype.
16451 * config/arc/arc.md (in_ret_delay_slot): Use
16452 arc_return_address_register function.
16453 (simple_return): Likewise.
16454 (p_return_i): Likewise.
16456 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16458 PR tree-optimization/81428
16459 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16460 can't be built for those types.
16462 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16464 Remove stuff dead since r239246.
16466 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16467 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16468 (avr_inform_devices): Remove dead stuff.
16470 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16472 * config/arm/arm_neon.h: Fix softp typo.
16474 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16476 PR tree-optimization/81365
16477 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16478 aggregate moves onto bb predecessor edges, make sure there are no
16479 loads that could alias the lhs in between the start of bb and the
16482 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16485 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16486 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16487 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16489 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16491 PR tree-optimization/81396
16492 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16493 (init_symbolic_number): Initialize it to 1.
16494 (perform_symbolic_merge): Add n_ops from both operands into the new
16496 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16497 without base_addr as useless if they need more than one operation.
16498 (bswap_replace): Handle !bswap case for NULL base_addr.
16500 2017-07-17 Tom de Vries <tom@codesourcery.com>
16503 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16506 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16508 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16509 conditional builtin define __FIX_LEON3FT_B2BST.
16511 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16513 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16514 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16517 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16519 PR rtl-optimization/81424
16520 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16521 to remove potential trapping from operands if -fnon-call-exceptions.
16523 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16525 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16526 profile_proability for scalling.
16527 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16529 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16531 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16533 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16535 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16536 fixpoint arithmetics.
16538 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16540 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16541 fixpoint arithmetics.
16543 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16545 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16546 fixpoint arithmetics.
16548 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16550 * profile-count.h (profile_probability::from_reg_br_prob_note,
16551 profile_probability::to_reg_br_prob_note): New functions.
16552 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16553 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16554 * predict.c (probability_reliable_p): Update.
16555 (edge_probability_reliable_p): Update.
16556 (br_prob_note_reliable_p): Update.
16557 (invert_br_probabilities): Update.
16558 (add_reg_br_prob_note): New function.
16559 (combine_predictions_for_insn): Update.
16560 * asan.c (asan_clear_shadow): Update.
16561 * cfgbuild.c (compute_outgoing_frequencies): Update.
16562 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16563 (update_br_prob_note): Update.
16564 (rtl_verify_edges): Update.
16565 (purge_dead_edges): Update.
16566 (fixup_reorder_chain): Update.
16567 * emit-rtl.c (try_split): Update.
16568 * ifcvt.c (cond_exec_process_insns): Update.
16569 (cond_exec_process_if_block): Update.
16570 (dead_or_predicable): Update.
16571 * internal-fn.c (expand_addsub_overflow): Update.
16572 (expand_neg_overflow): Update.
16573 (expand_mul_overflow): Update.
16574 * loop-doloop.c (doloop_modify): Update.
16575 * loop-unroll.c (compare_and_jump_seq): Update.
16576 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16577 * predict.h: Update.
16578 * reorg.c (mostly_true_jump): Update.
16580 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16581 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16582 * config/arc/arc.c: (emit_unlikely_jump): Update.
16583 * config/arm/arm.c: (emit_unlikely_jump): Update.
16584 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16585 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16586 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16587 (ix86_print_operand): Update.
16588 (ix86_split_fp_branch): Update.
16589 (predict_jump): Update.
16590 * config/ia64/ia64.c (ia64_print_operand): Update.
16591 * config/mmix/mmix.c (mmix_print_operand): Update.
16592 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16593 (rs6000_expand_split_stack_prologue): Update.
16594 * config/rs6000/rs6000.c: Update.
16595 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16596 (s390_expand_vec_movstr): Update.
16597 (s390_expand_cs_tdsi): Update.
16598 (s390_expand_split_stack_prologue): Update.
16599 * config/sh/sh.c (sh_print_operand): Update.
16600 (expand_cbranchsi4): Update.
16601 (expand_cbranchdi4): Update.
16602 * config/sparc/sparc.c (output_v9branch): Update.
16603 * config/spu/spu.c (get_branch_target): Update.
16604 (ea_load_store_inline): Update.
16605 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16606 * config/tilepro/tilepro.c: Update.
16608 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16610 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16611 (gimplify_save_expr): Likewise.
16613 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16615 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16617 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16619 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16621 * timevar.def (TV_IPA_FNSUMMARY): Define.
16623 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16625 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16626 to back store errata sensitive sequence from being generated.
16627 (sqrtdf2_fix): Likewise.
16629 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16631 * tree-ssa-threadupdate.c (compute_path_counts,
16632 update_joiner_offpath_counts): Use profile_probability.
16634 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16637 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16639 * config/arm/arm-c.c (arm_cpu_builtins): Define
16640 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16642 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16645 array entries to represent __ieee128 versions of the
16646 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16647 scalar_extract_sig, and scalar_insert_exp built-in functions.
16648 (altivec_resolve_overloaded_builtin): Add special case handling
16649 for the __builtin_scalar_insert_exp function, as represented by
16650 the P9V_BUILTIN_VEC_VSIEDP constant.
16651 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16652 exponent support for __ieee128 argument.
16653 (VSESQP): Add scalar extract signature support for __ieee128
16655 (VSTDCNQP): Add scalar test negative support for __ieee128
16657 (VSIEQP): Add scalar insert exponent support for __int128 argument
16658 with __ieee128 result.
16659 (VSIEQPF): Add scalar insert exponent support for __ieee128
16660 argument with __ieee128 result.
16661 (VSTDCQP): Add scalar test data class support for __ieee128
16663 (VSTDCNQP): Add overload support for scalar test negative with
16664 __ieee128 argument.
16665 (VSTDCQP): Add overload support for scalar test data class
16666 __ieee128 argument.
16667 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16668 UNSPEC_VSX_SXSIGDP.
16669 (UNSPEC_VSX_SIEXPQP): New constant.
16670 (xsxexpqp): New insn for VSX scalar extract exponent quad
16672 (xsxsigqp): New insn for VSX scalar extract significand quad
16674 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16675 precision with floating point argument.
16676 (xststdcqp): New expand for VSX scalar test data class quad
16678 (xststdcnegqp): New expand for VSX scalar test negative quad
16680 (xststdcqp): New insn to match expansions for VSX scalar test data
16681 class quad precision and VSX scalar test negative quad precision.
16682 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16683 special case operand checking to enforce that second operand of
16684 VSX scalar test data class with quad precision argument is a 7-bit
16686 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16687 prototypes and descriptions of __ieee128 versions of
16688 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16689 scalar_test_data_class, and scalar_test_neg built-in functions.
16691 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16693 PR tree-optimization/81162
16694 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16695 replace a negate with an add.
16697 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16699 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16701 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16703 * config/arm/arm-c.c (arm_cpu_builtins): Define
16704 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16706 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16708 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16709 (armv8-r): Set ARM Cortex-R52 as default CPU.
16710 * config/arm/arm-tables.opt: Regenerate.
16711 * config/arm/arm-tune.md: Regenerate.
16712 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16714 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16715 extension for -mcpu=cortex-r52.
16717 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16719 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16720 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16721 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16722 (fp-armv8): Define it as FP_ARMv8 only.
16723 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16724 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16726 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16727 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16728 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16729 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16730 than TARGET_FPU_ARMV8.
16731 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16732 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16733 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16735 * config/arm/neon.md (neon_vrint): Likewise.
16736 (neon_vcvt): Likewise.
16737 (neon_<fmaxmin_op><mode>): Likewise.
16738 (<fmaxmin><mode>3): Likewise.
16739 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16740 * config/arm/predicates.md (arm_cond_move_operator): Check against
16741 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16743 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16745 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16746 to top of function.
16748 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16750 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16751 loop in comment with memset.
16753 2017-07-14 Martin Liska <mliska@suse.cz>
16755 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16756 * dwarf2out.c (is_java): Remove the function.
16757 (output_pubname): Remove usage of the function.
16758 (lower_bound_default): Remove usage of DW_LANG_Java.
16759 (gen_compile_unit_die): Likewise.
16760 * gcc.c: Remove compiler defaults for .java and .zip files.
16761 * gimple-expr.c (remove_suffix): Change as there's no longer
16762 extension than 4-letter one.
16763 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16764 (gimplify_save_expr): Likewise.
16765 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16766 as it's possible even for other languages than Java.
16767 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16768 * lto-opts.c (lto_write_options): Remove reference to Java.
16769 * opts.c (strip_off_ending): Update file extension handling.
16770 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16771 * tree-eh.c (lower_resx): Likewise.
16772 * tree.c (free_lang_data_in_type): Remove dead code.
16773 (find_decls_types_r): Likewise.
16774 (build_common_builtin_nodes): Remove Java from a comment.
16775 (verify_type): Remove dead code.
16776 * varasm.c (assemble_external): Remove Java from a comment.
16778 2017-07-14 Martin Liska <mliska@suse.cz>
16780 * opts.c (finish_options): Add quotes.
16781 (common_handle_option): Likewise.
16783 2017-07-14 Martin Liska <mliska@suse.cz>
16785 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16786 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16787 Remove N_SO_PASCAL.
16788 * dwarf2out.c (lower_bound_default): Do not handle
16790 (gen_compile_unit_die): Likewise.
16791 * gcc.c: Remove default extension binding for GNU Pascal.
16792 * stmt.c: Remove Pascal language from a comment.
16793 * xcoffout.c: Likewise.
16795 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16798 * diagnostic-show-locus.c (fixit_cmp): New function.
16799 (layout::layout): Sort m_fixit_hints.
16800 (column_range::column_range): Assert that the values are valid.
16801 (struct char_span): New struct.
16802 (correction::overwrite): New method.
16803 (struct source_line): New struct.
16804 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16805 calls in terms of classes source_line and char_span, and
16806 correction::overwrite.
16807 (selftest::test_overlapped_fixit_printing_2): New function.
16808 (selftest::diagnostic_show_locus_c_tests): Call it.
16810 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16812 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16813 early if there is no lhs.
16815 2017-07-13 Martin Liska <mliska@suse.cz>
16817 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16818 (gen_reference_type_die): Likewise.
16819 * stor-layout.c: Remove Pascal-related comment.
16821 2017-07-13 Martin Liska <mliska@suse.cz>
16823 * opts.c (finish_options): Add quotes to error messages.
16824 (parse_sanitizer_options): Likewise.
16826 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16828 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16830 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16832 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16834 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16836 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16838 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16840 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16843 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16844 provides the hardware capability bits, define the macro
16845 __BUILTIN_CPU_SUPPORTS__.
16846 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16847 if GLIBC does not provide the hardware capability bits. Add a
16848 gcc_unreachable call if the built-in cpu function is neither
16849 __builtin_cpu_is nor __builtin_cpu_supports.
16850 (rs6000_get_function_versions_dispatcher): Change the warning
16851 that an old GLIBC is used which does not export the capability
16852 bits to be an error.
16853 * doc/extend.texi (target_clones attribute): Document the
16854 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16855 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16856 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16857 the macros defined by GCC if the newer GLIBC is available.
16859 2017-07-12 Jeff Law <law@redhat.com>
16861 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16862 remaining includes slightly.
16863 * config/riscv/riscv-builtins.c: Include profile-count.h.
16865 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16868 * config/avr/avr.c (avr_set_current_function): In diagnostic
16869 messages: Quote keywords and (parts of) identifiers.
16870 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16873 2017-07-12 Carl Love <cel@us.ibm.com>
16875 * config/rs6000/rs6000-c.c: Add support for built-in functions
16876 vector bool char vec_revb (vector bool char);
16877 vector bool short vec_revb (vector short char);
16878 vector bool int vec_revb (vector bool int);
16879 vector bool long long vec_revb (vector bool long long);
16880 * doc/extend.texi: Update the built-in documentation file for the
16881 new built-in functions.
16883 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16885 * config/s390/s390.md: Remove movcc splitter.
16887 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16889 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16890 load/store on condition.
16892 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16895 * config/avr/avr.c (avr_encode_section_info)
16896 [progmem && !TREE_READONLY]: Error if progmem object needs
16899 2017-07-11 Michael Collison <michael.collison@arm.com>
16901 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16904 2017-07-11 Carl Love <cel@us.ibm.com>
16906 * config/rs6000/rs6000-c.c: Add support for builtins
16907 vector unsigned int vec_parity_lsbb (vector signed int);
16908 vector unsigned int vec_parity_lsbb (vector unsigned int);
16909 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16910 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16911 vector unsigned long long vec_parity_lsbb (vector signed long long);
16912 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16913 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16914 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16915 * doc/extend.texi: Update the built-in documentation file for the
16916 new built-in functions.
16918 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16920 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16921 (layout::m_primary_loc): New field.
16922 (layout::layout): Initialize new field. Move location filtering
16923 logic from here to...
16924 (layout::maybe_add_location_range): ...this new method. Add
16925 support for filtering to just the lines already specified by other
16927 (layout::will_show_line_p): New method.
16928 (gcc_rich_location::add_location_if_nearby): New method.
16929 (selftest::test_add_location_if_nearby): New test function.
16930 (selftest::diagnostic_show_locus_c_tests): Call it.
16931 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16934 2017-07-11 Tom de Vries <tom@codesourcery.com>
16936 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16937 (bb_first_real_insn): New function.
16938 (nvptx_single): Add extra initialization of broadcasted condition
16941 2017-07-11 Nathan Sidwell <nathan@acm.org>
16943 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16945 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16947 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16948 Binutils doc as TEXI will mess them up.
16949 * doc/invoke.texi (AVR Options): Same here.
16951 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16953 * config/sparc/sparc.opt (mfix-ut700): New option.
16954 (mfix-gr712rc): Likewise.
16955 (sparc_fix_b2bst): New variable.
16956 * doc/invoke.texi (SPARC options): Document them.
16957 (ARM options): Fix warnings.
16958 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16959 instructions to prevent sequences that can trigger the store-store
16960 errata for certain LEON3FT processors.
16961 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16962 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16963 * config/sparc/sparc.md (fix_b2bst): New attribute.
16964 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16966 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16969 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16971 (*rsqrtsf2_sse): Ditto.
16973 (div<mode>3): Macroize insn from divdf3 and divsf3
16974 using MODEF mode iterator.
16976 2017-07-10 Martin Sebor <msebor@redhat.com>
16978 PR tree-optimization/80397
16979 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16980 instead of testing for equality to INTEGER_TYPE.
16982 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16984 * config.gcc: Remove uclibc from arc target spec.
16986 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16988 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16990 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16993 * lto-wrapper.c (remove_option): New function.
16994 (merge_and_complain): Merge PIC/PIE options more realistically.
16996 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16998 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17002 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17003 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17004 * config.in: Regenerate.
17005 * configure: Regenerate.
17006 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17007 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17008 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17009 (TARGET_GASISR_PROLOGUES): ...target mask.
17010 * common/config/avr/avr-common.c
17011 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17012 Set -mgas-isr-prologues.
17013 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17014 INSERT_PASS_BEFORE for it.
17015 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17016 * config/avr/avr.c (avr_option_override)
17017 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17018 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17019 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17020 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17021 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17023 (make_avr_pass_pre_proep): New function.
17024 (emit_push_sfr) <treg>: Add argument to function and use it
17025 instead of TMP_REG.
17026 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17027 and set machine->gasisr.yes.
17028 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17029 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17030 __gcc_isr.n_pushed to .L__stack_usage.
17031 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17032 (avr_asm_final_postscan_insn): ...this new static function.
17033 * config/avr/avr.h (machine_function)
17034 <is_no_gccisr, use_L__stack_usage>: New fields.
17035 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17036 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17037 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17038 (gasisr, *gasisr): New expander and insn.
17039 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17040 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17041 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17043 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17045 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17048 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17050 Move jump-tables out of .text again.
17053 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17054 (ASM_OUTPUT_ADDR_VEC): New function.
17055 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17056 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17057 INSN_ADDRESSes as asm comment.
17058 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17059 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17060 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17061 * config/avr/avr.md (*tablejump): Adjust comment.
17062 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17063 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17065 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17066 (avr_output_addr_vec): New proto.
17067 (avr_log_t) <insn_addresses>: New field.
17069 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17072 * config/i386/i386.c (ix86_function_arg_advance): Set
17073 outgoing_args_on_stack to true if there are outgoing arguments
17075 (ix86_function_arg): Likewise.
17076 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17077 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17078 * config/i386/i386.h (machine_function): Add
17079 outgoing_args_on_stack.
17081 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17083 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17084 supporting pthreds.
17085 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17087 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17089 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17090 (REAL_H): Remove $(MACHMODE_H).
17091 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17093 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17094 $(MACHMODE_H) and double-int.h.
17095 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17097 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17100 2017-07-07 Andrew Pinski <apinski@cavium.com>
17102 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17103 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17105 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17107 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17108 Add warning if GCC was not configured to link against a GLIBC that
17109 exports the hardware capability bits.
17110 (make_resolver_func): Make resolver function private and not a
17111 COMDAT function. Create the name with clone_function_name instead
17112 of make_unique_name.
17115 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17116 correct operand in doing the split.
17118 2017-07-07 Carl Love <cel@us.ibm.com>
17120 * config/rs6000/rs6000-c: Add support for built-in function
17121 vector unsigned short vec_pack_to_short_fp32 (vector float,
17123 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17124 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17125 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17126 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17127 (convert_4f32_8i16): Add define_expand.
17128 * doc/extend.texi: Update the built-in documentation file for the
17129 new built-in function.
17131 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17133 * config/sparc/m8.md: New file.
17134 * config/sparc/sparc.md: Include m8.md.
17136 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17138 * config/sparc/sparc.opt: New option -mvis4b.
17139 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17140 (sparc_option_override): Handle VIS4B.
17141 (enum sparc_builtins): Define
17142 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17143 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17144 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17145 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17146 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17147 (check_constant_argument): New function.
17148 (sparc_vis_init_builtins): Define builtins
17149 __builtin_vis_dictunpack{8,16,32},
17150 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17151 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17152 __builtin_vis_fpcmpde{8,16,32}shl and
17153 __builtin_vis_fpcmpur{8,16,32}shl.
17154 (sparc_expand_builtin): Check that the constant operands to
17155 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17156 constant and in range.
17157 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17159 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17160 (SPARC_IMM5_P): Likewise.
17161 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17162 (enabled): Handle vis4b.
17163 (UNSPEC_DICTUNPACK): New unspec.
17164 (UNSPEC_FPCMPSHL): Likewise.
17165 (UNSPEC_FPUCMPSHL): Likewise.
17166 (UNSPEC_FPCMPDESHL): Likewise.
17167 (UNSPEC_FPCMPURSHL): Likewise.
17168 (cpu_feature): New CPU feature `vis4b'.
17169 (dictunpack{8,16,32}): New insns.
17170 (FPCSMODE): New mode iterator.
17171 (fpcscond): New code iterator.
17172 (fpcsucond): Likewise.
17173 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17174 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17175 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17176 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17177 * config/sparc/constraints.md: Define constraints `q' for unsigned
17178 2-bit integer constants and `t' for unsigned 5-bit integer
17180 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17182 (imm5_operand_dictunpack16): Likewise.
17183 (imm5_operand_dictunpack32): Likewise.
17184 (imm2_operand): Likewise.
17185 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17186 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17187 ditunpack* and fpcmp*shl builtins.
17189 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17191 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17192 * config.in: Add HAVE_AS_SPARC6 define.
17193 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17195 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17197 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17198 (CPP_CPU_SPEC): Handle m8.
17199 (ASM_CPU_SPEC): Likewise.
17200 * config/sparc/sparc-opts.h (enum processor_type): Add
17202 * config/sparc/sparc.c (m8_costs): New struct.
17203 (sparc_option_override): Handle TARGET_CPU_m8.
17204 (sparc32_initialize_trampoline): Likewise.
17205 (sparc64_initialize_trampoline): Likewise.
17206 (sparc_issue_rate): Likewise.
17207 (sparc_register_move_cost): Likewise.
17208 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17209 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17210 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17211 (CPP_CPU_SPEC): Handle M8.
17212 (ASM_CPU_SPEC): Likewise.
17213 (AS_M8_FLAG): Define.
17214 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17215 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17216 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17218 * configure: Regenerate.
17219 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17222 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17224 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17226 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17227 ("*movdi_insn_sp32"): Do not set v3pipe.
17228 ("*movsi_insn"): Likewise.
17229 ("*movdi_insn_sp64"): Likewise.
17230 ("*movsf_insn"): Likewise.
17231 ("*movdf_insn_sp32"): Likewise.
17232 ("*movdf_insn_sp64"): Likewise.
17233 ("*zero_extendsidi2_insn_sp64"): Likewise.
17234 ("*sign_extendsidi2_insn"): Likewise.
17235 ("*mov<VM32:mode>_insn"): Likewise.
17236 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17237 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17238 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17239 ("<vlop:code><VL:mode>3"): Likewise.
17240 ("*not_<vlop:code><VL:mode>3"): Likewise.
17241 ("*nand<VL:mode>_vis"): Likewise.
17242 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17243 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17244 ("one_cmpl<VL:mode>2"): Likewise.
17245 ("faligndata<VM64:mode>_vis"): Likewise.
17246 ("alignaddrsi_vis"): Likewise.
17247 ("alignaddrdi_vis"): Likweise.
17248 ("alignaddrlsi_vis"): Likewise.
17249 ("alignaddrldi_vis"): Likewise.
17250 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17251 ("bmaskdi_vis"): Likewise.
17252 ("bmasksi_vis"): Likewise.
17253 ("bshuffle<VM64:mode>_vis"): Likewise.
17254 ("cmask8<P:mode>_vis"): Likewise.
17255 ("cmask16<P:mode>_vis"): Likewise.
17256 ("cmask32<P:mode>_vis"): Likewise.
17257 ("pdistn<P:mode>_vis"): Likewise.
17258 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17260 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17262 * config/sparc/sparc.md ("subtype"): New insn attribute.
17263 ("*wrgsr_sp64"): Set insn subtype.
17264 ("*rdgsr_sp64"): Likewise.
17265 ("alignaddrsi_vis"): Likewise.
17266 ("alignaddrdi_vis"): Likewise.
17267 ("alignaddrlsi_vis"): Likewise.
17268 ("alignaddrldi_vis"): Likewise.
17269 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17270 ("fexpand_vis"): Likewise.
17271 ("fpmerge_vis"): Likewise.
17272 ("faligndata<VM64:mode>_vis"): Likewise.
17273 ("bshuffle<VM64:mode>_vis"): Likewise.
17274 ("cmask8<P:mode>_vis"): Likewise.
17275 ("cmask16<P:mode>_vis"): Likewise.
17276 ("cmask32<P:mode>_vis"): Likewise.
17277 ("fchksm16_vis"): Likewise.
17278 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17279 ("fmean16_vis"): Likewise.
17280 ("fp<plusminus_insn>64_vis"): Likewise.
17281 ("<plusminus_insn>v8qi3"): Likewise.
17282 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17283 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17284 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17285 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17286 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17287 ("*movqi_insn"): Likewise.
17288 ("*movhi_insn"): Likewise.
17289 ("*movsi_insn"): Likewise.
17290 ("movsi_pic_gotdata_op"): Likewise.
17291 ("*movdi_insn_sp32"): Likewise.
17292 ("*movdi_insn_sp64"): Likewise.
17293 ("movdi_pic_gotdata_op"): Likewise.
17294 ("*movsf_insn"): Likewise.
17295 ("*movdf_insn_sp32"): Likewise.
17296 ("*movdf_insn_sp64"): Likewise.
17297 ("*zero_extendhisi2_insn"): Likewise.
17298 ("*zero_extendqihi2_insn"): Likewise.
17299 ("*zero_extendqisi2_insn"): Likewise.
17300 ("*zero_extendqidi2_insn"): Likewise.
17301 ("*zero_extendhidi2_insn"): Likewise.
17302 ("*zero_extendsidi2_insn_sp64"): Likewise.
17303 ("ldfsr"): Likewise.
17304 ("prefetch_64"): Likewise.
17305 ("prefetch_32"): Likewise.
17306 ("tie_ld32"): Likewise.
17307 ("tie_ld64"): Likewise.
17308 ("*tldo_ldub_sp32"): Likewise.
17309 ("*tldo_ldub1_sp32"): Likewise.
17310 ("*tldo_ldub2_sp32"): Likewise.
17311 ("*tldo_ldub_sp64"): Likewise.
17312 ("*tldo_ldub1_sp64"): Likewise.
17313 ("*tldo_ldub2_sp64"): Likewise.
17314 ("*tldo_ldub3_sp64"): Likewise.
17315 ("*tldo_lduh_sp32"): Likewise.
17316 ("*tldo_lduh1_sp32"): Likewise.
17317 ("*tldo_lduh_sp64"): Likewise.
17318 ("*tldo_lduh1_sp64"): Likewise.
17319 ("*tldo_lduh2_sp64"): Likewise.
17320 ("*tldo_lduw_sp32"): Likewise.
17321 ("*tldo_lduw_sp64"): Likewise.
17322 ("*tldo_lduw1_sp64"): Likewise.
17323 ("*tldo_ldx_sp64"): Likewise.
17324 ("*mov<VM32:mode>_insn"): Likewise.
17325 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17326 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17328 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17330 * config/sparc/sparc.md ("type"): New insn type viscmp.
17331 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17333 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17334 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17335 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17336 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17338 ("n7_vis_logical_11cycle"): Likewise.
17339 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17340 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17341 * config/sparc/niagara.md ("niag_vis"): Likewise.
17342 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17343 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17345 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17347 * config/sparc/sparc.md: New instruction type `bmask'.
17348 (bmaskdi_vis): Use the `bmask' type.
17349 (bmasksi_vis): Likewise.
17350 * config/sparc/ultra3.md (us3_array): Likewise.
17351 * config/sparc/niagara7.md (n7_array): Likewise.
17352 * config/sparc/niagara4.md (n4_array): Likewise.
17353 * config/sparc/niagara2.md (niag2_vis): Likewise.
17354 (niag3_vis): Likewise.
17355 * config/sparc/niagara.md (niag_vis): Likewise.
17357 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17359 * ipa-comdats.c: Remove optimize check from gate.
17360 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17361 for functions not optimized.
17362 (ipa_fn_summary_read): Skip optimize check.
17363 (ipa_fn_summary_write): Likewise.
17364 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17366 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17368 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17369 (check_callers): Give up on caller which is not optimized.
17370 (inline_small_functions): Likewise.
17371 (ipa_inline): Do not give up when not optimizing.
17372 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17373 away unoptimizes cdtors.
17374 (whole_program_function_and_variable_visibility): Do
17375 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17376 * ipa.c (process_references): Do not check optimize.
17377 (symbol_table::remove_unreachable_nodes): Update optimize check.
17378 (set_writeonly_bit): Update optimize check.
17379 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17380 (pass_ipa_single_use::gate): Remove.
17382 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17384 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17385 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17386 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17387 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17388 permute_load, permute_store, adjust_extract, adjust_splat,
17389 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17390 replace_swap_with_copy, dump_swap_insn_table,
17391 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17392 recombine_lvx_pattern, recombine_stvx_pattern,
17393 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17394 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17395 to file rs6000-p8swap.c.
17396 * config/rs6000/rs6000-p8swap.c: New file.
17397 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17398 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17399 and rs6000*-*-* targets.
17401 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17403 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17405 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17407 * lto-wrapper.c (merge_and_complain): Do not merge
17408 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17409 fsigned_zeros, ftrapping_math, fwrapv.
17410 (append_compiler_options): Do not track these options.
17411 (append_linker_options): Likewie
17413 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17415 * cgraphunit.c (cgraph_node::finalize_function): When
17416 !flag_toplevel_reorde set no_reorder flag.
17417 (varpool_node::finalize_decl): Likewise.
17418 (symbol_table::compile): Drop no toplevel reorder path.
17420 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17422 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17423 edges; zero probability is not better than uninitialized.
17425 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17427 * asan.h (asan_sanitize_allocas_p): Declare.
17428 * asan.c (asan_sanitize_allocas_p): New function.
17429 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17430 (handle_builtin_alloca): Likewise.
17431 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17432 if !asan_sanitize_allocas_p.
17433 * params.def (asan-instrument-allocas): Add new option.
17434 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17435 * opts.c (common_handle_option): Disable allocas sanitization for
17438 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17440 * asan.c: Include gimple-fold.h.
17441 (get_last_alloca_addr): New function.
17442 (handle_builtin_stackrestore): Likewise.
17443 (handle_builtin_alloca): Likewise.
17444 (asan_emit_allocas_unpoison): Likewise.
17445 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17446 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17447 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17448 (instrument_builtin_call): Pass gimple iterator to
17449 get_mem_refs_of_builtin_call.
17450 (last_alloca_addr): New global.
17451 * asan.h (asan_emit_allocas_unpoison): Declare.
17452 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17453 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17454 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17455 if function calls alloca.
17456 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17457 * gimple-fold.h (replace_call_with_value): Declare.
17458 * internal-fn.c: Include asan.h.
17459 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17460 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17462 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17464 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17465 (C_SELFTEST_FLAGS): New.
17466 (CPP_SELFTEST_FLAGS): New.
17467 (SELFTEST_DEPS): New, from deps of s-selftest.
17468 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17469 (CPP_SELFTEST_DEPS): New.
17470 (selftest): Add dependency on s-selftest-c++.
17471 (s-selftest): Rename to...
17472 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17473 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17474 than SELFTEST_FLAGS.
17475 (selftest-gdb): Rename to...
17476 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17478 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17479 (selftest-valgrind): Rename to...
17480 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17482 (selftest-valgrind): Reintroduce as an alias for
17483 selftest-c-valgrind.
17484 (s-selftest-c++): New.
17485 (selftest-c++-gdb): New.
17486 (selftest-c++-valgrind): New.
17488 2017-07-06 Olivier Hainque <hainque@adacore.com>
17490 * gcc.c (process_command): When deciding if undefined variables
17491 should be ignored when processing specs, accept "gcc -v" as well.
17493 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17495 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17496 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17498 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17500 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17501 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17502 * config/arm/arm-tables.opt: Regenerate.
17503 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17505 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17507 2017-07-06 Carl Love <cel@us.ibm.com>
17509 * ChangeLog: Clean up from mid air collision
17511 2017-07-06 Carl Love <cel@us.ibm.com>
17513 * config/rs6000/rs6000-c.c: Add support for built-in functions
17514 vector signed int vec_subc (vector signed int, vector signed int);
17515 vector signed __int128 vec_subc (vector signed __int128,
17516 vector signed __int128);
17517 vector unsigned __int128 vec_subc (vector unsigned __int128,
17518 vector unsigned __int128);
17519 vector signed int vec_sube (vector signed int, vector signed int,
17520 vector signed int);
17521 vector unsigned int vec_sube (vector unsigned int,
17522 vector unsigned int,
17523 vector unsigned int);
17524 vector signed __int128 vec_sube (vector signed __int128,
17525 vector signed __int128,
17526 vector signed__int128);
17527 vector unsigned __int128 vec_sube (vector unsigned __int128,
17528 vector unsigned __int128,
17529 vector unsigned __int128);
17530 vector signed int vec_subec (vector signed int, vector signed int,
17531 vector signed int);
17532 vector unsigned int vec_subec (vector unsigned int,
17533 vector unsigned int,
17534 vector unsigned int);
17535 vector signed __int128 vec_subec (vector signed __int128,
17536 vector signed __int128,
17537 vector signed__int128);
17538 vector unsigned __int128 vec_subec (vector unsigned __int128,
17539 vector unsigned __int128,
17540 vector unsigned __int128);
17541 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17542 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17543 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17544 BU_ALTIVEC_OVERLOAD_X definitions.
17545 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17546 * doc/extend.texi: Update the built-in documentation file for the new
17547 built-in functions.
17549 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17552 * diagnostic-show-locus.c (layout::layout): Use start and finish
17553 spelling location for the start and finish of each range.
17554 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17555 Add unused aspect param.
17556 * input.c (expand_location_1): Add "aspect" param, and use it
17557 to access the correct part of the location.
17558 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17560 (expand_location_to_spelling_point): Likewise.
17561 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17562 param, and pass it to expand_location_1.
17564 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17566 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17567 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17568 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17569 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17570 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17571 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17572 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17573 _mm_maskz_getmant_ss): New intrinsics.
17574 (__builtin_ia32_getexpss128_mask): Changed to ...
17575 __builtin_ia32_getexpss128_round ... this.
17576 (__builtin_ia32_getexpsd128_mask): Changed to ...
17577 __builtin_ia32_getexpsd128_round ... this.
17578 * config/i386/i386-builtin-types.def
17579 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17580 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17581 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17582 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17583 __builtin_ia32_getmantss_mask_round): New builtins.
17584 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17585 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17586 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17587 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17588 * config/i386/sse.md
17589 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17590 avx512f_sgetexp<mode><mask_scalar_name>
17591 <round_saeonly_scalar_name> ... this.
17592 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17593 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17594 vgetexp<ssescalarmodesuffix>
17595 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17596 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17597 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17598 avx512f_vgetmant<mode><mask_scalar_name>
17599 <round_saeonly_scalar_name> ... this.
17600 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17601 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17602 vgetmant<ssescalarmodesuffix>
17603 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17604 %0<mask_scalar_operand4>, %1, %2
17605 <round_saeonly_scalar_mask_op4>, %3} ... this.
17606 * config/i386/subst.md (mask_scalar_operand4,
17607 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17608 round_saeonly_scalar_nimm_predicate): New subst attributes.
17610 2017-07-06 Julia Koval <julia.koval@intel.com>
17612 * config/i386/i386.c (ix86_erase_embedded_rounding):
17613 Remove code for old rounding pattern.
17615 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17617 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17619 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17621 * doc/sourcebuild.texi (Test Directives, Variants of
17622 dg-require-support): Add documentation for dg-require-stack-check.
17624 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17626 * config/i386/subst.md (mask_scalar, round_scalar,
17627 round_saeonly_scalar): New meta-templates.
17628 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17629 round_scalar_mask_operand3, round_scalar_mask_op3,
17630 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17631 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17632 round_saeonly_scalar_constraint,
17633 round_saeonly_scalar_prefix): New subst attribute.
17634 * config/i386/sse.md
17635 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17636 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17637 <round_scalar_name> ... this.
17638 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17639 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17640 <round_scalar_name> ... this.
17641 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17642 <sse>_vm<code><mode>3<mask_scalar_name>
17643 <round_saeonly_scalar_name> ... this.
17644 (v<plusminus_mnemonic><ssescalarmodesuffix>
17645 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17646 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17647 v<plusminus_mnemonic><ssescalarmodesuffix>
17648 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17649 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17650 (v<multdiv_mnemonic><ssescalarmodesuffix>
17651 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17652 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17653 v<multdiv_mnemonic><ssescalarmodesuffix>
17654 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17655 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17656 (v<maxmin_float><ssescalarmodesuffix>
17657 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17658 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17659 v<maxmin_float><ssescalarmodesuffix>
17660 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17661 %0<mask_scalar_operand3>, %1, %<iptr>2
17662 <round_saeonly_scalar_mask_op3>} ... this.
17664 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17666 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17667 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17669 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17670 Alan Hayward <alan.hayward@arm.com>
17671 David Sherwood <david.sherwood@arm.com>
17673 * combine.c (simplify_if_then_else): Remove "enum" before
17675 * compare-elim.c (can_eliminate_compare): Likewise.
17676 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17678 (aarch64_lookup_simd_builtin_type): Likewise.
17679 (aarch64_simd_builtin_type): Likewise.
17680 (aarch64_init_simd_builtin_types): Likewise.
17681 (aarch64_simd_expand_args): Likewise.
17682 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17684 (aarch64_reverse_mask): Likewise.
17685 (aarch64_simd_emit_reg_reg_move): Likewise.
17686 (aarch64_gen_adjusted_ldpstp): Likewise.
17687 (aarch64_ccmp_mode_to_code): Likewise.
17688 (aarch64_operands_ok_for_ldpstp): Likewise.
17689 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17690 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17692 (aarch64_min_divisions_for_recip_mul): Likewise.
17693 (aarch64_reassociation_width): Likewise.
17694 (aarch64_get_condition_code_1): Likewise.
17695 (aarch64_simd_emit_reg_reg_move): Likewise.
17696 (aarch64_simd_attr_length_rglist): Likewise.
17697 (aarch64_reverse_mask): Likewise.
17698 (aarch64_operands_ok_for_ldpstp): Likewise.
17699 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17700 (aarch64_gen_adjusted_ldpstp): Likewise.
17701 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17703 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17704 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17705 (arm_lookup_simd_builtin_type): Likewise.
17706 (arm_simd_builtin_type): Likewise.
17707 (arm_init_simd_builtin_types): Likewise.
17708 (arm_expand_builtin_args): Likewise.
17709 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17710 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17711 (ft32_setup_incoming_varargs): Likewise.
17712 (ft32_function_arg): Likewise.
17713 (ft32_function_arg_advance): Likewise.
17714 (ft32_pass_by_reference): Likewise.
17715 (ft32_arg_partial_bytes): Likewise.
17716 (ft32_valid_pointer_mode): Likewise.
17717 (ft32_addr_space_pointer_mode): Likewise.
17718 (ft32_addr_space_legitimate_address_p): Likewise.
17719 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17721 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17722 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17723 (iamcu_alignment): Likewise.
17724 (canonicalize_vector_int_perm): Likewise.
17725 (ix86_noce_conversion_profitable_p): Likewise.
17726 (ix86_mpx_bound_mode): Likewise.
17727 (ix86_operands_ok_for_move_multiple): Likewise.
17728 * config/microblaze/microblaze-protos.h
17729 (microblaze_expand_conditional_branch_reg): Likewise.
17730 * config/microblaze/microblaze.c
17731 (microblaze_expand_conditional_branch_reg): Likewise.
17732 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17734 (rs6000_reassociation_width): Likewise.
17735 (rs6000_invalid_binary_op): Likewise.
17736 (fusion_p9_p): Likewise.
17737 (emit_fusion_p9_load): Likewise.
17738 (emit_fusion_p9_store): Likewise.
17739 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17741 (riscv_hard_regno_mode_ok_p): Likewise.
17742 (riscv_address_insns): Likewise.
17743 (riscv_split_symbol): Likewise.
17744 (riscv_legitimize_move): Likewise.
17745 (riscv_function_value): Likewise.
17746 (riscv_hard_regno_nregs): Likewise.
17747 (riscv_expand_builtin): Likewise.
17748 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17749 (riscv_build_integer): Likewise.
17750 (riscv_split_integer): Likewise.
17751 (riscv_legitimate_constant_p): Likewise.
17752 (riscv_cannot_force_const_mem): Likewise.
17753 (riscv_regno_mode_ok_for_base_p): Likewise.
17754 (riscv_valid_base_register_p): Likewise.
17755 (riscv_valid_offset_p): Likewise.
17756 (riscv_valid_lo_sum_p): Likewise.
17757 (riscv_classify_address): Likewise.
17758 (riscv_legitimate_address_p): Likewise.
17759 (riscv_address_insns): Likewise.
17760 (riscv_load_store_insns): Likewise.
17761 (riscv_force_binary): Likewise.
17762 (riscv_split_symbol): Likewise.
17763 (riscv_force_address): Likewise.
17764 (riscv_legitimize_address): Likewise.
17765 (riscv_move_integer): Likewise.
17766 (riscv_legitimize_const_move): Likewise.
17767 (riscv_legitimize_move): Likewise.
17768 (riscv_address_cost): Likewise.
17769 (riscv_subword): Likewise.
17770 (riscv_output_move): Likewise.
17771 (riscv_canonicalize_int_order_test): Likewise.
17772 (riscv_emit_int_order_test): Likewise.
17773 (riscv_function_arg_boundary): Likewise.
17774 (riscv_pass_mode_in_fpr_p): Likewise.
17775 (riscv_pass_fpr_single): Likewise.
17776 (riscv_pass_fpr_pair): Likewise.
17777 (riscv_get_arg_info): Likewise.
17778 (riscv_function_arg): Likewise.
17779 (riscv_function_arg_advance): Likewise.
17780 (riscv_arg_partial_bytes): Likewise.
17781 (riscv_function_value): Likewise.
17782 (riscv_pass_by_reference): Likewise.
17783 (riscv_setup_incoming_varargs): Likewise.
17784 (riscv_print_operand): Likewise.
17785 (riscv_elf_select_rtx_section): Likewise.
17786 (riscv_save_restore_reg): Likewise.
17787 (riscv_for_each_saved_reg): Likewise.
17788 (riscv_register_move_cost): Likewise.
17789 (riscv_hard_regno_mode_ok_p): Likewise.
17790 (riscv_hard_regno_nregs): Likewise.
17791 (riscv_class_max_nregs): Likewise.
17792 (riscv_memory_move_cost): Likewise.
17793 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17794 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17795 (rl78_addr_space_address_mode): Likewise.
17796 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17798 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17799 (rs6000_reassociation_width): Likewise.
17800 (rs6000_invalid_binary_op): Likewise.
17801 (fusion_p9_p): Likewise.
17802 (emit_fusion_p9_load): Likewise.
17803 (emit_fusion_p9_store): Likewise.
17804 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17805 (ok_for_simple_move_operands): Likewise.
17806 (ok_for_simple_move_strict_operands): Likewise.
17807 (ok_for_simple_arith_logic_operands): Likewise.
17808 (visium_legitimize_reload_address): Likewise.
17809 (visium_select_cc_mode): Likewise.
17810 (output_cbranch): Likewise.
17811 (visium_split_double_move): Likewise.
17812 (visium_expand_copysign): Likewise.
17813 (visium_expand_int_cstore): Likewise.
17814 (visium_expand_fp_cstore): Likewise.
17815 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17816 (visium_function_arg): Likewise.
17817 (visium_function_arg_advance): Likewise.
17818 (visium_libcall_value): Likewise.
17819 (visium_setup_incoming_varargs): Likewise.
17820 (visium_legitimate_constant_p): Likewise.
17821 (visium_legitimate_address_p): Likewise.
17822 (visium_legitimize_address): Likewise.
17823 (visium_secondary_reload): Likewise.
17824 (visium_register_move_cost): Likewise.
17825 (visium_memory_move_cost): Likewise.
17826 (prepare_move_operands): Likewise.
17827 (ok_for_simple_move_operands): Likewise.
17828 (ok_for_simple_move_strict_operands): Likewise.
17829 (ok_for_simple_arith_logic_operands): Likewise.
17830 (visium_function_value_1): Likewise.
17831 (rtx_ok_for_offset_p): Likewise.
17832 (visium_legitimize_reload_address): Likewise.
17833 (visium_split_double_move): Likewise.
17834 (visium_expand_copysign): Likewise.
17835 (visium_expand_int_cstore): Likewise.
17836 (visium_expand_fp_cstore): Likewise.
17837 (visium_split_cstore): Likewise.
17838 (visium_select_cc_mode): Likewise.
17839 (visium_split_cbranch): Likewise.
17840 (output_cbranch): Likewise.
17841 (visium_print_operand_address): Likewise.
17842 * expmed.c (flip_storage_order): Likewise.
17843 * expmed.h (emit_cstore): Likewise.
17844 (flip_storage_order): Likewise.
17845 * genrecog.c (validate_pattern): Likewise.
17846 * hsa-gen.c (gen_hsa_addr): Likewise.
17847 * internal-fn.c (expand_arith_overflow): Likewise.
17848 * ira-color.c (allocno_copy_cost_saving): Likewise.
17849 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17850 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17851 (process_invariant_for_inheritance): Likewise.
17852 * lra-eliminations.c (move_plus_up): Likewise.
17853 * omp-low.c (lower_oacc_reductions): Likewise.
17854 * simplify-rtx.c (simplify_subreg): Likewise.
17855 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17856 (TARGET_CHKP_BOUND_MODE): Likewise..
17857 * targhooks.c (default_chkp_bound_mode): Likewise.
17858 (default_setup_incoming_vararg_bounds): Likewise.
17859 * targhooks.h (default_chkp_bound_mode): Likewise.
17860 (default_setup_incoming_vararg_bounds): Likewise.
17861 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17862 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17863 (have_whole_vector_shift): Likewise.
17864 * tree-vect-stmts.c (vectorizable_load): Likewise.
17865 * doc/tm.texi: Regenerate.
17867 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17869 Graceful degrade if Binutils PR21472 is not available.
17872 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17873 .rodata in flash test fails.
17874 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17875 * confgure: Regenerate.
17876 * config.in: Regenerate.
17877 * config/avr/avr.c (avr_asm_named_section)
17878 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17879 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17880 (avr_asm_init_sections): Same.
17882 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17884 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17885 (fma<VH:mode>4_intrinsic): Likewise.
17886 (*fmsub<VCVTF:mode>4): Likewise.
17887 (*fmsub<VH:mode>4_intrinsic): Likewise.
17889 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17892 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17893 Don't depend on "optimize > 0".
17894 (out_movhi_r_mr, out_movqi_mr_r): Same.
17895 (out_movhi_mr_r, out_movqi_r_mr): Same.
17896 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17897 io_address_operand on "optimize > 0".
17899 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17901 * tree-loop-distribution.c: Add general explanantion on the pass.
17902 (generate_loops_for_partition): Mark distributed loop.
17903 (pg_add_dependence_edges): New parameter. Handle alias data
17904 dependence specially and record it in the parameter if asked.
17905 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17906 (init_partition_graph_vertices, add_partition_graph_edge): New.
17907 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17908 (free_partition_graph_vdata, build_partition_graph): New.
17909 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17910 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17911 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17912 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17913 (version_for_distribution_p, finalize_partitions): New.
17914 (distribute_loop): Handle alias data dependence specially. Factor
17915 out loop fusion code as functions and call these functions.
17917 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17919 * tree-loop-distribution.c (classify_partition): New parameter and
17920 better handle reduction statement.
17921 (rdg_build_partitions): Revise comment.
17922 (distribute_loop): Compute statements in all partitions and pass it
17923 to classify_partition.
17925 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17927 * tree-loop-distribution.c (enum partition_type): New.
17928 (struct partition): New field type.
17929 (partition_merge_into): Add parameter. Update partition type.
17930 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17931 (build_rdg_partition_for_vertex): Compute partition type.
17932 (rdg_build_partitions): Dump partition type.
17933 (distribute_loop): Update calls to partition_merge_into.
17935 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17937 * tree-loop-distribution.c (struct ddr_hasher): New.
17938 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17940 (classify_partition): Call get_data_dependence.
17941 (pg_add_dependence_edges): Ditto.
17942 (distribute_loop): Release data dependence hash table.
17944 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17946 * tree-loop-distribution.c (ref_base_address): Delete.
17947 (similar_memory_accesses): Rename ...
17948 (share_memory_accesses): ... to this. Check if partitions access
17949 the same memory reference.
17950 (distribute_loop): Call share_memory_accesses.
17952 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17954 * tree-loop-distribution.c (struct partition): New field recording
17955 its data reference.
17956 (partition_alloc, partition_free): Init and release data refs.
17957 (partition_merge_into): Merge data refs.
17958 (build_rdg_partition_for_vertex): Collect data refs for partition.
17959 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17961 (distribute_loop): Remve data refs from vertice data of partition
17964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17966 * tree-loop-distribution.c (params.h): Include header file.
17967 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17968 (datarefs_vec): New global var.
17969 (create_rdg_vertices): Use datarefs_vec directly.
17970 (free_rdg): Don't free data references.
17971 (build_rdg): Update use. Don't free data references.
17972 (distribute_loop): Compute global variable for data references.
17973 Bail out if there are too many data references.
17975 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17977 * tree-loop-distribution.c (loop_nest): New global var.
17978 (build_rdg): Use loop directly, rather than loop nest.
17979 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17981 (distribute_loop): Compute global variable loop nest. Update use.
17983 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17985 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17986 (partition_merge_into): New parameter. Dump reason for fusion.
17987 (distribute_loop): Update use of partition_merge_into.
17989 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17991 * tree-loop-distribution.c (bb_top_order_index): New.
17992 (bb_top_order_index_size, bb_top_order_cmp): New.
17993 (stmts_from_loop): Use topological order.
17994 (pass_loop_distribution::execute): Compute and release topological
17995 order for basic blocks.
17997 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17999 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18002 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18004 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18005 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18006 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18007 * internal-fn.def (LOOP_DIST_ALIAS): New.
18008 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18009 (fold_loop_internal_call): ... this.
18010 (vect_loop_dist_alias_call): New function.
18011 (set_uid_loop_bbs): Call fold_loop_internal_call.
18012 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18015 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18018 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18019 Require dead FLAGS_REG at the beginning of a peephole.
18021 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18024 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18025 arguments in the call to __builtin_ia32_sbb_u32.
18026 (_subborrow_u64): Swap _X and _Y arguments in the call to
18027 __builtin_ia32_sbb_u64.
18029 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18032 * tree-vrp.c (compare_assert_loc): Turn into a function template
18033 with stable template parameter. Only test if a->e is NULL,
18034 !a->e == !b->e has been verified already. Use e == NULL or
18035 e != NULL instead of e or ! e tests. If stable is true, don't use
18036 iterative_hash_expr, on the other side allow a or b or both NULL
18037 and sort the NULLs last.
18038 (process_assert_insertions): Sort using compare_assert_loc<false>
18039 instead of compare_assert_loc, later sort using
18040 compare_assert_loc<true> before calling process_assert_insertions_for
18041 in a loop. Use break instead of continue once seen NULL pointer.
18043 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18045 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18046 Cortex-R7 and Cortex-R8 processors.
18048 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18050 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18051 uninitialized while src is not.
18053 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18055 * common/config/arm/arm-common.c: Adjust include path for
18057 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18058 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18059 (arm-cpu-data.h): Likewise.
18060 (arm-cpu-cdata.h): Likewise.
18061 * config/arm/arm-cpu.h: Delete.
18062 * config/arm/arm-cpu-cdata.h: Delete.
18063 * config/arm/arm-cpu-data.h: Delete.
18065 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18067 * config/arm/arm-cpus.in (cortex-a55): New.
18068 (cortex-a75): Likewise.
18069 (cortex-a75.cortex-a55): Likewise.
18070 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18072 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18073 * config/arm/arm-cpu-cdata.h: Regenerate.
18074 * config/arm/arm-cpu-data.h: Regenerate.
18075 * config/arm/arm-cpu.h: Regenerate.
18076 * config/arm/arm-tables.opt: Regenerate.
18077 * config/arm/arm-tune.md: Regenerate.
18079 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18081 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18083 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18085 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18088 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18090 PR tree-optimization/81292
18091 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18092 full_string_p, also call adjust_related_strinfos if the adjustment
18093 is simple, otherwise invalidate related strinfos.
18095 2017-07-04 Martin Liska <mliska@suse.cz>
18098 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18099 newly created variable as DECL_IGNORED_P.
18101 2017-07-04 Martin Liska <mliska@suse.cz>
18104 * ipa-inline.c (inline_small_functions):
18105 Use xstrdup_for_dump.
18107 2017-07-04 Tom de Vries <tom@codesourcery.com>
18109 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18111 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18114 * config/darwin.c (darwin_function_switched_text_sections):
18117 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18119 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18121 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18123 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18125 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18127 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18128 min_profitable_iters, and th as inclusive lower bounds.
18129 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18130 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18131 for min_profitable_iters and min_profitable_estimate.
18132 (vect_transform_loop): Treat th as an inclusive lower bound.
18133 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18135 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18138 * config/darwin.c (darwin_function_switched_text_sections):
18139 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18140 in two pieces, and suppress the use of buf.
18142 2017-07-03 Nathan Sidwell <nathan@acm.org>
18144 * hash-table.h (hash_table_mod1): Fix indentation.
18146 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18148 PR middle-end/81290
18149 * predict.c (force_edge_cold): Be more careful about propagation
18151 * profile-count.h (profile_probability::guessed,
18152 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18154 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18156 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18158 * doc/invoke.texi (rcpc architecture extension): Document it.
18160 2017-07-03 Richard Biener <rguenther@suse.de>
18162 PR tree-optimization/60510
18163 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18164 the scalar reduction PHI and use it.
18165 (vectorizable_reduction): Properly guard the single_defuse_cycle
18166 path for non-SLP reduction chains where we cannot use it.
18167 Rework reduc_def/index and vector type deduction. Rework
18168 vector operand gathering during reduction op code-gen.
18169 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18170 chains dissolve the chain and leave it to non-SLP reduction
18173 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18175 * tree-data-ref.h (dr_alignment): Declare.
18176 * tree-data-ref.c (dr_alignment): New function.
18177 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18178 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18180 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18182 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18184 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18185 and base_misalignment fields.
18186 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18187 * tree-data-ref.c: Include builtins.h.
18188 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18189 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18190 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18191 * tree-vect-data-refs.c: Include tree-cfg.h.
18192 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18193 fields instead of calculating an alignment here.
18194 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18195 innermost_loop_behavior fields.
18197 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18199 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18201 (DR_STEP_ALIGNMENT): New macro.
18202 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18203 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18204 (create_data_ref): Print it.
18205 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18206 to tell whether the step preserves vector (mis)alignment.
18207 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18208 Move the check for an integer step and generalise to all INTEGER_CST.
18209 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18210 Print the outer step alignment.
18212 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18214 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18215 with offset_alignment.
18216 (DR_ALIGNED_TO): Delete.
18217 (DR_OFFSET_ALIGNMENT): New macro.
18218 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18219 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18220 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18221 (create_data_ref): Likewise.
18222 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18223 (vect_analyze_data_refs): Likewise.
18224 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18225 creating dummy innermost behavior.
18227 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18229 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18230 with a "innermost_loop_behavior *" and refeence tree.
18231 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18232 (create_data_ref): Update call accordingly.
18233 * tree-predcom.c (find_looparound_phi): Likewise.
18235 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18237 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18238 fields with dr_wrt_vec_loop.
18239 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18240 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18241 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18242 (vect_dr_behavior): New function.
18243 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18244 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18245 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18246 track whether the step preserves the misalignment.
18247 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18248 Use vect_dr_behavior.
18249 (vect_setup_realignment): Update call accordingly.
18250 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18251 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18252 call to vect_create_addr_base_for_vector_ref.
18253 (vect_create_cond_for_align_checks): Likewise.
18254 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18255 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18256 (vect_recog_mask_conversion_pattern): Likewise.
18257 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18258 (new_stmt_vec_info): Remove redundant zeroing.
18260 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18262 * common/config/arm/arm-common.c (arm_be8_option): New function.
18263 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18264 (ISA_ARMv6): Add isa_bit_be8.
18265 * config/arm/arm.h (arm_be8_option): Add prototype.
18266 (BE8_SPEC_FUNCTION): New define.
18267 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18268 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18269 (mlittle-endian): Similarly.
18270 (mbe8, mbe32): New options.
18271 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18272 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18274 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18276 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18278 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18280 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18281 (cleanup_tree_cfg_bb): Use it.
18282 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18284 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18286 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18289 * loop-doloop.c (add_test): Update profile.
18291 2017-07-03 Martin Liska <mliska@suse.cz>
18294 * sanopt.c (rewrite_usage_of_param): New function.
18295 (sanitize_rewrite_addressable_params): Likewise.
18296 (pass_sanopt::execute): Call rewrite_usage_of_param.
18298 2017-07-03 Richard Biener <rguenther@suse.de>
18300 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18301 back to using VIEW_CONVERT_EXPR.
18303 2017-07-03 Martin Liska <mliska@suse.cz>
18306 * doc/extend.texi: Document when a resolver function is
18307 generated for target_clones.
18309 2017-07-03 Martin Liska <mliska@suse.cz>
18311 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18312 and shadow memory of auto variables which are subject of
18313 use-after-scope sanitization.
18314 (asan_expand_mark_ifn): Add do set only when is_poison.
18316 2016-07-03 Richard Biener <rguenther@suse.de>
18318 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18320 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18321 (vectorizable_reduction): Perform reduction PHI creation when
18322 visiting a reduction PHI and adjust and simplify code generation
18323 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18324 (vect_transform_loop): Visit reduction PHIs.
18325 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18326 defs into the SLP tree.
18327 (vect_build_slp_tree): Reduction defs terminate the recursion.
18328 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18330 (vect_get_vec_defs_for_stmt_copy): Export.
18331 (vect_get_vec_defs): Likewise.
18332 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18334 (vect_get_vec_defs_for_stmt_copy): Declare.
18335 (vect_get_vec_defs): Likewise.
18337 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18339 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18340 parameter with a "loop" parameter and use it instead of the
18341 loop containing DR_STMT. Don't check simple_iv when doing
18342 BB analysis. Describe the two analysis modes in the comment.
18344 2017-07-03 Tom de Vries <tom@codesourcery.com>
18346 PR tree-optimization/69468
18347 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18348 (find_same_succ_bb): Handle ignore_edge_flags.
18350 2017-07-03 Tom de Vries <tom@codesourcery.com>
18352 PR tree-optimization/81192
18353 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18355 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18357 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18359 2017-07-03 Tom de Vries <tom@codesourcery.com>
18361 PR tree-optimization/81192
18362 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18363 BB_SAME_SUCC (bb) == NULL.
18365 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18367 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18370 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18372 * dumpfile.c: Include profile-count.h
18373 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18375 (insert_cond_bb): Update profile.
18376 * tree-cfg.h (insert_cond_bb): Update prototype.
18377 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18378 * tree-dump.c: Do not include tree-cfg.
18380 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18382 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18384 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18386 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18389 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18391 * tree-complex.c (expand_complex_div_wide): update profile.
18393 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18394 Alan Hayward <alan.hayward@arm.com>
18395 David Sherwood <david.sherwood@arm.com>
18397 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18398 (CORETYPES_H): New define.
18399 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18400 (insn-modes-inline.h, s-modes-inline-h): New rules.
18401 (generated_files): Add insn-modes-inline.h.
18402 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18403 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18404 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18405 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18406 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18407 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18408 (build/gencodes.o, build/genconditions.o): Likewise.
18409 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18410 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18411 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18412 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18413 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18414 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18415 * coretypes.h: Include everything up to real.h for generators.
18416 Include insn-modes.h first. Include wide-int-print.h after
18417 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18418 * machmode.h: Don't include insn-modes.h here.
18419 * function-tests.c: Remove includes of signop.h, machmode.h,
18420 double-int.h and wide-int.h.
18422 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18424 * optc-save-gen.awk: Likewise.
18425 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18426 * godump.c: Remove include of wide-int-print.h.
18427 * pretty-print.h: Likewise.
18428 * wide-int-print.cc: Likewise.
18429 * wide-int.cc: Likewise.
18430 * hash-map-tests.c: Remove include of signop.h.
18431 * hash-set-tests.c: Likewise.
18432 * rtl-tests.c: Likewise.
18433 * mkconfig.sh: Remove include of machmode.h.
18434 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18436 (emit_insn_modes_inline_h): ...this new function. Emit the code
18437 into an insn-modes-inline.h header file, adding appropriate
18438 include guards and end comments.
18439 (emit_insn_modes_c_header): Remove include of machmode.h.
18440 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18442 (main): Handle -i flag and call emit_insn_modes_inline_h when
18445 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18447 * tree-ssa-strlen.c (strinfo): Rename the length field to
18448 nonzero_chars. Add a full_string_p field.
18449 (compare_nonzero_chars, zero_length_string_p): New functions.
18450 (get_addr_stridx): Add an offset_out parameter.
18451 Use compare_nonzero_chars.
18452 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18453 (new_strinfo): Update after above changes to strinfo.
18454 (set_endptr_and_length): Set full_string_p.
18455 (get_string_length): Update after above changes to strinfo.
18456 (unshare_strinfo): Update call to new_strinfo.
18457 (maybe_invalidate): Likewise.
18458 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18459 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18460 as a uhwi instead of an shwi. Update after above changes to
18461 strinfo and new_strinfo.
18462 (zero_length_string): Assert that chainsi contains full strings.
18463 Use zero_length_string_p. Update call to new_strinfo.
18464 (adjust_related_strinfos): Update after above changes to strinfo.
18465 Copy full_string_p from origsi.
18466 (adjust_last_stmt): Use zero_length_string_p.
18467 (handle_builtin_strlen): Update after above changes to strinfo and
18468 new_strinfo. Install the lhs as the string length if the previous
18469 entry didn't describe a full string.
18470 (handle_builtin_strchr): Update after above changes to strinfo
18472 (handle_builtin_strcpy): Likewise.
18473 (handle_builtin_strcat): Likewise.
18474 (handle_builtin_malloc): Likewise.
18475 (handle_pointer_plus): Likewise.
18476 (handle_builtin_memcpy): Likewise. Track nonzero characters
18477 that aren't necessarily followed by a nul terminator.
18478 (handle_char_store): Likewise.
18480 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18482 PR tree-optimization/80769
18483 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18484 for malloc and calloc. Document the new invariant that all related
18485 strinfos have delayed lengths or none do.
18486 (verify_related_strinfos): Move earlier in file.
18487 (set_endptr_and_length): New function, split out from...
18488 (get_string_length): ...here. Also set the lengths of related
18490 (zero_length_string): Assert that chainsi has known (rather than
18492 (adjust_related_strinfos): Likewise.
18494 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18496 PR tree-optimization/81136
18497 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18498 assert that two references with the same misalignment have the same
18499 compile-time misalignment if those compile-time misalignments
18502 2017-07-01 Andi Kleen <ak@linux.intel.com>
18504 * print-tree.c (print_node): Print all attributes.
18506 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18508 * cfg.c (scale_bbs_frequencies): New function.
18509 * cfg.h (scale_bbs_frequencies): Declare it.
18510 * cfgloopanal.c (single_likely_exit): Cleanup.
18511 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18513 (scale_loop_profile): Likewise.
18514 (loop_version): Likewise.
18515 (create_empty_loop_on_edge): Update.
18516 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18517 scale_loop_frequencies, scale_loop_profile, loopify,
18518 loop_version): Update prototypes.
18519 * modulo-sched.c (sms_schedule): Update.
18520 * predict.c (unlikely_executed_edge_p): Also check probability.
18521 (probably_never_executed_edge_p): Fix typo.
18522 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18523 * tree-parloops.c (gen_parallel_loop): Update.
18524 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18525 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18526 * tree-ssa-loop-split.c (split_loop): Update.
18527 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18528 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18529 (vect_loop_versioning): Update.
18530 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18532 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18534 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18536 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18538 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18539 to keep profile consistent.
18541 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18543 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18544 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18545 * profile-count.h (max_safe_multiplier): Make unsigned.
18546 (profile_count::guessed_zero): New.
18548 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18550 * bb-reorder.c (fix_up_crossing_landing_pad,
18551 fix_crossing_conditional_branches): Use make_single_succ_edge
18552 to keep profile consistent.
18554 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18556 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18559 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18562 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18563 the right scopes, make sure cond_jump isn't preserved between multiple
18564 iterations. Search for fallthru edge whenever there are 3+ edges and
18565 use find_fallthru_edge for it.
18567 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18569 Patch by Alexander Monakov <amonakov@ispras.ru>
18570 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18571 probabilities consistently.
18573 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18575 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18576 emit_cmp_and_jump_insns.
18578 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18581 * tree-inline.c (expand_call_inline): Combine profile statuses.
18583 2017-06-30 Andrew Pinski <apinski@cavium.com>
18585 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18586 fold_stmt returned true.
18588 2017-06-30 Nathan Sidwell <nathan@acm.org>
18590 * ggc.h (empty_string): Delete.
18591 * cfgexpand.c (expand_asm_stmt): Use plain "".
18592 * optabs.c (expand_asm_memory_barrier): Likewise.
18593 * stringpool.c (empty_string): Delete.
18594 (digit_vector, digit_string): Delete.
18595 (ggc_alloc_string): Use plain "", don't optimize single digit
18596 strings. Use ggc_alloc_atomic.
18598 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18600 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18601 comparison set and one other set, use the cost of the non-comparison
18604 2017-06-30 Nathan Sidwell <nathan@acm.org>
18606 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18609 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18611 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18612 loops. Remove now unneeded calls to gimple_switch_set_label() that
18613 just set removed labels to NULL_TREE.
18615 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18617 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18618 (set_range_info): ...here. Only call set_range_info_raw if domain
18620 (set_nonzero_bits): Call set_range_info_raw.
18621 * tree-ssanames.h (set_range_info_raw): New.
18623 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18626 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18627 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18628 of nonimmediate_operand and <store_mask_constraint> instead of m
18629 for the input operand. For V8FI iterator, always split if input
18630 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18631 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18632 <store_mask_predicate> instead of register_operand and
18633 <store_mask_constraint> instead of v for the input operand. Make
18634 sure both operands aren't MEMs for if not <mask_applied>.
18636 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18638 * lto-wrapper.c (copy_file) Close both file descriptors before
18641 2017-06-30 Martin Liska <mliska@suse.cz>
18644 * multiple_target.c (create_dispatcher_calls): Make ifunc
18645 also for function that don't have calls or are not referenced.
18647 2017-06-30 Richard Biener <rguenther@suse.de>
18649 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18650 analyze the first scalar stmt. Move vector type computation
18651 for the BB case here from ...
18652 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18653 live operation processing in the SLP case properly.
18655 2017-06-30 Richard Biener <rguenther@suse.de>
18657 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18659 2017-06-30 Martin Liska <mliska@suse.cz>
18662 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18663 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18665 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18667 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18670 2017-06-30 Richard Biener <rguenther@suse.de>
18672 PR tree-optimization/81249
18673 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18674 condition reduction result to original scalar type.
18676 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18678 * profile-count.h (enum profile_quality): Fix typos and whitespace
18681 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18683 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18684 type for branch probabilities.
18686 2017-06-29 Julian Brown <julian@codesourcery.com>
18687 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18689 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18690 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18691 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18692 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18694 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18696 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18697 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18698 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18699 CC usage from generic code to here.
18700 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18701 CC usage into the target macros.
18703 2017-06-29 Maya Rashish <coypu@sdf.org>
18705 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18708 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18710 * arm/arm-builtins.c: Include profile-count.h
18711 * except.c (sjlj_emit_function_enter): Use
18712 profile_probability::unlikely.
18714 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18716 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18717 and tocrel_offset be pointer args rather than implicitly using
18719 (legitimate_constant_pool_address_p, rs6000_emit_move,
18720 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18721 tocrel_offset and use in toc_relative_expr_p call.
18722 (print_operand, print_operand_address): Use static tocrel_base_oac
18723 and tocrel_offset_oac.
18724 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18727 2017-06-29 Maya Rashish <coypu@sdf.org>
18729 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18731 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18733 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18734 objects, take into account only the alignment of 'op0' and 'mode1' if
18737 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18739 * ccmp.c (ccmp_tree_comparison_p): New function.
18740 (ccmp_candidate_p): Update to use above function.
18741 (get_compare_parts): New function.
18742 (expand_ccmp_next): Update to use new functions.
18743 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18745 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18746 take mode as argument.
18747 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18748 * expr.c (expand_expr_real_1): Pass mode as argument.
18750 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18752 * combine.c (combine_instructions): Print insns to dump_file, together
18755 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18757 * asan.c (asan_emit_stack_protection): Update.
18758 (create_cond_insert_point): Update.
18759 * auto-profile.c (afdo_propagate_circuit): Update.
18760 * basic-block.h (struct edge_def): Turn probability to
18761 profile_probability.
18762 (EDGE_FREQUENCY): Update.
18763 * bb-reorder.c (find_traces_1_round): Update.
18764 (better_edge_p): Update.
18765 (sanitize_hot_paths): Update.
18766 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18767 (make_single_succ_edge): Update.
18768 (check_bb_profile): Update.
18769 (dump_edge_info): Update.
18770 (update_bb_profile_for_threading): Update.
18771 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18772 probabilitycount to 0.
18773 * cfgbuild.c (compute_outgoing_frequencies): Update.
18774 * cfgcleanup.c (try_forward_edges): Update.
18775 (outgoing_edges_match): Update.
18776 (try_crossjump_to_edge): Update.
18777 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18778 (expand_gimple_tailcall): Update.
18779 (construct_init_block): Use make_single_succ_edge.
18780 (construct_exit_block): Use make_single_succ_edge.
18781 * cfghooks.c (verify_flow_info): Update.
18782 (redirect_edge_succ_nodup): Update.
18783 (split_edge): Update.
18784 (account_profile_record): Update.
18785 * cfgloopanal.c (single_likely_exit): Update.
18786 * cfgloopmanip.c (scale_loop_profile): Update.
18787 (set_zero_probability): Remove.
18788 (duplicate_loop_to_header_edge): Update.
18789 * cfgloopmanip.h (loop_version): Update prototype.
18790 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18791 (force_nonfallthru_and_redirect): Update.
18792 (update_br_prob_note): Update.
18793 (rtl_verify_edges): Update.
18794 (purge_dead_edges): Update.
18795 (rtl_lv_add_condition_to_bb): Update.
18796 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18797 * cgraphunit.c (init_lowered_empty_function): Update.
18798 (cgraph_node::expand_thunk): Update.
18799 * cilk-common.c: Include profile-count.h
18800 * dojump.c (inv): Remove.
18801 (jumpifnot): Update.
18802 (jumpifnot_1): Update.
18803 (do_jump_1): Update.
18805 (do_jump_by_parts_greater_rtx): Update.
18806 (do_compare_rtx_and_jump): Update.
18807 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18808 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18809 * dwarf2cfi.c: Include profile-count.h
18810 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18811 (sjlj_emit_dispatch_table): Likewise.
18812 * explow.c: Include profile-count.h
18813 * expmed.c (emit_store_flag_force): Update.
18814 (do_cmp_and_jump): Update.
18815 * expr.c (compare_by_pieces_d::generate): Update.
18816 (compare_by_pieces_d::finish_mode): Update.
18817 (emit_block_move_via_loop): Update.
18818 (store_expr_with_bounds): Update.
18819 (store_constructor): Update.
18820 (expand_expr_real_2): Update.
18821 (expand_expr_real_1): Update.
18822 * expr.h (try_casesi, try_tablejump): Update prototypes.
18823 * gimple-pretty-print.c (dump_probability): Update.
18824 (dump_profile): New.
18825 (dump_gimple_label): Update.
18826 (dump_gimple_bb_header): Update.
18827 * graph.c (draw_cfg_node_succ_edges): Update.
18828 * hsa-gen.c (convert_switch_statements): Update.
18829 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18830 (find_if_case_1): Update.
18831 (find_if_case_2): Update.
18832 * internal-fn.c (expand_arith_overflow_result_store): Update.
18833 (expand_addsub_overflow): Update.
18834 (expand_neg_overflow): Update.
18835 (expand_mul_overflow): Update.
18836 (expand_vector_ubsan_overflow): Update.
18837 * ipa-cp.c (good_cloning_opportunity_p): Update.
18838 * ipa-split.c (split_function): Use make_single_succ_edge.
18839 * ipa-utils.c (ipa_merge_profiles): Update.
18840 * loop-doloop.c (add_test): Update.
18841 (doloop_modify): Update.
18842 * loop-unroll.c (compare_and_jump_seq): Update.
18843 (unroll_loop_runtime_iterations): Update.
18844 * lra-constraints.c (lra_inheritance): Update.
18845 * lto-streamer-in.c (input_cfg): Update.
18846 * lto-streamer-out.c (output_cfg): Update.
18847 * mcf.c (adjust_cfg_counts): Update.
18848 * modulo-sched.c (sms_schedule): Update.
18849 * omp-expand.c (expand_omp_for_init_counts): Update.
18850 (extract_omp_for_update_vars): Update.
18851 (expand_omp_ordered_sink): Update.
18852 (expand_omp_for_ordered_loops): Update.
18853 (expand_omp_for_generic): Update.
18854 (expand_omp_for_static_nochunk): Update.
18855 (expand_omp_for_static_chunk): Update.
18856 (expand_cilk_for): Update.
18857 (expand_omp_simd): Update.
18858 (expand_omp_taskloop_for_outer): Update.
18859 (expand_omp_taskloop_for_inner): Update.
18860 * omp-simd-clone.c (simd_clone_adjust): Update.
18861 * optabs.c (expand_doubleword_shift): Update.
18862 (expand_abs): Update.
18863 (emit_cmp_and_jump_insn_1): Update.
18864 (expand_compare_and_swap_loop): Update.
18865 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18866 * predict.c (predictable_edge_p): Update.
18867 (edge_probability_reliable_p): Update.
18868 (set_even_probabilities): Update.
18869 (combine_predictions_for_insn): Update.
18870 (combine_predictions_for_bb): Update.
18871 (propagate_freq): Update.
18872 (estimate_bb_frequencies): Update.
18873 (force_edge_cold): Update.
18874 * profile-count.c (profile_count::dump): Add missing space into dump.
18875 (profile_count::debug): Add newline.
18876 (profile_count::differs_from_p): Explicitly convert to unsigned.
18877 (profile_count::stream_in): Update.
18878 (profile_probability::dump): New member function.
18879 (profile_probability::debug): New member function.
18880 (profile_probability::differs_from_p): New member function.
18881 (profile_probability::differs_lot_from_p): New member function.
18882 (profile_probability::stream_in): New member function.
18883 (profile_probability::stream_out): New member function.
18884 * profile-count.h (profile_count_quality): Rename to ...
18885 (profile_quality): ... this one.
18886 (profile_probability): New.
18887 (profile_count): Update.
18888 * profile.c (compute_branch_probabilities): Update.
18889 * recog.c (peep2_attempt): Update.
18890 * sched-ebb.c (schedule_ebbs): Update.
18891 * sched-rgn.c (find_single_block_region): Update.
18892 (compute_dom_prob_ps): Update.
18893 (schedule_region): Update.
18894 * sel-sched-ir.c (compute_succs_info): Update.
18895 * stmt.c (struct case_node): Update.
18896 (do_jump_if_equal): Update.
18897 (get_outgoing_edge_probs): Update.
18898 (conditional_probability): Update.
18899 (emit_case_dispatch_table): Update.
18900 (expand_case): Update.
18901 (expand_sjlj_dispatch_table): Update.
18902 (emit_case_nodes): Update.
18903 * targhooks.c: Update.
18904 * tracer.c (better_p): Update.
18905 (find_best_successor): Update.
18906 * trans-mem.c (expand_transaction): Update.
18907 * tree-call-cdce.c: Update.
18908 * tree-cfg.c (gimple_split_edge): Upate.
18909 (move_sese_region_to_fn): Upate.
18910 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18911 * tree-eh.c (lower_resx): Upate.
18912 (cleanup_empty_eh_move_lp): Upate.
18913 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18914 * tree-inline.c (copy_edges_for_bb): Update.
18915 (copy_cfg_body): Update.
18916 * tree-parloops.c (gen_parallel_loop): Update.
18917 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18918 (gimple_gen_time_profiler): Update.
18919 * tree-ssa-dce.c (remove_dead_stmt): Update.
18920 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18921 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18922 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18923 (unloop_loops): Update.
18924 (try_peel_loop): Update.
18925 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18926 * tree-ssa-loop-split.c (connect_loops): Update.
18927 (split_loop): Update.
18928 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18929 (hoist_guard): Update.
18930 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18931 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18932 (value_replacement): Update.
18933 * tree-ssa-reassoc.c (branch_fixup): Update.
18934 * tree-ssa-tail-merge.c (replace_block_by): Update.
18935 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18936 (create_edge_and_update_destination_phis): Update.
18937 (compute_path_counts): Update.
18938 (recompute_probabilities): Update.
18939 (update_joiner_offpath_counts): Update.
18940 (freqs_to_counts_path): Update.
18941 (duplicate_thread_path): Update.
18942 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18943 (struct switch_conv_info): Update.
18944 (gen_inbound_check): Update.
18945 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18946 (vect_do_peeling): Update.
18947 (vect_loop_versioning): Update.
18948 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18949 (optimize_mask_stores): Update.
18950 * ubsan.c (ubsan_expand_null_ifn): Update.
18951 * value-prof.c (gimple_divmod_fixed_value): Update.
18952 (gimple_divmod_fixed_value_transform): Update.
18953 (gimple_mod_pow2): Update.
18954 (gimple_mod_pow2_value_transform): Update.
18955 (gimple_mod_subtract): Update.
18956 (gimple_mod_subtract_transform): Update.
18957 (gimple_ic): Update.
18958 (gimple_stringop_fixed_value): Update.
18959 (gimple_stringops_transform): Update.
18960 * value-prof.h: Update.
18962 2017-06-29 Carl Love <cel@us.ibm.com>
18964 * config/rs6000/rs6000-c.c: Add support for built-in functions
18965 vector signed int vec_signed (vector float);
18966 vector signed long long vec_signed (vector double);
18967 vector signed int vec_signed2 (vector double, vector double);
18968 vector signed int vec_signede (vector double);
18969 vector signed int vec_signedo (vector double);
18970 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18971 instruction generator.
18972 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18973 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18974 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18976 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18977 vunsignede_v2df): Add define_expands.
18978 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18979 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18980 VEC_UNSIGNEDO): Add definitions.
18981 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18982 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18983 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18984 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18985 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18986 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18987 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18988 vec_unsignede, vec_unsignedo): Add builtin defines.
18989 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18991 * doc/extend.texi: Update the built-in documentation file for the
18992 new built-in functions.
18994 2017-06-29 Richard Biener <rguenther@suse.de>
18996 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18997 reduction chains to LOOP_VINFO_REDUCTIONS.
18998 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18999 SLP reductions after processing reduction chains.
19001 2017-06-29 Nathan Sidwell <nathan@acm.org>
19003 * builtins.c (fold_builtin_FUNCTION): Use
19004 lang_hooks.decl_printable_name.
19006 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19008 PR middle-end/81194
19009 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19010 with only one label.
19011 * stmt.c (expand_case): Assert NCASES is greater than one.
19013 2017-06-29 Richard Biener <rguenther@suse.de>
19015 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19017 (group_case_labels): Likewise.
19018 (find_taken_edge): Push sanity checking on val to workers...
19019 (find_taken_edge_cond_expr): ... here
19020 (find_taken_edge_switch_expr): ... and here, handle cases
19021 with just a default label.
19022 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19023 (group_case_labels): Likewise.
19024 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19025 group_case_labels does anything cleanup the CFG again.
19027 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19029 PR tree-optimization/81196
19030 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19031 exit condition comparing two IVs.
19033 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19035 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19036 profile to the dummy entry at the end of the list of architectures.
19037 * config/arm/arm-cpu-cdata.h: Regenerated.
19039 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19040 Michael Collison <michael.collison@arm.com>
19043 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19045 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19046 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19047 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19048 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19049 of shift when the shift amount is masked with constant equal to
19050 the size of the mode.
19051 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19054 2017-06-29 Martin Liska <mliska@suse.cz>
19056 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19058 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19061 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19064 (edge_growth_cache_entry::edge_growth_cache_entry): New
19066 (reset_edge_growth_cache): Update to use constructor.
19068 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19070 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19071 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19072 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19074 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19076 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19077 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19079 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19081 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19082 (*-linux-uclibc*): Add t-uclibc tmake_file.
19083 * config/t-musl: New.
19084 * config/t-uclibc: New.
19086 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19088 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19090 (gen_comm_data): Emit architectural setting of arch_prof.
19091 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19093 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19094 (armv8-m.base, armv8-m.main): Likewise.
19095 * arm-protos.h (arm_build_target): Add profile field.
19096 (arch_option): Likewise.
19097 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19099 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19100 arm_active_target.profile.
19102 2017-06-28 Richard Biener <rguenther@suse.de>
19104 PR middle-end/81227
19105 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19106 TYPE_OVERFLOW_WRAPS.
19107 * match.pd (negate_expr_p): Likewise.
19108 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19109 fold_build2, not fold_binary.
19111 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19113 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19114 Convert memory address to Pmode.
19115 (aarch64_print_operand): Assert MEM operands are always Pmode.
19117 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19120 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19121 Remove redundant if.
19122 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19123 * config/arm/aarch-common-protos.h
19124 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19125 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19127 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19130 * multiple_target.c (create_dispatcher_calls): Set the default
19131 clone to be static, not public.
19133 2017-06-28 Richard Biener <rguenther@suse.de>
19135 * tree-vect-loop.c (vectorizable_reduction): Move special
19136 cond reduction IV var creation ...
19137 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19138 parameter. Use STMT_VINFO_VECTYPE.
19139 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19142 2017-06-28 Martin Liska <mliska@suse.cz>
19145 * ipa-visibility.c (non_local_p): Handle visibility.
19147 2017-06-28 Martin Liska <mliska@suse.cz>
19150 * common.opt: Add IntegerRange to various options.
19151 * opt-functions.awk (integer_range_info): New function.
19152 * optc-gen.awk: Add integer_range_info to cl_options struct.
19153 * opts-common.c (decode_cmdline_option): Handle
19154 CL_ERR_INT_RANGE_ARG.
19155 (cmdline_handle_error): Likewise.
19156 * opts.c (print_filtered_help): Show valid interval in
19157 when --help is provided.
19158 * opts.h (struct cl_option): Add range_min and range_max fields.
19159 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19161 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19163 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19164 (x * C EQ/NE y * C): New transformation.
19166 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19168 * genmultilib (combination_space): Accept '+' in option names.
19170 2017-06-28 Martin Liska <mliska@suse.cz>
19173 * asan.c (instrument_derefs): Bail out inner references
19174 that are hard register variables.
19176 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19179 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19180 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19182 2017-06-28 Richard Biener <rguenther@suse.de>
19184 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19185 (vect_get_slp_defs): Adjust.
19186 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19188 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19190 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19191 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19192 (vectorizable_reduction): Adjust.
19193 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19195 (vect_get_slp_defs): Likewise.
19196 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19197 (vectorizable_bswap): Adjust.
19198 (vectorizable_call): Likewise.
19199 (vectorizable_conversion): Likewise.
19200 (vectorizable_assignment): Likewise.
19201 (vectorizable_shift): Likewise.
19202 (vectorizable_operation): Likewise.
19203 (vectorizable_store): Likewise.
19204 (vectorizable_condition): Likewise.
19205 (vectorizable_comparison): Likewise.
19207 2017-06-28 Michael Collison <michael.collison@arm.com>
19210 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19212 (arm_gen_movmemqi): Removed unused variable 'i'.
19213 Convert 'for' loop into 'while' loop.
19214 (arm_expand_prologue): Remove last unnecessary set of insn.
19215 (thumb_pop): Remove unused variable 'pushed_words'.
19216 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19218 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19220 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19221 * config/s390/s390-protos.h: Add prototype of
19222 s390_rel_address_ok_p.
19223 * config/s390/s390.c (s390_got_symbol): New function.
19224 (s390_rel_address_ok_p): New function.
19225 (legitimize_pic_address): Use s390_rel_address_ok_p.
19226 (s390_load_got): Use s390_got_symbol.
19227 (s390_option_override): Issue error if
19228 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19229 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19231 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19233 2017-06-27 Andrew Pinski <apinski@cavium.com>
19235 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19236 (X * copysign (1.0, X)): New pattern.
19237 (X * copysign (1.0, -X)): New pattern.
19238 (copysign (-1.0, CST)): New pattern.
19240 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19242 * genmultilib (combination_space): Remove variable.
19243 Validate reuse rules against regular expression for any sequence
19244 of multilib options in any order.
19246 2017-06-27 Michael Collison <michael.collison@arm.com>
19248 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19249 call aarch64_split_simd_combine.
19250 * (aarch64_combine_internal<mode>): Delete pattern.
19251 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19252 Allow register and subreg operands.
19254 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19256 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19257 specific need, just fallback on defaults.
19258 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19260 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19261 Olivier Hainque <hainque@adacore.com>
19263 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19265 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19266 targets. Pick a default if no particular attempt applied.
19267 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19270 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19272 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19273 (x86_64-wrs-vxworks7): Likewise.
19275 2017-06-27 Marek Polacek <polacek@redhat.com>
19278 * ubsan.c (instrument_null): Check get_base_address's result for null.
19280 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19282 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19284 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19286 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19287 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19288 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19289 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19290 New function types.
19291 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19292 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19293 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19294 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19295 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19296 BUILT_IN_FEUPDATEENV): New builtins.
19297 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19298 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19299 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19300 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19302 (builtin_structptr_types): Adjust size.
19303 * tree.c (builtin_structptr_types): Add four entries.
19305 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19306 Olivier Hainque <hainque@adacore.com>
19308 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19309 (TLS_SYM): New local macro, forcing reference to __tls__ on
19310 link command lines for VxWorks 7 RTPs, triggering initialization
19312 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19313 OS features TLS support, true for RTPs on VxWorks 7.
19314 * config/vxworks.c (vxworks_override_options): Setup emutls
19317 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19319 * predict.c (test_prediction_value_range): Use -1U instead of -1
19320 to avoid narrowing conversion warning.
19321 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19322 to avoid narrowing conversion warning.
19323 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19325 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19327 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19329 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19330 64bit configurations.
19331 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19332 (SIZE_TYPE): Likewise.
19333 * config/vxworks.c (vxworks_emutls_var_fields): Use
19334 long_unsigned_type_node instead of unsigned_type_node as the offset
19335 field type, which is "pointer" mode in emutls.c.
19337 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19340 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19342 PR middle-end/81207
19343 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19344 gimple_vuse copying separately from gimple_vdef copying.
19346 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19348 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19351 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19352 Olivier Hainque <hainque@adacore.com>
19354 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19355 for all vxworks7 targets.
19356 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19357 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19358 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19359 variations for VX6/VX7 and 32/64bits later on in ...
19360 (VXWORKS_LIB_SPEC): Leverage new macros.
19361 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19362 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19364 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19366 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19367 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19369 2017-06-26 Carl Love <cel@us.ibm.com>
19371 * config/rs6000/rs6000-c.c: Add support for built-in functions
19372 vector bool char vec_reve (vector bool char);
19373 vector signed char vec_reve (vector signed char);
19374 vector unsigned char vec_reve (vector unsigned char);
19375 vector bool int vec_reve (vector bool int);
19376 vector signed int vec_reve (vector signed int);
19377 vector unsigned int vec_reve (vector unsigned int);
19378 vector bool long long vec_reve (vector bool long long);
19379 vector signed long long vec_reve (vector signed long long);
19380 vector unsigned long long vec_reve (vector unsigned long long);
19381 vector bool short vec_reve (vector bool short);
19382 vector signed short vec_reve (vector signed short);
19383 vector double vec_reve (vector double);
19384 vector float vec_reve (vector float);
19385 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19386 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19387 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19388 (altivec_vreve): New pattern.
19389 * config/rs6000/altivec.h (vec_reve): New define.
19390 * doc/extend.texi (vec_rev): Update the built-in documentation file
19391 for the new built-in functions.
19393 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19395 PR tree-optimization/71815
19396 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19398 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19400 (slsr_process_phi): Likewise.
19401 (replace_uncond_cands_and_profitable_phis): Don't replace a
19402 multiply candidate with a stride of 1 (copy or cast).
19403 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19405 (lowest_cost_path): Likewise.
19406 (total_savings): Likewise.
19408 2017-06-26 Richard Biener <rguenther@suse.de>
19411 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19412 Use def_builtin_pure for all gather builtins.
19414 2017-06-26 Richard Biener <rguenther@suse.de>
19416 PR tree-optimization/81203
19417 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19418 non-dominating BBs.
19420 2017-06-26 Marek Polacek <polacek@redhat.com>
19423 * doc/invoke.texi: Document -Wmultistatement-macros.
19425 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19427 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19428 arm_neon_ok_no_float_abi effective target.
19430 2017-06-26 Richard Biener <rguenther@suse.de>
19432 PR tree-optimization/80928
19433 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19434 (copy_bbs): Set BB_DUPLICATED flag early.
19435 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19437 (execute_on_shrinking_pred): Likewise.
19438 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19439 BB_DUPLICATED blocks.
19440 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19441 iterate over all PHIs considering removal of *gsi.
19443 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19445 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19448 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19450 * config/rs6000/rs6000-string.c: (expand_block_clear,
19451 do_load_for_compare, select_block_compare_mode,
19452 compute_current_alignment, expand_block_compare,
19453 expand_strncmp_align_check, expand_strn_compare,
19454 expand_block_move, rs6000_output_load_multiple)
19455 Move functions related to string/block move/compare
19456 to a separate file.
19457 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19458 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19459 for this function which is now used in two files.
19460 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19461 * config.gcc: Add rs6000-string.o to extra_objs for
19462 targets powerpc*-*-* and rs6000*-*-*.
19464 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19467 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19468 32-bit, since indexed is not valid for DImode.
19469 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19470 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19471 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19472 (define_peephole2 for Altivec d-form store): Likewise.
19475 * multiple_target.c (create_dispatcher_calls): Only create the
19476 dispatcher call if the function is the default clone of a
19477 versioned function.
19479 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19481 PR middle-end/80902
19482 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19483 a call, force the call to not be a tail call.
19485 2017-06-23 Jeff Law <law@redhat.com>
19487 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19490 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19492 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19493 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19494 handling for early expansion of vector shifts (sl,sr,sra,rl).
19495 (builtin_function_type): Add vector shift right instructions
19496 to the unsigned argument list.
19498 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19500 rtl-optimizatoin/79286
19501 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19502 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19503 trap. PIC register plus a const unspec without offset can never trap.
19505 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19507 * tree.h (builtin_structptr_type): New type.
19508 (builtin_structptr_types): Declare new array.
19509 * tree.c (builtin_structptr_types): New array.
19510 (free_lang_data, build_common_tree_nodes): Use it.
19512 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19515 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19518 2017-06-22 Matt Turner <mattst88@gmail.com>
19520 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19521 Lake models to skylake case. Assume skylake for unknown
19522 models with clflushopt.
19524 2017-06-22 Jeff Law <law@redhat.com>
19526 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19527 frame sizes that do not satisfy aarch64_uimm12_shift.
19529 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19531 * profile-count.h (apply_probability,
19532 apply_scale, probability_in): Fix checks for zero.
19534 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19536 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19537 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19539 2016-06-22 Richard Biener <rguenther@suse.de>
19541 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19542 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19543 REDUC_MAX_EXPR support.
19544 (vectorizable_reduction): Likewise.
19545 (vect_create_epilog_for_reduction): Likewise.
19547 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19549 * match.pd (A / (1 << B) -> A >> B): New.
19550 * generic-match-head.c: Include optabs-tree.h.
19551 * gimple-match-head.c: Likewise.
19552 * optabs-tree.h (target_supports_op_p): New.
19553 * optabs-tree.c (target_supports_op_p): New.
19555 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19557 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19558 $gcc_cv_ld --help output.
19559 (gcc_cv_ld_demangle): Likewise.
19560 (gcc_cv_ld_eh_frame_hdr): Likewise.
19561 (gcc_cv_ld_pie): Likewise.
19562 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19563 (gcc_cv_ld_buildid): Likewise.
19564 (gcc_cv_ld_sysroot): Likewise.
19565 (ld_bndplt_support): Likewise.
19566 (ld_pushpopstate_support): Likewise.
19567 * configure: Regenerate.
19568 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19570 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19573 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19574 operands indexes to avoid gap between operands and match_dups.
19576 2017-06-21 Andrew Pinski <apinski@cavium.com>
19578 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19579 Increment Arith_shift and Arith_shift_reg by 1.
19580 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19582 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19583 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19584 (aarch64_strip_extend): Add new argument and test for it.
19585 (aarch64_cheap_mult_shift_p): New function.
19586 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19587 add a cost if it is true.
19588 Update calls to aarch64_strip_extend.
19589 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19591 2017-06-21 Andrew Pinski <apinski@cavium.com>
19593 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19595 (thunderxt88): Likewise.
19596 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19597 (thunderx_prefetch_tune): New variable.
19598 (thunderx2t99_prefetch_tune): Update for the correct values.
19599 (thunderxt88_tunings): New variable.
19600 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19601 generic_prefetch_tune.
19602 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19604 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19606 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19607 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19608 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19609 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19610 (aarch64_atomic_cas<mode>, GPI): Likewise.
19612 2017-06-21 Martin Liska <mliska@suse.cz>
19614 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19615 statements on cold and hot labels.
19616 * predict.c (tree_estimate_probability_bb): Remove the
19617 prediction from this place.
19619 2017-06-21 Martin Liska <mliska@suse.cz>
19621 PR tree-optimization/79489
19622 * gimplify.c (maybe_add_early_return_predict_stmt): New
19624 (gimplify_return_expr): Call the function.
19625 * predict.c (tree_estimate_probability_bb): Remove handling
19627 * predict.def: Update comment about early return predictor.
19628 * gimple-predict.h (is_gimple_predict): New function.
19629 * predict.def: Change default value of early return to 66.
19630 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19632 * passes.def: Put pass_strip_predict_hints to the beginning of
19635 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19637 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19638 FUNCTION_DECL declarations.
19639 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19641 (dwaf2out_decl): Likewise.
19642 * godump.c (go_early_global_decl): Skip call to the real debug hook
19643 for FUNCTION_DECL declarations.
19644 * passes.c (rest_of_decl_compilation): Skip call to the
19645 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19646 -fdump-go-spec is passed.
19648 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19650 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19651 Reorder for compactness.
19652 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19653 (def_builtin_pure, def_builtin_pure2): New functions.
19654 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19656 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19658 * match.pd (nop_convert): New predicate.
19659 ((A +- CST1) +- CST2): Allow some NOP conversions.
19661 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19664 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19665 with ctors/dtors if GOVD_SHARED is set.
19667 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19669 * config/aarch64/aarch64.md (movti_aarch64):
19670 Emit mov rather than orr.
19671 (movtf_aarch64): Likewise.
19672 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19673 Emit mov rather than orr.
19675 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19677 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19678 Swap alternatives, make integer dup more expensive.
19680 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19682 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19683 Return true for non-tls symbols.
19685 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19687 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19688 (cortex-a75): Likewise.
19689 (cortex-a75.cortex-a55): Likewise.
19690 * config/aarch64/aarch64-tune.md: Regenerate.
19691 * doc/invoke.texi (-mtune): Document new values for -mtune.
19693 2017-06-21 Tom de Vries <tom@codesourcery.com>
19695 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19696 stack_size feature.
19697 (Effective-Target Keywords, Other attributes): Suggest using
19698 dg-add-options stack_size feature to get stack limit in stack_size
19699 effective target documentation.
19701 2017-06-21 Julian Brown <julian@codesourcery.com>
19702 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19704 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19705 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19706 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19708 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19709 attribute type list for neon_multiply.
19710 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19711 attribute type list for neon_multiply.
19712 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19713 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19714 attribute type list for neon_multiply.
19715 * config/arm/types.md (crypto_pmull): Add.
19716 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19717 attribute type list.
19719 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19721 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19724 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19726 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19727 to make sure not to dereference a NULL cost_classes_ptr pointer.
19729 2017-06-20 Carl Love <cel@us.ibm.com>
19731 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19732 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19733 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19734 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19735 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19736 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19737 VMULOSW): New enum "unspec" values.
19738 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19739 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19740 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19741 altivec_vmulosw): New patterns.
19742 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19743 VMULOSW): Add definitions.
19745 2017-06-20 Julia Koval <julia.koval@intel.com>
19747 * config/i386/i386.c: Fix rounding expand for new pattern.
19748 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19750 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19752 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19753 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19755 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19757 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19760 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19762 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19763 architecture, then by implementer ID.
19764 * config/aarch64/aarch64-tune.md: Regenerate.
19766 2017-06-20 Richard Biener <rguenther@suse.de>
19768 PR middle-end/81097
19769 * fold-const.c (split_tree): Fold to type before negating.
19771 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19773 * diagnostic-show-locus.c
19774 (selftest::test_fixit_deletion_affecting_newline): New function.
19775 (selftest::diagnostic_show_locus_c_tests): Call it.
19777 2017-06-20 Andreas Schwab <schwab@suse.de>
19780 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19783 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19785 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19786 __ARM_FEATURE_COPROC according to support.
19788 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19790 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19791 Rewritten to avoid overflow for > 32-bit pointers.
19794 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19795 by removing enum keyword.
19796 (ubsan_type_descriptor): Likewise. Formatting fix.
19799 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19800 splitter): Require TARGET_SSE2 in the condition.
19802 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19805 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19806 for doing vector set of SFmode on ISA 3.0.
19807 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19808 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19810 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19811 SFmode value into a V4SF variable that was extracted from another
19812 V4SF variable without converting the element to double precision
19813 and back to single precision vector format.
19814 (vsx_insert_extract_v4sf_p9_2): Likewise.
19816 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19818 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19819 in UWHI to avoid undefined overflow.
19822 * ubsan.h (enum ubsan_encode_value_phase): New.
19823 (ubsan_encode_value): Change second argument to
19824 enum ubsan_encode_value_phase with default value of
19825 UBSAN_ENCODE_VALUE_GENERIC.
19826 * ubsan.c (ubsan_encode_value): Change second argument to
19827 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19828 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19829 create_tmp_var_raw instead of create_tmp_var and use a
19831 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19832 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19833 ubsan_encode_value callers.
19836 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19837 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19838 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19840 2017-06-19 Richard Biener <rguenther@suse.de>
19842 PR middle-end/81118
19843 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19844 estimates if we changed anything.
19846 2017-06-19 Richard Biener <rguenther@suse.de>
19848 PR tree-optimization/80887
19849 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19850 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19851 simplified lookups, then reset mprts_hook.
19852 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19854 (try_to_simplify): Likewise.
19856 2017-06-19 Martin Liska <mliska@suse.cz>
19859 * gimplify.c (gimplify_switch_expr):
19860 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19862 2017-06-19 Martin Liska <mliska@suse.cz>
19864 * doc/install.texi: Document that PGO runs in 4 stages.
19866 2017-06-19 Martin Liska <mliska@suse.cz>
19869 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19870 to dispatcher function name.
19871 * multiple_target.c (replace_function_decl): New function.
19872 (create_dispatcher_calls): Redirect both edges and references.
19874 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19876 * profile-count.c (profile_count::dump): Dump quality.
19877 (profile_count::differs_from_p): Update for unsigned val.
19878 * profile-count.h (profile_count_quality): New enum.
19879 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19881 2017-06-19 Richard Biener <rguenther@suse.de>
19883 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19884 struct function as arg.
19885 (estimate_numbers_of_iterations): Export overload with loop arg.
19886 (free_numbers_of_iterations_estimates_loop): Use an overload of
19887 free_numbers_of_iterations_estimates instead.
19888 * tree-cfg.c (remove_bb): Adjust.
19889 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19890 * tree-parloops.c (gen_parallel_loop): Likewise.
19891 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19893 (tree_unroll_loops_completely): Likewise.
19894 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19895 Use an overload instead and export.
19896 (estimated_loop_iterations): Adjust.
19897 (max_loop_iterations): Likewise.
19898 (likely_max_loop_iterations): Likewise.
19899 (estimate_numbers_of_iterations): Take struct function as arg
19901 (loop_exits_before_overflow): Adjust.
19902 (free_numbers_of_iterations_estimates_loop): Use an overload.
19903 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19904 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19906 2017-06-19 Richard Biener <rguenther@suse.de>
19909 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19910 RANGE_EXPR conservatively.
19912 2017-06-16 Carl Love <cel@us.ibm.com>
19914 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19915 definitions for vec_float, vec_float2, vec_floato,
19916 vec_floate built-ins.
19917 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19918 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19920 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19921 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19922 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19923 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19924 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19925 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19926 vec_floato): Add builtin defines.
19927 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19928 Update the built-in documentation file for the new built-in
19931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19933 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19934 (mthumb): Mark as the negative of -marm.
19936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19938 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19940 (ARM Options, -mtune): Document that this accepts the same
19941 extension options as -mcpu.
19942 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19946 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19947 permitted extensions.
19949 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19951 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19952 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19953 (armv8-m.main): Add option +nodsp.
19954 * config/arm/arm-cpu-cdata.h: Regenerated.
19956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19958 * config/arm/t-fuchsia: New file.
19959 * config.gcc (arm*-*-fuchsia*): Use it.
19961 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19963 * config/arm/t-symbian: Rewrite for new option infrastructure.
19965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19967 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19968 (MULTILIB_REQUIRED): Likewise.
19970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19972 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19973 (MULTILIB_RESUE): Likewise.
19974 (MULTILIB_MATCHES): Likewise.
19975 (MULTLIB_REQUIRED): Likewise.
19977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19979 * config/arm/t-rtems: Rewrite for new option framework.
19981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19983 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19984 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19985 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19986 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19987 * config/arm/t-multilib: ... here.
19988 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19989 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19990 armv7-a and armv8*-a when A-profile libraries have not been built.
19991 * config/arm/t-rmprofile: Rewrite.
19993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19995 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19996 with a backslash. Remove the backslash after substituting unescaped
19998 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20002 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20003 the driver to pass through the default setting of -mfloat-abi.
20004 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20006 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20007 (all_feat_combs): New rule.
20008 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20010 * config/arm/t-aprofile: Rewrite.
20012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20014 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20015 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20016 fpu is not specified by the user/command-line.
20017 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20018 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20019 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20020 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20021 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20022 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20026 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20027 * config/arm/t-arm-elf: Rewritten.
20029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20031 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20032 have some floating-point instructions.
20033 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20034 (TARGET_MAYBE_HARD_FLOAT): New macro.
20035 * config/arm/arm-builtins.c (arm_init_builtins): Use
20036 TARGET_MAYBE_HARD_FLOAT.
20037 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20039 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20041 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20042 (configargs.h): Include it.
20043 (arm_print_hint_for_fpu_option): New function.
20044 (arm_parse_fpu_option): New function.
20045 (candidate_extension): New class.
20046 (arm_canon_for_multilib): New function.
20047 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20048 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20049 (ARCH_CANONICAL_SPECS): New macro.
20050 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20054 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20055 are set after handling multilib fragments. Set target_cpu_default2
20058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20060 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20062 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20063 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20065 (check_arch): Likewise.
20066 * config/arm/arm.c (arm_configure_build_target): Handle
20067 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20068 options in the default.
20070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20072 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20073 when an option is an alias of another.
20074 * config/arm/parsecpu.awk (optalias): New parser token.
20075 (gen_comm_data): Mark non-alias options as such. Emit entries
20076 for extension aliases.
20077 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20078 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20079 (armv6kz, armv6zk, armv6t2): Likewise.
20080 (armv7): Make vfpv3-d16 an alias.
20081 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20083 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20084 Sort in canonical order.
20085 (armv8-a): Sort in canonical order.
20086 (armv8.1-a, armv8.2-a): Likewise.
20087 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20089 (cortex-a9): Sort in canonical order.
20090 * config/arm/arm.c (selftests.h): Include it.
20091 (arm_test_cpu_arch_data): New function.
20092 (arm_run_self_tests): New function.
20093 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20094 (targetm): Move declaration to the end of the file.
20095 * arm-cpu-cdata.h: Regenerated.
20097 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20099 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20100 call to target_mode_check describing the type of option passed.
20101 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20102 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20103 arm_parse_cpu_option_name to match parameters against list of
20105 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20106 arm_arch_core_flags data structure.
20107 * config/arm/arm-cpu_cdata.h: Regenerated.
20109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20111 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20113 (arm_print_hint_for_cpu_option): Likewise.
20114 (arm_print_hint_for_arch_option): Likewise.
20115 (arm_parse_cpu_option_name): Likewise.
20116 (arm_parse_arch_option_name): Likewise.
20117 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20118 of entries in the all_fpus list.
20119 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20120 (arm_parse_cpu_option_name): Declare.
20121 (arm_parse_arch_option_name): Declare.
20122 (arm_parse_option_features): Declare.
20123 (arm_intialize_isa): Declare.
20124 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20126 (gen_comm_data): ... here. Make definitions non-static.
20127 * config/arm/arm-cpu-data.h: Regenerated.
20128 * config/arm/arm-cpu-cdata.h: Regenerated.
20130 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20132 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20133 (cpu_arch_extension): New structure.
20134 (cpu_arch_option, arch_option, cpu_option): New structures.
20135 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20136 architecture types.
20137 (gen_data): Generate new format data tables.
20138 * config/arm/arm.c (cpu_tune): New structure.
20139 (cpu_option, processors): Delete.
20140 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20141 (arm_print_hint_for_cpu_option): ... this and ...
20142 (arm_print_hint_for_arch_option): ... this.
20143 (arm_parse_arch_cpu_name): Delete. Replace with ...
20144 (arm_parse_cpu_option_name): ... this and ...
20145 (arm_parse_arch_option_name): ... this.
20146 (arm_unrecognized_feature): Change type of target parameter to
20148 (arm_parse_arch_cpu_features): Delete. Replace with ...
20149 (arm_parse_option_features): ... this.
20150 (arm_configure_build_target): Rework to use new configuration data
20152 (arm_print_tune_info): Rework for new configuration data tables.
20153 * config/arm/arm-cpu-data.h: Regenerated.
20154 * config/arm/arm-cpu.h: Regenerated.
20156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20158 * Makefile.in (OBJS): Move sbitmap.o from here ...
20159 (OBJS-libcommon): ... to here.
20161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20163 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20164 (ISA_ALL_CRYPTO): New macro.
20165 (ISA_ALL_SIMD): New macro
20166 (ISA_ALL_FP): New macro.
20167 * config/arm/arm.c (fpu_bitlist): Update initializer.
20168 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20170 (arm9e): Add fpu. Add option for nofp
20171 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20172 (arm926ej-s, arm1026ej-s): Likewise.
20173 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20174 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20175 neon-fp16, neon-vfpv4, nofp and nosimd.
20176 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20177 (cortex-a8): Add fpu. Add option for nofp.
20178 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20179 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20180 (cortex-r4f): Add fpu.
20181 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20182 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20184 (cortex-r8): Likewise.
20185 (cortex-m4): Add fpu. Add option for nofp.
20186 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20187 (cortex-a17.cortex-a7): Likewise.
20188 (cortex-a32): Add fpu. Add options for crypto and nofp.
20189 (cortex-a35, cortex-a53): Likewise.
20190 (cortex-a57): Add fpu. Add option for crypto.
20191 (cortex-a72, cortex-a73): Likewise.
20192 (exynos-m1): Likewise.
20193 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20194 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20195 (cortex-m33): Add fpu. Add option for nofp.
20196 * config/arm/arm-cpu-cdata.h: Regenerated
20197 * config/arm/arm-cpu-data.h: Regenerated.
20199 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20201 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20202 (armv5te, armv5tej): Likewise.
20203 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20204 (armv7): Add options fp and vfpv3-d16.
20205 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20206 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20208 (armv7ve): Likewise.
20209 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20210 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20211 (armv8-a): Add nocrypto option.
20212 (armv8.1-a, armv8.2-a): Likewise.
20213 (armv8-m.main): add options fp, fp.dp and nofp.
20215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20217 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20219 (armv8-a+crc): Delete.
20220 (armv8.1-a): Add options simd, crypto and nofp.
20221 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20222 (armv8.2-a+fp16): Delete.
20223 (armv8-m.main): Add option dsp.
20224 (armv8-m.main+dsp): Delete.
20225 (cortex-a8): Add fpu. Add nofp option.
20226 (cortex-a9): Add fpu. Add nofp and nosimd options.
20227 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20228 link to main cpu and architecture data structures.
20229 (gen_comm_data): Only put isa attributes from the main architecture
20231 (option): New statement for architecture and CPU entries.
20232 * arm.c (struct cpu_option): New structure.
20233 (struct processors): Add entry for options.
20234 (arm_unrecognized_feature): New function.
20235 (arm_parse_arch_cpu_name): Ignore any characters after the first
20237 (arm_parse_arch_cpu_feature): New function.
20238 (arm_configure_build_target): Separate out any CPU and architecture
20239 features and parse separately. Don't error out if -mfpu=auto is
20240 used with only an architecture string.
20241 (arm_print_asm_arch_directives): New function.
20242 (arm_file_start): Call it.
20243 * config/arm/arm-cpu-cdata.h: Regenerated.
20244 * config/arm/arm-cpu-data.h: Likewise.
20245 * config/arm/arm-tables.opt: Likewise.
20247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20249 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20250 assembler when it is not -mfpu=auto.
20252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20254 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20255 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20256 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20257 (ASM_CPU_SPEC): Rewrite.
20258 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20259 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20260 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20261 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20262 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20263 copied string is NUL-terminated. Also strip any characters prefixed
20265 (arm_rewrite_selected_arch): New function.
20266 (arm_rewrite_march): New function.
20268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20270 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20271 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20272 (march, mcpu, mtune): Convert to string-based options.
20273 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20274 (arm_parse_arch_cpu_name): New function.
20275 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20276 identify selected architecture or CPU.
20277 (arm_option_save): New function.
20278 (TARGET_OPTION_SAVE): Redefine.
20279 (arm_option_restore): Restore string options.
20280 (arm_option_print): Print string options.
20282 2017-06-16 Martin Sebor <msebor@redhat.com>
20284 PR tree-optimization/80933
20285 PR tree-optimization/80934
20286 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20287 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20288 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20289 (gimple_fold_builtin): Call them.
20291 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20293 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20294 as unlikely; update profile.
20296 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20298 * predict.c (force_edge_cold): Handle declaring edges impossible
20301 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20303 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20305 (try_unroll_loop_completely): Fix reporting.
20307 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20309 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20311 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20314 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20315 if given a non-constant argument for an intrinsic which requires a
20318 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20320 * profile.c (compare_freqs): New function.
20321 (branch_prob): Sort edge list.
20322 (find_spanning_tree): Assume that the list is priority sorted.
20324 2017-06-16 Richard Biener <rguenther@suse.de>
20326 PR tree-optimization/81090
20327 * passes.def (pass_record_bounds): Remove.
20328 * tree-pass.h (make_pass_record_bounds): Likewise.
20329 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20330 make_pass_record_bounds): Likewise.
20331 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20332 not free niter estimates at the beginning but at the end.
20333 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20335 2017-06-16 Richard Biener <rguenther@suse.de>
20337 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20338 initializer to workaround ICE in host GCC 4.8.
20340 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20342 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20344 (clone_inlined_nodes): Update.
20346 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20348 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20349 prefetch settings, and enable prefetching by default at -O3.
20351 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20353 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20354 Set flag_prefetch_loop_arrays according to tuning data.
20356 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20358 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20359 New tune structure.
20360 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20361 [Unrelated to main purpose of the patch] Place the pointer field last
20362 to enable type checking errors when tune structure are wrongly merged.
20363 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20364 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20365 (thunderx2t99_prefetch_tune): New tune constants.
20366 (tune_params *_tunings): Update all tunings (no functional change).
20367 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20368 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20369 from tunings structures.
20371 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20374 * ubsan.c (instrument_null): Add T argument, use it instead
20375 of computing it based on IS_LHS.
20376 (instrument_object_size): Likewise.
20377 (pass_ubsan::execute): Adjust instrument_null and
20378 instrument_object_size callers to pass gimple_get_lhs or
20379 gimple_assign_rhs1 result to it. Use instrument_null instead of
20380 calling get_base_address and instrument_mem_ref. Handle
20381 aggregate call arguments for object-size sanitization.
20383 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20385 PR tree-optimization/81089
20386 * tree-vrp.c (is_masked_range_test): Validate operands of
20389 2017-06-15 Martin Sebor <msebor@redhat.com>
20392 * dumpfile.c (dump_register): Avoid calling memset to initialize
20393 a class with a default ctor.
20394 * gcc.c (struct compiler): Remove const qualification.
20395 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20396 * hash-table.h: Ditto.
20397 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20399 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20400 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20402 * params.h (struct param_info): Make struct members non-const.
20403 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20404 with default initialization.
20405 * vec.h (vec_copy_construct, vec_default_construct): New helper
20407 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20408 with vec_copy_construct.
20409 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20410 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20411 * doc/invoke.texi (-Wclass-memaccess): Document.
20413 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20415 * emit-rtl.h (is_leaf): Update comment about local
20416 register allocator.
20418 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20421 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20422 for a variable to have a section before checking if the section has a
20424 Set section to.persistent if persistent attribute is set.
20425 Warn if .persistent attribute is used on an automatic variable.
20427 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20429 PR rtl-optimization/80474
20430 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20432 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20434 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20437 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20439 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20440 (casesi): Emit bounds checking as RTL.
20441 (casesi_internal_mips16_<mode>): Remove bounds checking.
20443 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20445 * config/xtensa/xtensa.c (xtensa_option_override): Append
20446 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20447 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20448 xtensa_doloop_hooks): Define unconditionally.
20449 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20451 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20452 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20453 for it in xtensa_option_override.
20454 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20455 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20457 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20459 * doc/cppopts.texi: Document '-' special value to -MF.
20461 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20463 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20464 (cortex_a53_fconst): Likewise.
20465 (cortex_a53_fpmul): Likewise.
20466 (cortex_a53_f_load_64): Likewise.
20467 (cortex_a53_f_load_many): Likewise.
20468 (cortex_a53_advsimd_alu): Likewise.
20469 (cortex_a53_advsimd_alu_q): Likewise.
20470 (cortex_a53_advsimd_mul): Likewise.
20471 (cortex_a53_advsimd_mul_q): Likewise.
20472 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20473 Add missing fmul, r2f_cvt and fconst cases.
20475 2017-06-14 Richard Biener <rguenther@suse.de>
20477 PR middle-end/81088
20478 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20480 (fold_binary_loc): When associating do not treat pre-existing
20481 TREE_OVERFLOW on literal constants as a reason to allow
20482 TREE_OVERFLOW on associated literal constants.
20484 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20486 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20487 (MASK_FEATURES): New macro.
20488 * config/sparc/sparc.c (sparc_option_override): Remove the special
20489 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20491 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20493 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20494 a division of 0 if non-call exceptions are enabled.
20496 2017-06-14 Andrew Pinski <apinski@cavium.com>
20497 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20500 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20501 Improve vector initialization code gen for only variable case.
20503 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20505 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20507 2017-06-14 Richard Biener <rguenther@suse.de>
20509 PR tree-optimization/81083
20510 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20513 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20515 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20516 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20517 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20518 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20519 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20520 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20522 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20524 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20525 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20527 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20529 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20531 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20533 * config/rs6000/t-rtems: Don't handle SPE.
20535 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20537 * config/rs6000/t-linux: Don't handle SPE.
20539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20541 * config/rs6000/eabispe.h: Delete file.
20543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20545 * config/rs6000/t-spe: Delete file.
20547 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20549 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20550 (rs6000_legitimate_offset_address_p): Return false for anything in
20551 V2SImode or V2SFmode.
20553 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20555 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20556 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20557 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20559 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20560 (rs6000_legitimate_offset_address_p): Ditto.
20561 (rs6000_emit_move): Ditto.
20562 (rs6000_init_builtins): Remove V4HI_type_node.
20564 2017-06-13 Martin Liska <mliska@suse.cz>
20567 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20568 (gate_asan): Likewise.
20569 * asan.h (asan_no_sanitize_address_p): Remove the function.
20570 (sanitize_flags_p): New function.
20571 * builtins.def: Fix coding style.
20572 * common.opt: Use renamed enum value.
20573 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20574 * doc/extend.texi: Document no_sanitize attribute.
20575 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20576 to SANITIZE_UNDEFINED_NONDEFAULT.
20577 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20578 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20579 Use sanitize_flags_p.
20580 * gimplify.c (gimplify_function_tree): Likewise.
20581 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20582 * opts.c (parse_no_sanitize_attribute): New function.
20583 (common_handle_option): Use renamed enum value.
20584 * opts.h (parse_no_sanitize_attribute): Declare.
20585 * tree.c (sanitize_flags_p): New function.
20586 * tree.h: Declared here.
20587 * tsan.c: Use sanitize_flags_p.
20588 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20589 (instrument_mem_ref): Likewise.
20590 (instrument_bool_enum_load): Likewise.
20591 (do_ubsan_in_current_function): Remove the function.
20592 (pass_ubsan::execute): Use sanitize_flags_p.
20593 * ubsan.h: Remove do_ubsan_in_current_function
20594 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20595 (dump_function_to_file): Use it here.
20597 2017-06-13 Martin Jambor <mjambor@suse.cz>
20599 PR tree-optimization/80803
20600 PR tree-optimization/81063
20601 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20602 (propagate_subaccesses_across_link): Enqueue subtree whenever
20603 necessary instead of relying on the caller.
20605 2017-06-13 Martin Jambor <mjambor@suse.cz>
20607 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20608 that have a first_link.
20609 (sort_and_splice_var_accesses): Do not check first_link before
20611 (subtree_mark_written_and_enqueue): Likewise.
20612 (propagate_all_subaccesses): Likewise and do not stop at first
20613 parent with a first_link.
20615 2017-06-13 Martin Jambor <mjambor@suse.cz>
20617 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20620 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20622 * match.pd: New pattern.
20624 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20626 * tree-vrp.c (is_masked_range_test): New function.
20627 (register_edge_assert_for): Determine ranges for
20630 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20632 PR tree-optimization/67328
20633 * fold-const.c (maskable_range_p): New function.
20634 (build_range_check): Generate bittests if possible.
20636 2017-06-13 Martin Liska <mliska@suse.cz>
20638 * gimple-pretty-print.c (dump_probability): Add new argument.
20639 (dump_edge_probability): Dump both probability and count.
20640 (dump_gimple_label): Likewise.
20641 (dump_gimple_bb_header): Likewise.
20643 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20646 * config/avr/avr-devices.c: Fix indentation.
20647 * config/avr/gen-avr-mmcu-specs.c: Dito.
20649 2017-06-13 Richard Biener <rguenther@suse.de>
20651 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20652 instead get vector type from stmt_info.
20653 (vectorizable_reduction): Adjust. Remove dead code.
20655 2017-06-13 Richard Biener <rguenther@suse.de>
20657 PR middle-end/81065
20658 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20659 case of C * (x * C2 + C3).
20660 (fold_addr_of_array_ref_difference): Properly fold index difference.
20662 2017-06-12 David S. Miller <davem@davemloft.net>
20665 * config/sparc/sparc.md (return expander): Emit frame blockage if
20666 function uses alloca.
20668 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20670 * combine.c (make_field_assignment): Check len rather than the mode
20671 precision when calling force_to_mode.
20673 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20675 Support multilibs and devices that see flash in RAM address range.
20678 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20679 (avr_mcu_t) <flash_pm_offset>: New field.
20680 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20681 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20682 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20683 (AVR_TINY_PM_OFFSET): Remove macro.
20684 * config/avr/avr.opt (-mshort-calls): New option.
20685 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20686 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20687 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20688 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20689 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20690 instead of avr_arch->have_jmp_call.
20691 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20692 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20693 avr_arch->flash_pm_offset to define.
20694 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20695 new field flash_pm_offset. Add entry for avrxmega3.
20696 (avr_texinfo): Add entry for avrxmega3.
20697 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20698 attiny212, attiny214,
20699 attiny412, attiny414, attiny416, attiny417,
20700 attiny814, attiny816, attiny817,
20701 attiny1614, attiny1616, attiny1617,
20702 attiny3214, attiny3216, attiny3217.
20703 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20704 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20705 (avr_print_operand_address) [AVR_TINY]: Same.
20706 (avr_asm_init_sections) <readonly_data_section>: Only patch
20707 callback if avr_arch->flash_pm_offset = 0.
20708 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20709 for rodata if avr_arch->flash_pm_offset != 0.
20710 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20711 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20712 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20713 (m_options): Append opt_rcall.
20714 (m_dirnames): Append dir_rcall.
20715 * config/avr/t-multilib: Regenerate.
20717 * configure.ac [target=avr]: Check whether avrxmega3 default
20718 linker description file works as needed.
20719 * configure: Regenerate.
20720 * doc/avr-mmcu.texi: Regenerate.
20721 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20722 <__AVR_ARCH__>: Document avrxmega3 and 103.
20723 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20724 <__AVR_SHORT_CALLS__>: Document it.
20725 <__AVR_PM_BASE_ADDRESS__>: Document it.
20726 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20727 (AVR Variable Attributes) <progmem>: Document this is
20728 not needed for avrxmega3.
20729 (AVR Named Address Spaces) <__flash>: Dito.
20731 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20733 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20735 2017-06-12 Doug Rupp <rupp@adacore.com>
20737 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20738 Append vxworks-stdint.h to the tm_file list.
20739 * config/vxworks-stdint.h: New file.
20741 2017-06-12 Martin Liska <mliska@suse.cz>
20743 PR tree-optimization/81041
20744 * tree-profile.c (gimple_gen_ic_func_profiler):
20745 Create an extra BB in profile-generate
20746 (gimple_gen_time_profiler): Likewise.
20748 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20750 PR tree-optimization/81003
20751 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20752 (update_range_test): Use it instead of force_gimple_operand_gsi.
20754 2017-06-12 Richard Biener <rguenther@suse.de>
20756 PR tree-optimization/81053
20757 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20758 with backedge value not defined in loop. Simplify def stmt
20761 2017-06-11 Tom de Vries <tom@codesourcery.com>
20764 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20766 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20767 nvptx_cannot_force_const_mem.
20769 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20771 * opts.c (finish_options): Move test for flag_split_stack after
20772 it has been initialized.
20774 2017-06-11 Jason Merrill <jason@redhat.com>
20776 * tree.h (id_equal): New.
20777 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20778 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20779 instead of strcmp of IDENTIFIER_POINTER.
20781 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20783 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20784 (mark_all_inlined_calls_cdtor): Fix formating.
20785 (inline_transform): Rescale profile before inlining.
20787 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20789 * cgraph.h (cgraph_edge::clone): Update prototype.
20790 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20791 (cgraph_node::create_clone): Update.
20792 (cgraph_node::create_version_clone): Update.
20793 * tree-inline.c (copy_bb): Update.
20794 (expand_call_inline): Update.
20796 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20798 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20799 factored out from ...
20800 (rs6000_emit_prologue): ... here.
20802 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20804 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20805 factored out from ...
20806 (rs6000_emit_prologue): ... here.
20808 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20810 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20812 (handle_missing_profiles): Fix computation of tp_first_run.
20813 (counts_to_freqs): Do not touch freqs when count is 0.
20815 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20817 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20820 2017-06-10 Tom de Vries <tom@codesourcery.com>
20822 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20823 attributes): Document signal effective target.
20825 2017-06-10 Tom de Vries <tom@codesourcery.com>
20827 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20828 Document effective target stack_size.
20830 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20832 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20833 to the edit_context if they can be auto-applied.
20835 2017-06-9 Ian Lance Taylor <iant@golang.org>
20837 * opts.c (finish_options): If -fsplit-stack, disable implicit
20838 -forder-blocks-and-partition.
20839 * doc/invoke.texi (Optimize Options): Document that when using
20840 -fsplit-stack -forder-blocks-and-partition is not implicitly
20843 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20845 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20846 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20847 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20848 * builtins.def (abort, trap, unreachable): Declare cold.
20849 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20850 * tree-core.h (ECF_COLD): New.
20851 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20852 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20854 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20856 * predict.c (unlikely_executed_stmt_p): Cleanup.
20858 2017-06-09 Richard Biener <rguenther@suse.de>
20860 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20861 model if the ref is always written to.
20863 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20865 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20867 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20869 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20872 2017-06-09 Tom de Vries <tom@codesourcery.com>
20875 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20876 "target cannot support label values" when encountering LABEL_REF.
20878 2017-06-09 Martin Liska <mliska@suse.cz>
20880 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20881 (gimple_gen_ic_func_profiler): Emit direct comparison
20882 of __gcov_indirect_call_callee with NULL.
20883 (gimple_gen_time_profiler): Change probability from
20884 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20886 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20888 * profile.c (edge_gcov_counts): Turn to pointer.
20889 (compute_branch_probabilities, compute_branch_probabilities): Update.
20890 (branch_prob): Do not clear edge_gcov_count.
20891 * profile.h (edge_gcov_counts): Turn to pointer.
20892 (edge_gcov_count): Update.
20894 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20896 * gimple.h (gimple_check_failed): Mark cold.
20898 2017-06-09 Richard Biener <rguenther@suse.de>
20900 PR tree-optimization/66623
20901 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20902 refactor check_reduction into two parts, properly computing
20903 whether we have to check reduction validity for outer loop
20906 2017-06-09 Richard Biener <rguenther@suse.de>
20908 PR tree-optimization/79483
20909 * graphite-scop-detection.c (order): New global.
20910 (get_order): Compute bb to order mapping that satisfies code
20911 generation constraints.
20912 (cmp_pbbs): New helper.
20913 (build_scops): Start domwalk at entry block, sort generated
20916 2017-06-09 Richard Biener <rguenther@suse.de>
20918 PR middle-end/81007
20919 * ipa-polymorphic-call.c
20920 (ipa_polymorphic_call_context::restrict_to_inner_class):
20921 Skip FIELD_DECLs with error_mark_node type.
20922 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20925 2017-06-09 Martin Liska <mliska@suse.cz>
20927 * predict.c (struct branch_predictor): New struct.
20928 (test_prediction_value_range): New test.
20929 (predict_c_tests): New function.
20930 * selftest-run-tests.c (selftest::run_tests): Run the function.
20931 * selftest.h: Declare new tests.
20933 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20936 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20937 gen_add3_insn did not fail.
20938 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20939 r0, construct that number in a temporary reg and add that reg to r0.
20940 If asked to put the result in r0 as well, fail.
20942 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20945 for early expansion of vec_eqv.
20947 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20949 PR middle-end/81005
20950 * ubsan.c (instrument_null): Avoid pointless code temporary.
20951 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20953 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20957 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20960 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20963 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20965 * predict.c (unlikely_executed_edge_p): Move ahead.
20966 (probably_never_executed_edge_p): Use it.
20968 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20970 PR middle-end/79988
20971 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20972 gimple_call_builtin_p call.
20974 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20976 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20977 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20978 rtl_check_failed_type2, rtl_check_failed_code1,
20979 rtl_check_failed_code2, rtl_check_failed_code_mode,
20980 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20981 rtvec_check_failed_bounds, rtl_check_failed_flag,
20982 _fatal_insn_not_found, _fatal_insn): Likewise.
20983 * tree.h (tree_contains_struct_check_failed,
20984 tree_check_failed, tree_not_check_failed,
20985 tree_class_check_failed, tree_range_check_failed,
20986 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20987 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20988 tree_operand_check_failed, omp_clause_check_failed,
20989 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20992 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20994 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20995 flag_branch_probabilities.
20996 * ipa-inline.c (edge_badness): Likewise.
20997 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20998 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20999 * predict.c (maybe_hot_frequency_p): Likewise.
21000 (probably_never_executed): Likewise.
21001 * sched-ebb.c (schedule_ebbs): Likewise.
21002 * sched-rgn.c (find_single_block_region): Likewise.
21003 * tracer.c (tail_duplicate): Likewise.
21005 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21007 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21008 longer requires x_flag_profile_use.
21010 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21012 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21013 instead of flag_reorder_blocks_and_partition.
21014 * dbxout.c (dbxout_function_end): Likewise.
21015 * dwarf2out.c (gen_subprogram_die): Likewise.
21016 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21017 * hw-doloop.c (reorg_loops): Likewise.
21018 * varasm.c (assemble_start_function,
21019 assemble_end_function): Likewise.
21020 (decide_function_section): Do not check for
21021 flag_reorder_blocks_and_partition.
21023 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21025 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21027 (chkp_get_hard_register_fake_addr_expr): Ditto.
21028 (chkp_build_addr_expr): Add check for hard reg case.
21029 (chkp_parse_array_and_component_ref): Ditto.
21030 (chkp_find_bounds_1): Ditto.
21031 (chkp_process_stmt): Don't generate bounds store for
21034 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21036 * predict.c (maybe_hot_bb_p): Do not check profile status.
21037 (maybe_hot_edge_p): Likewise.
21038 (probably_never_executed): Check for zero counts even if profile
21040 (unlikely_executed_edge_p): New function.
21041 (unlikely_executed_stmt_p): New function.
21042 (unlikely_executed_bb_p): New function.
21043 (set_even_probabilities): Use unlikely predicates.
21044 (combine_predictions_for_bb): Likewise.
21045 (predict_paths_for_bb): Likewise.
21046 (predict_paths_leading_to_edge): Likewise.
21047 (determine_unlikely_bbs): New function.
21048 (estimate_bb_frequencies): Use it.
21049 (compute_function_frequency): Use zero counts even if profile is
21051 * profile-count.h: Fix typo.
21053 2017-08-08 Julia Koval <julia.koval@intel.com>
21055 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21056 _mm512_mask_cvtsepi16_storeu_epi8,
21057 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21058 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21059 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21060 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21061 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21062 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21063 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21064 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21065 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21066 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21067 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21068 __builtin_ia32_pmovuswb256mem_mask,
21069 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21070 __builtin_ia32_pmovwb512mem_mask): New builtins.
21072 2017-08-08 Julia Koval <julia.koval@intel.com>
21074 PR target/73350,80862
21075 * config/i386/subst.md (round): Fix round pattern.
21076 * config/i386/i386.c (ix86_erase_embedded_rounding):
21077 Fix erasing rounding for the fixed pattern.
21079 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21081 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21083 2017-06-08 Martin Liska <mliska@suse.cz>
21085 PR gcov-profile/80911
21086 * gcov.c (block_info::block_info): New constructor.
21088 2017-06-07 Carl Love <cel@us.ibm.com>
21090 * config/rs6000/rs6000-c: The return type of the following
21091 built-in functions was implemented as int not long long. Fix sign
21092 of return value for the unsigned version of vec_mulo and vec_mule.
21093 vector unsigned long long vec_bperm (vector unsigned long long,
21094 vector unsigned char)
21095 vector signed long long vec_mule (vector signed int,
21097 vector unsigned long long vec_mule (vector unsigned int,
21098 vector unsigned int)
21099 vector signed long long vec_mulo (vector signed int,
21101 vector unsigned long long vec_mulo (vector unsigned int,
21102 vector unsigned int)
21103 * doc/extend.texi: Fix the documentation for the built-in
21106 2017-06-07 Carl Love <cel@us.ibm.com>
21109 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21112 2017-06-07 Carl Love <cel@us.ibm.com>
21114 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21115 support, Generate doublehv for signed int/float for BE case only.
21117 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21119 * doc/invoke.texi (mcx16): Rewrite.
21121 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21123 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21124 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21125 *mov<mode>_softfloat, and an anonymous splitter): Use
21126 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21128 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21130 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21132 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21133 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21134 (rs6000_debug_reg_global): Adjust.
21135 (rs6000_init_hard_regno_mode_ok): Adjust.
21136 (rs6000_dbx_register_number): Adjust.
21137 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21138 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21139 Remove SPE_ACC and SPEFSCR.
21140 (REG_ALLOC_ORDER): Ditto.
21141 (FRAME_POINTER_REGNUM): Change to 111.
21142 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21143 (REG_CLASS_NAMES): Ditto.
21144 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21145 (REGISTER_NAMES): Ditto.
21146 (ADDITIONAL_REG_NAMES): Ditto.
21147 (rs6000_reg_names): Ditto.
21148 * config/rs6000/rs6000.md: Renumber some register number
21151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21153 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21155 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21156 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21158 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21159 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21160 Delete the SPE high registers.
21161 (REG_ALLOC_ORDER): Ditto.
21162 (enum reg_class): Remove SPE_HIGH_REGS.
21163 (REG_CLASS_NAMES): Ditto.
21164 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21165 (REGISTER_NAMES): Ditto.
21166 (rs6000_reg_names): Ditto.
21167 * doc/tm.texi.in: Remove SPE as example.
21168 * doc/tm.texi: Regenerate.
21170 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21172 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21173 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21174 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21175 * config/rs6000/rs6000.md (type): Remove "brinc".
21177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21179 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21180 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21181 * config/rs6000/linuxspe.h: Delete file.
21182 * config/rs6000/rs6000.md: Don't include spe.md.
21183 * config/rs6000/spe.h: Delete file.
21184 * config/rs6000/spe.md: Delete file.
21185 * config/rs6000/t-rs6000: Remove spe.md.
21187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21189 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21190 (reg_or_none500mem_operand): Delete.
21191 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21192 instead of reg_or_none500mem_operand.
21194 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21196 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21197 handling of SPE flags.
21198 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21200 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21202 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21204 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21205 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21206 paired_divv2sf3): Similar.
21207 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21208 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21209 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21211 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21212 Rename the paired_* instruction patterns.
21213 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21215 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21216 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21217 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21218 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21219 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21220 PAIRED_VECTOR_MODE.
21221 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21222 (spe_func_has_64bit_regs_p): Delete.
21223 (spe_expand_predicate_builtin): Delete.
21224 (spe_expand_evsel_builtin): Delete.
21225 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21226 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21227 (invalid_e500_subreg): Delete.
21228 (rs6000_legitimize_address): Always force_reg op2 as well, for
21229 paired single memory accesses.
21230 (rs6000_member_type_forces_blk): Delete.
21231 (rs6000_spe_function_arg): Delete.
21232 (rs6000_expand_unop_builtin): Delete SPE handling.
21233 (rs6000_expand_binop_builtin): Ditto.
21234 (spe_expand_stv_builtin): Delete.
21235 (bdesc_2arg_spe): Delete.
21236 (spe_expand_builtin): Delete.
21237 (spe_expand_predicate_builtin): Delete.
21238 (spe_expand_evsel_builtin): Delete.
21239 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21240 (spe_init_builtins): Delete.
21241 (spe_func_has_64bit_regs_p): Delete.
21242 (savres_routine_name): Delete "info" parameter. Adjust callers.
21243 (rs6000_emit_stack_reset): Ditto.
21244 (rs6000_dwarf_register_span): Delete.
21245 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21246 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21247 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21249 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21251 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21252 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21253 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21254 mulv2sf3, divv2sf3): Delete expanders.
21256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21258 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21260 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21262 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21263 * config/rs6000/rs6000.c: Ditto.
21265 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21267 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21268 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21269 comparison_operator.
21271 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21273 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21274 * config/rs6000/rs6000.opt: Ditto.
21275 * config/rs6000/t-rtems: Ditto.
21277 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21279 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21280 TARGET_E500_SINGLE by 0, simplify.
21281 * config/rs6000/rs6000.c: Ditto.
21282 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21283 (spe_build_register_parallel): Delete.
21284 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21285 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21286 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21287 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21288 (E500_CONVERT): Delete.
21289 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21291 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21293 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21294 * config/rs6000/dfp.md: Ditto.
21295 (negdd2, *negdd2_fpr): Merge.
21296 (absdd2, *absdd2_fpr): Merge.
21297 (negtd2, *negtd2_fpr): Merge.
21298 (abstd2, *abstd2_fpr): Merge.
21299 * config/rs6000/e500.h: Delete file.
21300 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21301 TARGET_FPRS by 1 and simplify.
21302 * config/rs6000/rs6000-c.c: Ditto.
21303 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21304 TARGET_DF_SPE by 0.
21305 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21307 * config/rs6000/rs6000.md: Ditto.
21308 (floatdidf2, *floatdidf2_fpr): Merge.
21309 (move_from_CR_gt_bit): Delete.
21310 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21311 (E500_CR_IOR_COMPARE): Delete.
21312 (All patterns that require !TARGET_FPRS): Delete.
21313 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21315 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21317 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21319 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21321 * graphds.c (add_edge): Intitialize edge's attached data.
21322 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21323 pointer parameter. Call pointed function on each edge during
21324 graph traversing. Skip traversing the edge when the function
21326 (graphds_dfs, graphds_scc): Ditto.
21327 (for_each_edge): New parameter. Pass the new parameter to callback
21329 * graphds.h (skip_edge_callback): New function pointer type.
21330 (graphds_dfs, graphds_scc): New function pointer parameter.
21331 (graphds_edge_callback, for_each_edge): New parameter.
21333 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21335 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21336 out code checking if runtime alias check is possible to below ...
21337 Call the new function.
21338 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21339 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21341 2017-06-07 Marek Polacek <polacek@redhat.com>
21344 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21345 TYPE_OVERFLOW_WRAPS check.
21347 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21349 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21350 if versioning is required.
21351 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21352 peeling with the check for versioning.
21354 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21356 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21357 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21358 Set true to new parameter if new ssa variable is defined.
21359 (vect_gen_vector_loop_niters): Refactor. Set range information
21360 for the new vector loop bound variable.
21361 (vect_do_peeling): Ditto.
21363 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21365 * tree-affine.c (ssa.h): Include header file.
21366 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21367 has wrapping overflow behavior.
21369 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21371 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21373 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21375 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21376 (tree_to_aff_combination): ... here.
21378 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21380 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21381 reg_pressure model function.
21382 (ivopts_global_cost_for_size): Delete.
21383 (determine_set_costs, iv_ca_recount_cost): Call new model function
21384 ivopts_estimate_reg_pressure.
21386 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21388 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21389 expensive than udiv. Remove floating point cases from mod.
21391 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21393 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21394 Increase idiv cost.
21396 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21398 * config/aarch64/aarch64.md
21399 (copysignsf3): Fix mask generation.
21401 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21403 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21405 (class dump_manager): Add register_dumps method.
21406 * dumpfile.c: Include langhooks.h.
21407 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21408 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21409 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21410 (dump_manager::dump_register): Start with 512 entries instead of 32.
21411 (dump_manager::register_dumps): New method.
21412 * toplev.c (general_init): Instead of invoking register_dumps
21413 langhook, invoke register_dumps method on the dump manager.
21414 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21417 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21419 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21420 Say that # requires an associated define_split to exist, and that
21421 the define_split must be suitable for use after register allocation.
21423 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21425 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21426 (compute_outgoing_frequencies): Also initialize zero counts.
21427 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21428 around loops; preserve more of profile when nothing changes.
21430 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21432 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21434 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21435 * config/arm/arm-cpu-cdata.h: Regenerate.
21436 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21437 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21438 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21439 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21440 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21442 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21444 * config/arm/t-rmprofile: Likewise.
21445 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21447 2017-06-06 David S. Miller <davem@davemloft.net>
21450 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21451 blockage if function uses alloca.
21453 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21455 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21456 New "uid" fields to hold pretty-print IDs of group and ref.
21457 Memory references are now identified as <group_id>:<ref_id>
21458 instead of using [random] addresses.
21459 (dump_mem_details): Simplify, no functional change.
21460 (dump_mem_ref): Simplify and make output more concise.
21461 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21462 (find_or_create_group): Initialize group uid.
21463 (record_ref): Initialize ref uid. Improve debug output.
21464 (prune_group_by_reuse, should_issue_prefetch_p,)
21465 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21466 (mark_nontemporal_store, determine_loop_nest_reuse):
21467 Improve debug output.
21469 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21471 * dbgcnt.def (prefetch): New debug counter.
21472 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21473 (schedule_prefetches): Stop issueing prefetches if debug counter
21476 2017-06-06 Tom de Vries <tom@codesourcery.com>
21478 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21479 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21481 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21483 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21484 Use aarch64_reg_or_zero predicate for operand 4.
21485 (aarch64_compare_and_swap<mode> define_insn_and_split):
21486 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21487 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21489 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21491 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21492 (arm_compute_save_core_reg_mask): This.
21493 (thumb1_compute_save_reg_mask): Rename into ...
21494 (thumb1_compute_save_core_reg_mask): This.
21495 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21496 (arm_compute_frame_layout): Likewise.
21498 2017-06-06 Richard Biener <rguenther@suse.de>
21500 PR tree-optimization/80974
21501 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21502 keep or clear leaders SSA info.
21504 2017-06-06 Tom de Vries <tom@codesourcery.com>
21506 * config/nvptx/nvptx.c (split_mode_p): New function.
21507 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21509 2017-06-06 Tom de Vries <tom@codesourcery.com>
21511 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21513 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21516 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21519 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21521 * shrink-wrap.c (handle_simple_exit): Update profile.
21522 (try_shrink_wrapping): Upate profile.
21524 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21526 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21527 (tree_guess_outgoing_edge_probabilities): New.
21528 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21529 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21531 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21533 * ipa-split.c (split_function): Initialize return bb profile.
21535 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21537 * profile.c (compute_branch_probabilities): Also initialize
21538 EXIT_BLOCK profile.
21540 2017-06-06 Richard Biener <rguenther@suse.de>
21542 PR tree-optimization/80928
21543 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21544 (vect_analyze_loop_operations): Properly guard analysis for
21546 (vect_transform_loop): Likewise.
21547 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21548 (vect_model_induction_cost): Do not cost for pure SLP.
21549 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21550 of induction in inner loop vectorization.
21551 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21552 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21553 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21555 (vect_analyze_slp_cost_1): Cost induction.
21556 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21557 (vect_get_slp_vect_defs): Likewise.
21558 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21559 (vect_transform_stmt): Handle SLP reductions.
21560 * tree-vectorizer.h (vectorizable_induction): Adjust.
21562 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21564 * config/rs6000/rs6000.c (make_resolver_func): Update
21565 init_lowered_empty_function call.
21567 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21569 * doc/invoke.texi: Document the -fprofile-abs-path option.
21570 * common.opt (fprofile-abs-path): New option.
21571 * gcov-io.h (gcov_write_filename): Declare.
21572 * gcov-io.c (gcov_write_filename): New function.
21573 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21574 * profile.c (output_location): Likewise.
21576 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21578 * shring-wrap.c: Revert accidental commit.
21580 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21582 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21584 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21586 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21588 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21589 profile in callgraph edge.
21590 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21591 (apply_scale): Likewise.
21592 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21593 Also scale profile when inlining function with zero profile.
21594 (initialize_cfun): Update exit block profile even when it is zero.
21595 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21596 when profile is read.
21598 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21600 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21601 (CLONE_*): New constants to define the processors we can generate
21602 code for with the target_clone attribute.
21603 (rs6000_clone_map): New array to identify which clone processors
21604 the current program is running on.
21605 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21606 target_clone attribute.
21607 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21608 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21609 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21610 (cpu_expand_builtin): Add support for target_clone attribute.
21611 (rs6000_valid_attribute_p): Allow "default" attribute.
21612 (get_decl_name): New debug function to simplify printing the
21613 current function name in debugging statements.
21614 (rs6000_clone_priority): New functions to support the target_clone
21615 attribute, and be able to generate code to switch between ISA 2.05
21616 through ISA 3.0 (power6 through power9).
21617 (rs6000_compare_version_priority): Likewise.
21618 (rs6000_get_function_versions_dispatcher): Likewise.
21619 (make_resolver_func): Likewise.
21620 (add_condition_to_bb): Likewise.
21621 (dispatch_function_versions): Likewise.
21622 (rs6000_generate_version_dispatcher_body): Likewise.
21623 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21624 (fusion_gpr_load_p): Fix a spacing issue.
21625 * doc/extend.texi (Common Function Attributes): Document that the
21626 PowerPC supports the target_clone attribute.
21628 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21630 * config/arm/arm.h: explain F symbol found in description of ARM
21631 register allocation in its legend.
21633 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21635 * config/mips/frame-header-opt.c: Include profile-count.h.
21636 * config/riscv/riscv.c: Include profile-count.h
21638 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21640 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21642 (sm_set_flag_if_changed): Add bbs field.
21643 (execute_sm_if_changed_flag_set): Pass BBS.
21644 (execute_sm): Update.
21646 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21648 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21651 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21653 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21654 (peephole2): New peephole2 to emit the above.
21655 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21657 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21659 * config/aarch64/aarch64.c (define_peephole2 above
21660 *sub_<shift>_<mode>): New peephole.
21662 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21664 * config/i386/i386.c (make_resolver_func): Update.
21665 * Makefile.in: Add profile-count.h and profile-count.o
21666 * auto-profile.c (afdo_indirect_call): Update to new API.
21667 (afdo_set_bb_count): Update.
21668 (afdo_propagate_edge): Update.
21669 (afdo_propagate_circuit): Update.
21670 (afdo_calculate_branch_prob): Update.
21671 (afdo_annotate_cfg): Update.
21672 * basic-block.h: Include profile-count.h
21673 (struct edge_def): Turn count to profile_count.
21674 (struct basic_block_def): Likewie.
21675 (REG_BR_PROB_BASE): Move to profile-count.h
21676 (RDIV): Move to profile-count.h
21677 * bb-reorder.c (max_entry_count): Turn to profile_count.
21678 (find_traces): Update.
21679 (rotate_loop):Update.
21680 (connect_traces):Update.
21681 (sanitize_hot_paths):Update.
21682 * bt-load.c (migrate_btr_defs): Update.
21683 * cfg.c (RDIV): Remove.
21684 (init_flow): Use alloc_block.
21685 (alloc_block): Uninitialize count.
21686 (unchecked_make_edge): Uninitialize count.
21687 (check_bb_profile): Update.
21688 (dump_edge_info): Update.
21689 (dump_bb_info): Update.
21690 (update_bb_profile_for_threading): Update.
21691 (scale_bbs_frequencies_int): Update.
21692 (scale_bbs_frequencies_gcov_type): Update.
21693 (scale_bbs_frequencies_profile_count): New.
21694 * cfg.h (update_bb_profile_for_threading): Update.
21695 (scale_bbs_frequencies_profile_count): Declare.
21696 * cfgbuild.c (compute_outgoing_frequencies): Update.
21697 (find_many_sub_basic_blocks): Update.
21698 * cfgcleanup.c (try_forward_edges): Update.
21699 (try_crossjump_to_edge): Update.
21700 * cfgexpand.c (expand_gimple_tailcall): Update.
21701 (construct_exit_block): Update.
21702 * cfghooks.c (verify_flow_info): Update.
21703 (dump_bb_for_graph): Update.
21704 (split_edge): Update.
21705 (make_forwarder_block): Update.
21706 (duplicate_block): Update.
21707 (account_profile_record): Update.
21708 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21709 (get_estimated_loop_iterations): Update.
21710 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21711 (single_likely_exit): Update.
21712 * cfgloopmanip.c (scale_loop_profile): Update.
21714 (set_zero_probability): Update.
21715 (lv_adjust_loop_entry_edge): Update.
21716 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21717 (purge_dead_edges): Update.
21718 (rtl_account_profile_record): Update.
21719 * cgraph.c (cgraph_node::create): Uninitialize count.
21720 (symbol_table::create_edge): Uninitialize count.
21721 (cgraph_update_edges_for_call_stmt_node): Update.
21722 (cgraph_edge::dump_edge_flags): Update.
21723 (cgraph_node::dump): Update.
21724 (cgraph_edge::maybe_hot_p): Update.
21725 * cgraph.h: Include profile-count.h
21726 (create_clone), create_edge, create_indirect_edge): Update.
21727 (cgraph_node): Turn count to profile_count.
21728 (cgraph_edge0: Likewise.
21729 (make_speculative, clone): Update.
21730 (create_edge): Update.
21731 (init_lowered_empty_function): Update.
21732 * cgraphclones.c (cgraph_edge::clone): Update.
21733 (duplicate_thunk_for_node): Update.
21734 (cgraph_node::create_clone): Update.
21735 * cgraphunit.c (cgraph_node::analyze): Update.
21736 (cgraph_node::expand_thunk): Update.
21737 * final.c (dump_basic_block_info): Update.
21738 * gimple-streamer-in.c (input_bb): Update.
21739 * gimple-streamer-out.c (output_bb): Update.
21740 * graphite.c (print_global_statistics): Update.
21741 (print_graphite_scop_statistics): Update.
21742 * hsa-brig.c: Include basic-block.h.
21743 * hsa-dump.c: Include basic-block.h.
21744 * hsa-gen.c (T sum_slice): Update.
21745 (convert_switch_statements):Update.
21746 * hsa-regalloc.c: Include basic-block.h.
21747 * ipa-chkp.c (chkp_produce_thunks): Update.
21748 * ipa-cp.c (struct caller_statistics): Update.
21749 (init_caller_stats): Update.
21750 (gather_caller_stats): Update.
21751 (ipcp_cloning_candidate_p): Update.
21752 (good_cloning_opportunity_p): Update.
21753 (get_info_about_necessary_edges): Update.
21754 (dump_profile_updates): Update.
21755 (update_profiling_info): Update.
21756 (update_specialized_profile): Update.
21757 (perhaps_add_new_callers): Update.
21758 (decide_about_value): Update.
21759 (ipa_cp_c_finalize): Update.
21760 * ipa-devirt.c (struct odr_type_warn_count): Update.
21761 (struct decl_warn_count): Update.
21762 (struct final_warning_record): Update.
21763 (possible_polymorphic_call_targets): Update.
21764 (ipa_devirt): Update.
21765 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21766 * ipa-icf.c (sem_function::merge): Update.
21767 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21768 * ipa-inline.c (compute_uninlined_call_time): Update.
21769 (compute_inlined_call_time): Update.
21770 (want_inline_small_function_p): Update.
21771 (want_inline_self_recursive_call_p): Update.
21772 (edge_badness): Update.
21773 (lookup_recursive_calls): Update.
21774 (recursive_inlining): Update.
21775 (inline_small_functions): Update.
21776 (dump_overall_stats): Update.
21777 (dump_inline_stats): Update.
21778 * ipa-profile.c (ipa_profile_generate_summary): Update.
21779 (ipa_propagate_frequency): Update.
21780 (ipa_profile): Update.
21781 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21782 * ipa-utils.c (ipa_merge_profiles): Update.
21783 * loop-doloop.c (doloop_modify): Update.
21784 * loop-unroll.c (report_unroll): Update.
21785 (unroll_loop_runtime_iterations): Update.
21786 * lto-cgraph.c (lto_output_edge): Update.
21787 (lto_output_node): Update.
21788 (input_node): Update.
21789 (input_edge): Update.
21790 (merge_profile_summaries): Update.
21791 * lto-streamer-in.c (input_cfg): Update.
21792 * lto-streamer-out.c (output_cfg): Update.
21793 * mcf.c (create_fixup_graph): Update.
21794 (adjust_cfg_counts): Update.
21795 (sum_edge_counts): Update.
21796 * modulo-sched.c (sms_schedule): Update.
21797 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21798 * predict.c (maybe_hot_count_p): Update.
21799 (probably_never_executed): Update.
21800 (dump_prediction): Update.
21801 (combine_predictions_for_bb): Update.
21802 (propagate_freq): Update.
21803 (handle_missing_profiles): Update.
21804 (counts_to_freqs): Update.
21805 (rebuild_frequencies): Update.
21806 (force_edge_cold): Update.
21807 * predict.h: Include profile-count.h
21808 (maybe_hot_count_p, counts_to_freqs): UPdate.
21809 * print-rtl-function.c: Do not include cfg.h
21810 * print-rtl.c: Include basic-block.h
21811 * profile-count.c: New file.
21812 * profile-count.h: New file.
21813 * profile.c (is_edge_inconsistent): Update.
21814 (correct_negative_edge_counts): Update.
21815 (is_inconsistent): Update.
21816 (set_bb_counts): Update.
21817 (read_profile_edge_counts): Update.
21818 (compute_frequency_overlap): Update.
21819 (compute_branch_probabilities): Update; Initialize and deinitialize
21821 (branch_prob): Update.
21822 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21823 (edge_gcov_count): New.
21824 (bb_gcov_count): New.
21825 * shrink-wrap.c (try_shrink_wrapping): Update.
21826 * tracer.c (better_p): Update.
21827 * trans-mem.c (expand_transaction): Update.
21828 (ipa_tm_insert_irr_call): Update.
21829 (ipa_tm_insert_gettmclone_call): Update.
21830 * tree-call-cdce.c: Update.
21831 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21832 (gimple_duplicate_sese_tail): Update.
21833 (gimple_account_profile_record): Update.
21834 (execute_fixup_cfg): Update.
21835 * tree-inline.c (copy_bb): Update.
21836 (copy_edges_for_bb): Update.
21837 (initialize_cfun): Update.
21838 (freqs_to_counts): Update.
21839 (copy_cfg_body): Update.
21840 (expand_call_inline): Update.
21841 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21842 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21843 (try_unroll_loop_completely): Update.
21844 (try_peel_loop): Update.
21845 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21846 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21847 * tree-ssa-loop-split.c (connect_loops): Update.
21848 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21849 * tree-ssa-reassoc.c (branch_fixup): Update.
21850 * tree-ssa-tail-merge.c (replace_block_by): Update.
21851 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21852 (compute_path_counts): Update.
21853 (update_profile): Update.
21854 (recompute_probabilities): Update.
21855 (update_joiner_offpath_counts): Update.
21856 (estimated_freqs_path): Update.
21857 (freqs_to_counts_path): Update.
21858 (clear_counts_path): Update.
21859 (ssa_fix_duplicate_block_edges): Update.
21860 (duplicate_thread_path): Update.
21861 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21862 (struct switch_conv_info): Update.
21863 * tree-tailcall.c (decrease_profile): Update.
21864 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21865 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21866 * value-prof.c (check_counter): Update.
21867 (gimple_divmod_fixed_value): Update.
21868 (gimple_mod_pow2): Update.
21869 (gimple_mod_subtract): Update.
21870 (gimple_ic_transform): Update.
21871 (gimple_stringop_fixed_value): Update.
21872 * value-prof.h (gimple_ic): Update.
21874 2017-06-02 Carl Love <cel@us.ibm.com>
21876 * config/rs6000/rs6000-c: Add support for built-in functions
21877 vector double vec_doublee (vector signed int);
21878 vector double vec_doublee (vector unsigned int);
21879 vector double vec_doublee (vector float);
21880 vector double vec_doubleh (vector signed int);
21881 vector double vec_doubleh (vector unsigned int);
21882 vector double vec_doubleh (vector float);
21883 vector double vec_doublel (vector signed int);
21884 vector double vec_doublel (vector unsigned int);
21885 vector double vec_doublel (vector float);
21886 vector double vec_doubleo (vector signed int);
21887 vector double vec_doubleo (vector unsigned int);
21888 vector double vec_doubleo (vector float);.
21889 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21890 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21892 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21893 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21894 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21896 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21897 vec_doublel, vec_doubleh.
21898 * doc/extend.texi: Update the built-in documentation file for the
21899 new built-in functions.
21901 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21904 * ipa-inline-analysis.c (free_growth_caches): Set
21905 edge_removal_hook_holder to NULL after removing it.
21907 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21909 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21910 comparision with zero.
21912 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21914 for early expansion of vec_min and vec_max builtins.
21915 (builtin_function_type): Add min/max unsigned variants to those
21916 identified as having unsigned arguments.
21918 2017-06-02 Olivier Hainque <hainque@adacore.com>
21920 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21922 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21924 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21925 Use VALL_F16 iterator rather than VALL.
21927 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21929 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21930 Emit CBNZ inside loop when doing a strong exchange and comparing
21931 against zero. Generate the CC flags after the loop.
21933 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21935 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21936 (dl_section_ref): New.
21937 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21938 On AIX, append an expression to subtract the size of the
21939 section length to dl_section_ref.
21941 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21943 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21944 for early expansion of vector absolute builtins.
21946 2017-06-02 Richard Biener <rguenther@suse.de>
21948 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21949 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21951 2017-06-02 Richard Biener <rguenther@suse.de>
21953 PR tree-optimization/80948
21954 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21955 stmt order as well.
21957 2017-06-02 Richard Biener <rguenther@suse.de>
21959 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21961 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21962 for inductions as relevant.
21964 2017-06-02 Richard Biener <rguenther@suse.de>
21966 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21967 (vectorizable_induction): ... this. Remove dead code.
21969 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21971 * builtins. (expand_builtin_alloca): Remove second parameter and
21972 infer its value from the first parameter instead.
21973 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21975 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21977 PR rtl-optimization/80903
21978 * loop-doloop.c (add_test): Unshare sequence.
21980 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21982 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21984 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21986 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21988 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21989 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21990 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21991 (xlogue_layout::get_stub_rtx): Make static.
21992 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21993 (xlogue_layout::compute_stub_managed_regs): Rename to...
21994 (xlogue_layout::count_stub_managed_regs): ...this.
21995 (xlogue_layout::is_stub_managed_reg): New function.
21996 (xlogue_layout::m_stub_names): Rename to...
21997 (xlogue_layout::s_stub_names): ...this, make static.
21998 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21999 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22000 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22001 xlogue_layout::s_stub_names): Instantiate statics.
22002 (stub_managed_regs): Remove.
22003 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22004 (disable_call_ms2sysv_xlogues): Rename to...
22005 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22006 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22008 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22009 change after reload_completed.
22010 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22012 (ix86_expand_prologue): Likewise.
22013 (ix86_expand_epilogue): Likewise.
22014 (ix86_expand_split_stack_prologue): Likewise.
22015 (ix86_compute_frame_layout): Remove frame parameter ...
22016 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22017 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22019 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22020 (ix86_frame): Move from here ...
22021 * config/i386/i386.h (ix86_frame): ... to here.
22022 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22023 complete ix86_frame data structure instead. Remove some_ld_name.
22025 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22027 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22028 symbols that hold a DECL_VALUE_EXPR.
22030 2017-06-01 Martin Jambor <mjambor@suse.cz>
22032 PR tree-optimization/80898
22033 * tree-sra.c (process_subtree_disqualification): Removed.
22034 (disqualify_candidate): Do not acll
22035 process_subtree_disqualification.
22036 (subtree_mark_written_and_enqueue): New function.
22037 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22038 RHS has been disqualified and re-queue LHS if necessary. Apart
22039 from that, ignore disqualified RHS.
22041 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22043 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22044 address fetch for z10 or later.
22046 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22048 * config/arc/arc.md (tst_movb): Add guard when splitting.
22050 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22052 * config/arc/arc.c (arc_can_eliminate): Test against
22053 arc_frame_pointer_needed.
22055 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22057 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22058 to prevent store reordering.
22059 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22060 (type): Add block type.
22061 (stack_tie): Define special instruction to be used in
22064 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22066 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22067 constraint. It is not valid for the pattern.
22068 (noncommutative_binary_comparison): Likewise.
22070 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22072 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22075 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22077 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22078 be used by the reg-alloc.
22080 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22082 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22083 reg-alloc when having mul64 or mul32x16 instructions.
22084 (mulsidi3): Likewise.
22085 (umulsidi3): Likewise.
22086 (mulsi32x16): New pattern.
22087 (mulsi64): Likewise.
22088 (mulsidi64): Likewise.
22089 (umulsidi64): Likewise.
22090 (MUL32x16_REG): Define.
22091 (mul64_600): Use MUL32x16_REG.
22092 (mac64_600): Likewise.
22093 (umul64_600): Likewise.
22094 (umac64_600): Likewise.
22096 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22098 * config/arc/arc.md (mulsi3_700): Make it commutative.
22100 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22102 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22104 (*sign_extendsidi2_insn): Likewise for movstosw.
22106 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22108 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22109 the type of the input discriminant value. Convert the
22110 discriminant value of signedness vary.
22112 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22114 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22115 Add to -Wall section.
22117 2017-06-01 Richard Biener <rguenther@suse.de>
22119 PR middle-end/66313
22120 * fold-const.c (fold_plusminus_mult_expr): If the factored
22121 factor may be zero use a wrapping type for the inner operation.
22122 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22123 and handle moved defs.
22124 (process_assignment): Properly guard the unary op case. Return a
22125 tri-state indicating that moving the stmt before the call may allow
22126 to continue. Pass through to_move.
22127 (find_tail_calls): Handle moving unrelated defs before
22130 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22133 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22134 splitter result in the canonical way.
22136 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22138 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22139 also for 32bit target. Update insn attributes.
22140 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22142 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22144 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22145 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22146 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22147 (_mm_maskz_min_ss): New intrinsics.
22149 2017-05-31 Martin Liska <mliska@suse.cz>
22151 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22152 Change comment style to one we normally use.
22153 (vectorizable_reduction): Likewise.
22154 (vectorizable_induction): Likewise.
22155 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22156 (vectorizable_call): Likewise.
22157 (vectorizable_simd_clone_call): Likewise.
22158 (vectorizable_conversion): Likewise.
22159 (vectorizable_assignment): Likewise.
22160 (vectorizable_shift): Likewise.
22161 (vectorizable_operation): Likewise.
22162 (vectorizable_store): Likewise.
22163 (vectorizable_load): Likewise.
22164 * tree-vectorizer.h: Likewise.
22166 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22168 * passes.c (emergency_dump_function): New.
22169 * tree-pass.h (emergency_dump_function): Declare.
22170 * plugin.c (plugins_internal_error_function): Remove.
22171 * plugin.h (plugins_internal_error_function): Remove declaration.
22172 * toplev.c (internal_error_function): New static function. Use it...
22173 (general_init): ...here.
22175 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22177 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22178 (arc_rtx_costs): Add costs for new patterns.
22179 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22180 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22182 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22184 * tree-ssa-strlen.c (get_next_strinfo): New function.
22185 (get_stridx_plus_constant): Use it.
22186 (zero_length_string): Likewise.
22187 (adjust_related_strinfos): Likewise.
22188 (adjust_last_stmt): Likewise.
22190 2017-05-31 Richard Biener <rguenther@suse.de>
22193 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22194 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22196 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22198 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22199 loop_vinfo argument and use of dependence distance vectors.
22200 Check instead whether the two references differ only in their
22201 initial value and assume that they have the same alignment if the
22202 difference is a multiple of the vector alignment.
22203 (vect_analyze_data_refs_alignment): Update call accordingly.
22205 2017-05-31 Martin Liska <mliska@suse.cz>
22208 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22210 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22212 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22213 (create_intersect_range_checks): Move from ...
22214 * tree-data-ref.c (create_intersect_range_checks_index)
22215 (create_intersect_range_checks): ... to here.
22216 (create_runtime_alias_checks): New function factored from ...
22217 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22218 here. Call above function.
22219 * tree-data-ref.h (create_runtime_alias_checks): New function.
22221 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22223 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22224 segment length for dr_b and compute it in wide_int.
22226 2017-05-31 Richard Biener <rguenther@suse.de>
22228 PR tree-optimization/80906
22229 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22230 and pass through iv_map.
22231 (copy_bb_and_scalar_dependences): Adjust.
22232 (translate_pending_phi_nodes): Likewise.
22233 (copy_loop_close_phi_args): Handle code-generating IVs instead
22236 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22238 * diagnostic-color.c (color_dict): Add "type-diff".
22239 (parse_gcc_colors): Update comment.
22240 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22241 -fdiagnostics-show-template-tree and -fno-elide-type.
22242 (GCC_COLORS): Add type-diff to example.
22244 (-fdiagnostics-show-template-tree): New.
22245 (-fno-elide-type): New.
22246 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22247 the pp_format_decoder callback. Call any m_format_postprocessor's
22249 (pretty_printer::pretty_printer): Initialize
22250 m_format_postprocessor.
22251 (pretty_printer::~pretty_printer): Delete any
22252 m_format_postprocessor.
22253 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22254 (class format_postprocessor): New class.
22255 (struct pretty_printer::format_decoder): Document the new parameters.
22256 (struct pretty_printer::m_format_postprocessor): New field.
22257 * tree-diagnostic.c (default_tree_printer): Update for new
22258 bool and const char ** params.
22259 * tree-diagnostic.h (default_tree_printer): Likewise.
22261 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22263 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22264 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22266 rs6000_gen_cell_microcode code.
22267 (rs6000_final_prescan_insn): Delete.
22268 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22269 "warn-cell-microcode" entries.
22270 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22271 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22272 throughout. Change cc_reg_not_micro_cr0_operand to
22273 cc_reg_not_cr0_operand throughout.
22274 (*extendhi<mode>2_noload): Delete.
22275 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22276 (mwarn-cell-microcode): Delete.
22277 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22278 -mgen-cell-microcode and -mwarn-cell-microcode.
22280 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22283 * config/i386/constraints.md (Yd): New constraint.
22285 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22286 and (?Yd, r) alternatives. Update insn attributes.
22287 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22288 and (?*Yd, r) alternatives. Update insn attributes.
22289 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22291 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22293 * gimplify.c (gimplify_modify_expr): Don't create a
22294 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22297 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22299 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22301 2017-05-30 Richard Biener <rguenther@suse.de>
22303 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22304 and reduc_def fields.
22305 (STMT_VINFO_REDUC_TYPE): New define.
22306 (STMT_VINFO_REDUC_DEF): Likewise.
22307 (vect_force_simple_reduction): Adjust prototype.
22308 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22309 (vect_is_simple_reduction): Remove check_reduction argument.
22310 (vect_force_simple_reduction): Adjust and set
22311 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22312 (vectorizable_reduction): Do not re-do reduction analysis
22313 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22314 * tree-parloops.c (gather_scalar_reductions): Adjust.
22316 2017-05-30 Richard Biener <rguenther@suse.de>
22318 PR middle-end/80901
22319 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22322 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22324 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22325 Introduce unknown_misalignment parameter and remove vf.
22326 (vect_peeling_hash_get_lowest_cost):
22327 Pass unknown_misalignment parameter.
22328 (vect_enhance_data_refs_alignment):
22329 Fix unsupportable data ref treatment.
22331 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22333 * tree-vect-data-refs.c (vect_get_data_access_cost):
22334 Workaround for SLP handling.
22335 (vect_enhance_data_refs_alignment):
22336 Compute costs for doing no peeling at all, compare to the best
22337 peeling costs so far and avoid peeling if cheaper.
22339 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22341 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22342 Return peeling info and set costs to zero for unlimited cost
22344 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22345 with unknown misalignment. Compute and costs for unknown
22346 misalignment, compare them to the costs for known misalignment
22347 and choose the cheapest for peeling.
22349 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22351 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22352 (vect_get_peeling_costs_all_drs): Create function.
22353 (vect_peeling_hash_get_lowest_cost):
22354 Use vect_get_peeling_costs_all_drs.
22355 (vect_peeling_supportable): Create function.
22356 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22358 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22360 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22361 DR_HAS_NEGATIVE_STEP.
22362 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22363 (vect_enhance_data_refs_alignment): Use.
22364 (vect_duplicate_ssa_name_ptr_info): Use.
22365 * tree-vectorizer.h (dr_misalignment): Use.
22366 (known_alignment_for_access_p): Use.
22368 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22371 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22373 (has_section_name): New function.
22375 2017-05-30 Martin Liska <mliska@suse.cz>
22378 * auto-profile.c (get_function_decl_from_block): Fix
22381 2017-05-30 Richard Biener <rguenther@suse.de>
22383 PR middle-end/80876
22384 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22386 2017-05-30 Martin Liska <mliska@suse.cz>
22388 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22389 * dumpfile.h (struct dump_file_info): Remove ctors.
22391 2017-05-30 Martin Liska <mliska@suse.cz>
22393 * predict.def: Fix GNU coding style.
22395 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22397 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22398 Mark 'to' argument with ATTRIBUTE_UNUSED.
22400 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22402 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22403 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22404 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22407 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22409 * doc/install.texi (Options specification): Restore entry of
22410 --enable-sjlj-exceptions.
22412 2017-05-27 Michael Eager <eager@eagercon.com>
22415 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22417 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22419 * config/microblaze/microblaze.h
22420 (FIXED_REGISTERS): Update in macro.
22421 (CALL_USED_REGISTERS): Update in macro.
22423 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22425 * doc/install.texi: Add links to macOS binary distributions.
22427 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22431 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22433 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22435 2017-05-26 Martin Liska <mliska@suse.cz>
22437 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22439 2017-05-26 Martin Liska <mliska@suse.cz>
22441 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22442 always leading ';; '.
22443 (dump_bb_info): Likewise.
22444 (brief_dump_cfg): Likewise.
22445 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22446 * dumpfile.c: Remove usage of TDF_VERBOSE.
22447 * dumpfile.h (enum dump_kind): Likewise.
22448 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22449 * print-tree.c (debug_verbose): Remove.
22450 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22451 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22452 * tree-diagnostic.c (default_tree_printer): Replace
22453 TDF_DIAGNOSTIC with TDF_SLIM.
22455 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22457 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22458 in parameter loop, rather than loop_vinfo.
22459 (create_intersect_range_checks): Ditto.
22460 (vect_create_cond_for_alias_checks): Update call to above functions.
22462 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22464 PR tree-optimization/80815
22465 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22466 for merging runtime alias checks. Handle negative DR_STEPs.
22468 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22470 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22472 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22473 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22474 out code pruning runtime alias checks.
22475 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22476 factored out from above.
22477 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22479 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22481 (prune_runtime_alias_test_list): New decalaration.
22483 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22485 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22486 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22487 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22488 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22489 (operator==, comp_dr_with_seg_len_pair): Ditto.
22490 (vect_prune_runtime_alias_test_list): Ditto.
22492 2017-05-26 Martin Liska <mliska@suse.cz>
22495 * params.def: Bound partial-inlining-entry-probability param.
22497 2017-05-26 Marek Polacek <polacek@redhat.com>
22500 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22503 2017-05-26 Richard Biener <rguenther@suse.de>
22505 PR tree-optimization/80842
22506 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22509 2017-05-26 Richard Biener <rguenther@suse.de>
22511 PR tree-optimization/80844
22512 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22514 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22516 * doc/md.texi (Machine Constraints): Update x86 family
22517 machine constraints section to match 'config/i386/constraints.md'.
22519 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22521 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22523 2017-05-25 Nathan Sidwell <nathan@acm.org>
22525 * doc/invoke.texi (--enable-languages): Update documentation.
22527 2017-05-25 Martin Liska <mliska@suse.cz>
22529 * dumpfile.c: Add TDF_FOLDING.
22530 * dumpfile.h (enum dump_kind): Likewise.
22531 * genmatch.c (dt_simplify::gen_1): Use it.
22533 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22535 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22537 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22539 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22540 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22542 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22544 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22545 * match.pd (X == C): Rewrite it here.
22546 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22547 with_certain_nonzero_bits2): New predicates.
22548 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22550 2017-05-24 Nathan Sidwell <nathan@acm.org>
22552 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22555 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22558 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22560 * config/powerpcspe: New port. Files are copied from the rs6000
22561 port, with "rs6000" in filenames replaced by "powerpcspe".
22563 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22565 PR rtl-optimization/80754
22566 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22568 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22570 * config/sparc/sparc.md (length): Return the correct value for -mflat
22571 sibcalls to match output_sibcall.
22573 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22577 * config/rs6000/rs6000.c (struct machine_function): Add new field
22579 (rs6000_get_separate_components): Init that field, use it.
22580 (rs6000_components_for_bb): Use the field.
22582 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22584 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22586 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22588 PR middle-end/80823
22589 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22591 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22594 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22595 address against address_operand predicate.
22596 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22598 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22600 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22601 parameters passed indirectly.
22603 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22605 * config/i386/i386.md (*movdi_internal): Remove SSE4
22606 alternative 18 (?r, *v). Update insn attributes.
22607 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22608 Update insn attributes.
22609 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22610 Update insn attributes.
22611 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22612 alternative 1 (r, v). Remove isa attribute.
22613 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22614 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22615 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22617 2017-05-23 Tom de Vries <tom@codesourcery.com>
22619 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22622 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22624 * cgraphunit.c (symbol_table::process_new_functions): Update.
22625 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22626 (inline_generate_summary): Rename to ...
22627 (ipa_fn_summary_generate): ... this one.
22628 (inline_read_summary): Rename to ...
22629 (ipa_fn_summary_read): ... this one.
22630 (inline_write_summary): Rename to ...
22631 (ipa_fn_summary_write): ... this one.
22632 (inline_free_summary): Rename to ...
22633 (ipa_free_fn_summary): ... this one.
22634 (pass_data_local_fn_summary, pass_local_fn_summary,
22635 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22636 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22637 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22638 make_pass_ipa_fn_summary): New.
22639 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22640 inline_write_summary, inline_free_summary): Remove.
22641 (ipa_free_fn_summary) : New.
22642 * ipa-inline.c (ipa_inline): Update.
22643 (pass_ipa_inline): Do not generate summaries.
22644 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22646 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22647 and add pass_ipa_fn_summary.
22648 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22650 (make_pass_inline_parameters): Remove.
22652 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22654 * omp-low.c (struct omp_context): Remove "default_kind" member.
22657 * omp-offload.c (execute_oacc_device_lower): Remove the
22658 parallelism dimensions function attributes for unparallelized
22659 OpenACC kernels constructs.
22661 2017-05-23 Martin Liska <mliska@suse.cz>
22663 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22665 (cgraph_edge::make_speculative): Likewise.
22666 (cgraph_edge::resolve_speculation): Likewise.
22667 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22668 (cgraph_node::dump): Likewise.
22669 * cgraph.h: Likewise.
22670 * cgraphunit.c (analyze_functions): Likewise.
22671 (symbol_table::compile): Likewise.
22672 * ipa-cp.c (print_all_lattices): Likewise.
22673 (determine_versionability): Likewise.
22674 (initialize_node_lattices): Likewise.
22675 (ipcp_verify_propagated_values): Likewise.
22676 (estimate_local_effects): Likewise.
22677 (update_profiling_info): Likewise.
22678 (create_specialized_node): Likewise.
22679 (perhaps_add_new_callers): Likewise.
22680 (decide_about_value): Likewise.
22681 (decide_whether_version_node): Likewise.
22682 (identify_dead_nodes): Likewise.
22683 (ipcp_store_bits_results): Likewise.
22684 * ipa-devirt.c (dump_targets): Likewise.
22685 (ipa_devirt): Likewise.
22686 * ipa-icf.c (sem_item::dump): Likewise.
22687 (sem_function::equals): Likewise.
22688 (sem_variable::equals): Likewise.
22689 (sem_item_optimizer::read_section): Likewise.
22690 (sem_item_optimizer::execute): Likewise.
22691 (congruence_class::dump): Likewise.
22692 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22693 (dump_inline_summary): Likewise.
22694 (estimate_node_size_and_time): Likewise.
22695 (inline_analyze_function): Likewise.
22696 * ipa-inline-transform.c (inline_call): Likewise.
22697 * ipa-inline.c (report_inline_failed_reason): Likewise.
22698 (want_early_inline_function_p): Likewise.
22699 (edge_badness): Likewise.
22700 (update_edge_key): Likewise.
22701 (inline_small_functions): Likewise.
22702 * ipa-profile.c (ipa_profile): Likewise.
22703 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22704 (ipa_make_edge_direct_to_target): Likewise.
22705 (remove_described_reference): Likewise.
22706 (ipa_impossible_devirt_target): Likewise.
22707 (propagate_controlled_uses): Likewise.
22708 (ipa_print_node_params): Likewise.
22709 (ipcp_transform_function): Likewise.
22710 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22711 (propagate_pure_const): Likewise.
22712 * ipa-reference.c (generate_summary): Likewise.
22713 (read_write_all_from_decl): Likewise.
22714 (propagate): Likewise.
22715 (ipa_reference_read_optimization_summary): Likewise.
22716 * ipa-utils.c (ipa_merge_profiles): Likewise.
22717 * ipa.c (walk_polymorphic_call_targets): Likewise.
22718 (symbol_table::remove_unreachable_nodes): Likewise.
22719 (ipa_single_use): Likewise.
22720 * passes.c (execute_todo): Likewise.
22721 * predict.c (drop_profile): Likewise.
22722 * symtab.c (symtab_node::get_dump_name): New function.
22723 (symtab_node::dump_name): Likewise.
22724 (symtab_node::dump_asm_name): Likewise.
22725 (symtab_node::dump_references): Likewise.
22726 (symtab_node::dump_referring): Likewise.
22727 (symtab_node::dump_base): Likewise.
22728 (symtab_node::debug_symtab): Likewise.
22729 * tree-sra.c (convert_callers_for_node): Likewise.
22730 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22731 * value-prof.c (init_node_map): Likewise.
22733 2017-05-23 Martin Liska <mliska@suse.cz>
22735 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22736 and symtab_node::debug_symtab to symbol_table::debug.
22737 * cgraphunit.c (analyze_functions): Use the renamed function.
22738 (symbol_table::compile): Likewise.
22739 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22740 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22741 * passes.c (execute_todo): Likewise.
22742 * symtab.c (symbol_table::dump): New function.
22743 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22745 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22747 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22748 that nonconst implies exec.
22750 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22752 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22753 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22754 (inline_edge_summary_vec): Turn into ...
22755 (ipa_call_summaries): ... this one.
22756 (redirect_to_unreachable, edge_set_predicate,
22757 evaluate_properties_for_edge, inline_summary_alloc,
22758 reset_ipa_call_summary, reset_inline_summary,
22759 inline_summary_t::duplicate): Update.
22760 (inline_edge_duplication_hook): Turn to ...
22761 (ipa_call_summary_t::duplicate): ... this one.
22762 (inline_edge_removal_hook): Turn to ...
22763 (ipa_call_summary_t::remove): ... this one.
22764 (dump_inline_edge_summary): Turn to ...
22765 (dump_ipa_call_summary): ... this one.
22766 (estimate_function_body_sizes): Update.
22767 (inline_update_callee_summaries): Update.
22768 (remap_edge_change_prob): Update.
22769 (remap_edge_summaries): Update.
22770 (inline_merge_summary): Update.
22771 (do_estimate_edge_time): Update.
22772 (inline_generate_summary): Update.
22773 (inline_read_section): Update.
22774 (inline_read_summary): Update.
22775 (inline_free_summary): Update.
22776 * ipa-inline.c (can_inline_edge_p): Update.
22777 (compute_inlined_call_time): Update.
22778 (want_inline_small_function_p): Update.
22779 (edge_badness): Update.
22780 (early_inliner): Update.
22781 * ipa-inline.h (inline_edge_summary): Turn to ...
22782 (ipa_call_summary): ... this one.
22783 (ipa_call_summary_t): New class.
22784 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22785 (ipa_call_summaries): New.
22786 (inline_edge_summary): Remove.
22787 (estimate_edge_growth): Update.
22788 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22789 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22790 * ipa-split.c (execute_split_functions): Update.
22791 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22793 2017-05-23 Tom de Vries <tom@codesourcery.com>
22795 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22796 attributes): Document rdrand effective target.
22798 2017-05-23 Tom de Vries <tom@codesourcery.com>
22800 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22801 attributes): Sort alphabetically.
22803 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22805 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22807 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22810 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22811 V2DF/V2DI splat into two separate patterns, one that handles
22812 registers, and the other that only handles memory. Drop support
22813 for splatting from a GPR on ISA 2.07 and then splitting the
22814 splat into direct move and splat.
22815 (vsx_splat_<mode>_reg): Likewise.
22816 (vsx_splat_<mode>_mem): Likewise.
22818 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22820 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22822 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22824 PR middle-end/80809
22825 * omp-low.c (finish_taskreg_remap): New function.
22826 (finish_taskreg_scan): If unit size of ctx->record_type
22827 is non-constant, unshare the size expression and replace
22828 decls in it with possible outer var refs.
22830 PR middle-end/80809
22831 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22832 GOVD_SHARED rather than GOVD_PRIVATE with it.
22833 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22834 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22836 PR middle-end/80853
22837 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22838 as last argument to build_outer_var_ref for pointer bases of array
22839 section reductions.
22841 2017-05-19 Martin Sebor <msebor@redhat.com>
22843 * print-tree.c (print_node): Print DECL_READ_P flag.
22845 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22847 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22848 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22849 * cgraph.c: Likewise.
22850 * cgraphunit.c: Likewise.
22851 * gengtype.c: Likewise.
22852 * ipa-cp.c: Likewise.
22853 * ipa-devirt.c: Likewise.
22854 * ipa-icf.c: Likewise.
22855 * ipa-predicate.c: Likewise.
22856 * ipa-profile.c: Likewise.
22857 * ipa-prop.c: Likewise.
22858 * ipa-split.c: Likewise.
22860 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22861 edge_predicate_pool, dump_inline_hints,
22862 inline_summary::account_size_time, redirect_to_unreachable,
22863 edge_set_predicate, set_hint_predicate,
22864 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22865 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22866 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22867 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22868 ipa_call_summary_t::remove, initialize_growth_caches,
22869 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22870 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22871 mark_modified, unmodified_parm_1, unmodified_parm,
22872 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22873 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22874 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22875 will_be_nonconstant_predicate, record_modified_bb_info,
22876 get_minimal_bb, record_modified, param_change_prob,
22877 phi_result_unknown_predicate, predicate_for_phi_result,
22878 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22879 estimate_function_body_sizes, compute_inline_parameters,
22880 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22881 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22882 inline_update_callee_summaries, remap_edge_change_prob,
22883 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22884 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22885 inline_analyze_function, inline_summary_t::insert,
22886 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22887 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22888 inline_free_summary): Move to ipa-fnsummary.h
22889 (predicate_t): Remove.
22890 * ipa-fnsummary.c: New file.
22891 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22892 (enum inline_hints_vals, inline_hints, agg_position_info,
22893 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22894 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22895 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22896 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22897 inline_read_summary, inline_write_summary, inline_free_summary,
22898 inline_analyze_function, initialize_inline_failed,
22899 inline_merge_summary, inline_update_overall_summary,
22900 compute_inline_parameters): Move to ipa-fnsummary.h
22901 * ipa-fnsummary.h: New file.
22902 * ipa-inline-transform.h: Include ipa-inline.h.
22903 * ipa-inline.c: LIkewise.
22905 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22907 * ipa-inline.c (edge_badness): Use inlined_time instead of
22908 inline_summaries->get.
22910 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22912 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22914 2017-05-22 Nathan Sidwell <nathan@acm.org>
22916 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22917 (fdump-lang): Document 'raw' option.
22918 * dumpfile.h (TDI_tu): Delete.
22919 * dumpfile.c (dump_files): Remove translation-unit.
22920 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22922 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22924 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22925 command option from $(AWK) call.
22926 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22927 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22928 [FORMAT]: Remove handling of variable.
22929 * config/avr/t-multilib: Regenerate.
22931 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22933 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22935 (dump_inline_summary): Do not print self_time.
22936 (estimate_function_body_sizes): Do not set self_time.
22937 (compute_inline_parameters): Likewise.
22938 (inline_read_section, inline_write_summary): Do not stream self_time.
22939 * ipa-inline.h (inline_summary): Drop self_time.
22941 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22943 * ipa-inline-analysis.c (account_size_time): Rename to ...
22944 (inline_summary::account_size_time): ... this one.
22945 (reset_ipa_call_summary): Turn to ...
22946 (ipa_call_summary::reset): ... this one.
22947 (reset_inline_summary): Turn to ...
22948 (inline_summary::reset): ... this one.
22949 (inline_summary_t::remove): Update.
22950 (inline_summary_t::duplicate): Update.
22951 (ipa_call_summary_t::remove): Update.
22952 (dump_inline_summary): Update.
22953 (estimate_function_body_sizes): Update.
22954 (compute_inline_parameters): Update.
22955 (estimate_node_size_and_time): Update.
22956 (inline_merge_summary): Update.
22957 (inline_update_overall_summary): Update.
22958 (inline_read_section): Update.
22959 (inline_write_summary): Update.
22960 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22961 add account_size_time and reset member functions.
22962 (ipa_call_summary): Add reset function.
22963 * ipa-predicate.h (predicate::operator &): Constify.
22965 2017-05-22 Richard Biener <rguenther@suse.de>
22967 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22969 2017-05-19 Jason Merrill <jason@redhat.com>
22971 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22973 2017-05-19 Marek Polacek <polacek@redhat.com>
22976 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22977 TYPE_OVERFLOW_WRAPS checks.
22979 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22981 * tree-core.h (enum omp_clause_default_kind): Add
22982 "OMP_CLAUSE_DEFAULT_PRESENT".
22983 * tree-pretty-print.c (dump_omp_clause): Handle it.
22984 * gimplify.c (enum gimplify_omp_var_data): Add
22985 "GOVD_MAP_FORCE_PRESENT".
22986 (gimplify_adjust_omp_clauses_1): Map it to
22987 "GOMP_MAP_FORCE_PRESENT".
22988 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22990 * gimplify.c (oacc_default_clause): Clarify.
22992 2017-05-19 Nathan Sidwell <nathan@acm.org>
22994 LANG_HOOK_REGISTER_DUMPS
22995 * toplev.c (general_init): Call register dump lang hook.
22996 * doc/invoke.texi: Document -fdump-lang option family.
22997 * dumpfile.c (dump_files): Remove class dump here.
22998 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22999 * dumpfile.h (tree_dump_index): Remove TDI_class.
23000 * langhooks-def.h (lhd_register_dumps): Declare.
23001 (LANG_HOOKS_REGISTER_DUMPS): Define.
23002 (LANG_HOOKS_INITIALIZER): Add it.
23003 * langhooks.c (lhd_register_dumps): Define.
23004 * langhooks.h (struct lang_hooks): Add register_dumps.
23006 2017-05-19 Nathan Sidwell <nathan@acm.org>
23008 * context.h (context::set_passes): New.
23009 * context.c (context::context): Do not create pass manager.
23010 * toplev.c (general_init): Create pass manager here.
23012 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23014 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23015 use this splitter if two add or or instructions would also work for
23016 the constant we want to generate.
23018 2017-05-19 Richard Biener <rguenther@suse.de>
23021 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23022 predicate evaluation.
23024 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23026 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23028 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23031 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23033 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23034 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23035 (inline_edge_summary_vec): Turn into ...
23036 (ipa_call_summaries): ... this one.
23037 (redirect_to_unreachable, edge_set_predicate,
23038 evaluate_properties_for_edge, inline_summary_alloc,
23039 reset_ipa_call_summary, reset_inline_summary,
23040 inline_summary_t::duplicate): Update.
23041 (inline_edge_duplication_hook): Turn to ...
23042 (ipa_call_summary_t::duplicate): ... this one.
23043 (inline_edge_removal_hook): Turn to ...
23044 (ipa_call_summary_t::remove): ... this one.
23045 (dump_inline_edge_summary): Turn to ...
23046 (dump_ipa_call_summary): ... this one.
23047 (estimate_function_body_sizes): Update.
23048 (inline_update_callee_summaries): Update.
23049 (remap_edge_change_prob): Update.
23050 (remap_edge_summaries): Update.
23051 (inline_merge_summary): Update.
23052 (do_estimate_edge_time): Update.
23053 (inline_generate_summary): Update.
23054 (inline_read_section): Update.
23055 (inline_read_summary): Update.
23056 (inline_free_summary): Update.
23057 * ipa-inline.c (can_inline_edge_p): Update.
23058 (compute_inlined_call_time): Update.
23059 (want_inline_small_function_p): Update.
23060 (edge_badness): Update.
23061 (early_inliner): Update.
23062 * ipa-inline.h (inline_edge_summary): Turn to ...
23063 (ipa_call_summary): ... this one.
23064 (ipa_call_summary_t): New class.
23065 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23066 (ipa_call_summaries): New.
23067 (inline_edge_summary): Remove.
23068 (estimate_edge_growth): Update.
23069 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23070 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23071 * ipa-split.c (execute_split_functions): Update.
23072 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23074 2017-05-19 Richard Biener <rguenther@suse.de>
23076 PR middle-end/80764
23077 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23079 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23081 * config/rs6000/rs6000.c (struct machine_function): Add field
23082 fpr_is_wrapped_separately.
23083 (rs6000_get_separate_components): Use 64 components. Handle the
23084 new FPR components.
23085 (rs6000_components_for_bb): Handle the FPR components.
23086 (rs6000_emit_prologue_components): Handle the FPR components.
23087 (rs6000_emit_epilogue_components): Handle the FPR components.
23088 (rs6000_set_handled_components): Handle the FPR components.
23089 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23090 that are already separately shrink-wrapped.
23091 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23092 that are already separately shrink-wrapped.
23094 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23097 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23100 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23101 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23102 cases where the register allocator uses a move and an offsettable
23103 memory operation to/from a FPR register on ISA 2.06/2.07.
23104 (define_peephole2 for Altivec d-form store): Likewise.
23106 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23109 * config/i386/mmx.md (*mov<mode>_internal): Enable
23110 alternatives 11, 12, 13 and 14 also for 32bit targets.
23111 Remove alternatives 15, 16, 17 and 18.
23112 * config/i386/sse.md (vec_concatv2di): Change
23113 alternative (!x, *y) to (x, ?!*Yn).
23115 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23117 * dumpfile.h (enum dump_kind): Remove stray comma.
23119 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23121 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23122 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23123 predicate::num_conditions
23124 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23125 (CHANGED): turn into predicate::changed.
23126 (agg_position_info): Move to ipa-predicate.h
23127 (add_condition, predicate::add_clause, predicate::operator &=,
23128 predicate::or_with, predicate::evaluate, predicate::probability,
23129 dump_condition, dump_clause, predicate::dump,
23130 predicate::remap_after_duplication, predicate::remap_after_inlining,
23131 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23132 (evaluate_conditions_for_known_args): Update.
23133 (set_cond_stmt_execution_predicate): Update.
23134 * ipa-inline.h: Include ipa-predicate.h
23135 (condition, inline_param_summary, conditions, agg_position_info,
23136 predicate): Move to ipa-predicate.h
23137 * ipa-predicate.c: New file.
23138 * ipa-predicate.h: New file.
23140 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23142 * final.c (leaf_function_p): Check we are not in a sequence.
23144 2017-05-18 Martin Liska <mliska@suse.cz>
23146 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23147 * dumpfile.c (dump_register): Use new enum dump_kind.
23148 (get_dump_file_name): Likewise.
23149 (dump_enable_all): Likewise.
23150 (dump_switch_p_1): Likewise.
23151 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23152 * dumpfile.h (enum dump_kind): New enum type.
23153 (struct dump_file_info): Create constructor and
23154 format fields and comments.
23155 * passes.c (pass_manager::register_one_dump_file):
23157 * statistics.c (statistics_early_init): Likewise.
23158 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23159 TDF_TREE with TDF_SLIM.
23160 (gather_memory_references_ref): Likewise.
23162 2017-05-18 Martin Liska <mliska@suse.cz>
23164 * vec.h (struct vnull): Use it.
23166 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23168 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23169 (true_predicate, false_predicate, true_predicate_p,
23170 false_predicate_p): Remove.
23171 (single_cond_predicate, not_inlined_predicate): Turn to member function
23173 (add_condition): Update.
23174 (add_clause): Turn to...
23175 (predicate::add_clause): ... this one; update; allow passing NULL
23177 (and_predicates): Turn to ...
23178 (predicate::operator &=): ... this one.
23179 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23180 (or_predicates): Turn to ...
23181 (predicate::or_with): ... this one.
23182 (evaluate_predicate): Turn to ...
23183 (predicate::evaluate): ... this one.
23184 (predicate_probability): Turn to ...
23185 (predicate::probability): ... this one.
23186 (dump_condition): Update.
23187 (dump_predicate): Turn to ...
23188 (predicate::dump): ... this one.
23189 (account_size_time): Update.
23190 (edge_set_predicate): Update.
23191 (set_hint_predicate): UPdate.
23192 (evaluate_conditions_for_known_args): Update.
23193 (evaluate_properties_for_edge): Update.
23194 (remap_predicate_after_duplication): Turn to...
23195 (predicate::remap_after_duplication): ... this one.
23196 (remap_hint_predicate_after_duplication): Update.
23197 (inline_summary_t::duplicate): UPdate.
23198 (dump_inline_edge_summary): Update.
23199 (dump_inline_summary): Update.
23200 (set_cond_stmt_execution_predicate): Update.
23201 (set_switch_stmt_execution_predicate): Update.
23202 (compute_bb_predicates): Update.
23203 (will_be_nonconstant_expr_predicate): Update.
23204 (will_be_nonconstant_predicate): Update.
23205 (phi_result_unknown_predicate): Update.
23206 (predicate_for_phi_result): Update.
23207 (array_index_predicate): Update.
23208 (estimate_function_body_sizes): Update.
23209 (estimate_node_size_and_time): Update.
23210 (estimate_ipcp_clone_size_and_time): Update.
23211 (remap_predicate): Rename to ...
23212 (predicate::remap_after_inlining): ... this one.
23213 (remap_hint_predicate): Update.
23214 (inline_merge_summary): Update.
23215 (inline_update_overall_summary): Update.
23216 (estimate_size_after_inlining): Update.
23217 (read_predicate): Rename to ...
23218 (predicate::stream_in): ... this one.
23219 (read_inline_edge_summary): Update.
23220 (write_predicate): Rename to ...
23221 (predicate::stream_out): ... this one.
23222 (write_inline_edge_summary): Update.
23223 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23224 (clause_t): Turn to uint32_t
23225 (predicate): Turn to class; implement constructor and operators
23227 (size_time_entry): Update.
23228 (inline_summary): Update.
23229 (inline_edge_summary): Update.
23231 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23233 * fold-const.c (fold_binary_loc): Move transformation...
23234 * match.pd (C - X CMP X): ... here.
23236 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23238 * config/sparc/sparc.c (sparc_option_override): Set function
23239 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23240 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23242 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23244 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23246 2017-05-18 Marek Polacek <polacek@redhat.com>
23249 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23250 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23251 gimple_assign_load_p.
23253 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23255 PR middle-end/80692
23256 * real.c (do_compare): Give decimal_do_compare preference over
23257 comparing just the signs.
23259 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23261 * doc/md.texi (Canonicalization of Instructions): Describe the
23262 canonical form of instructions that inherently set a condition
23265 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23267 PR middle-end/80775
23268 * tree-cfg.c: Move deletion of unreachable case statements to after
23269 the merging of consecutive case labels.
23271 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23273 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23274 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23275 restoring of callee-saved registers.
23277 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23279 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23280 * config/visium/visium.c (single_set_and_flags): Likewise.
23281 * config/visium/visium.md (Substitutions): Likewise.
23283 2017-05-17 Martin Liska <mliska@suse.cz>
23285 * cfg.c: Introduce dump_flags_t type and
23286 use it instead of int type.
23288 * cfghooks.c: Likewise.
23289 * cfghooks.h (struct cfg_hooks): Likewise.
23290 * cfgrtl.c: Likewise.
23291 * cfgrtl.h: Likewise.
23292 * cgraph.c (cgraph_node::get_body): Likewise.
23293 * coretypes.h: Likewise.
23294 * domwalk.c: Likewise.
23295 * domwalk.h: Likewise.
23296 * dumpfile.c (struct dump_option_value_info): Likewise.
23297 (dump_enable_all): Likewise.
23298 (dump_switch_p_1): Likewise.
23299 (opt_info_switch_p): Likewise.
23300 * dumpfile.h (enum tree_dump_index): Likewise.
23301 (struct dump_file_info): Likewise.
23302 * genemit.c: Likewise.
23303 * generic-match-head.c: Likewise.
23304 * gengtype.c (open_base_files): Likewise.
23305 * gimple-pretty-print.c: Likewise.
23306 * gimple-pretty-print.h: Likewise.
23307 * graph.c (print_graph_cfg): Likewise.
23308 * graphite-scop-detection.c (dot_all_sese): Likewise.
23309 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23310 * loop-unroll.c (report_unroll): Likewise.
23311 * passes.c (pass_manager::register_one_dump_file): Likewise.
23312 * print-tree.c: Likewise.
23313 * statistics.c: Likewise.
23314 * tree-cfg.c: Likewise.
23315 * tree-cfg.h: Likewise.
23316 * tree-dfa.c: Likewise.
23317 * tree-dfa.h: Likewise.
23318 * tree-dump.c (dump_function): Likewise.
23319 * tree-dump.h (struct dump_info): Likewise.
23320 * tree-pretty-print.c: Likewise.
23321 * tree-pretty-print.h: Likewise.
23322 * tree-ssa-live.c: Likewise.
23323 * tree-ssa-live.h: Likewise.
23324 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23325 * tree-vect-loop.c: Likewise.
23326 * tree-vect-slp.c: Likewise.
23328 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23329 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23331 PR tree-optimization/80457
23332 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23333 of all arguments to a statement as scalar_to_vec operations.
23334 (vectorizable_call): Adjust call to vect_model_simple_cost for
23336 (vectorizable_conversion): Likewise.
23337 (vectorizable_assignment): Likewise.
23338 (vectorizable_shift): Likewise.
23339 (vectorizable_operation): Likewise.
23340 (vectorizable_comparison): Likewise.
23341 (vect_is_simple_cond): Record the def types for operands.
23342 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23343 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23344 for statement argument count.
23346 2017-05-16 Carl Love <cel@us.ibm.com>
23348 * config/rs6000/rs6000-c: Add support for built-in functions
23349 vector unsigned long long vec_bperm (vector unsigned long long,
23350 vector unsigned char)
23351 vector signed long long vec_mule (vector signed int,
23353 vector unsigned long long vec_mule (vector unsigned int,
23354 vector unsigned int)
23355 vector signed long long vec_mulo (vector signed int,
23357 vector unsigned long long vec_mulo (vector unsigned int,
23358 vector unsigned int)
23359 vector signed char vec_sldw (vector signed char,
23360 vector signed char,
23362 vector unsigned char vec_sldw (vector unsigned char,
23363 vector unsigned char,
23365 vector signed short vec_sldw (vector signed short,
23366 vector signed short,
23368 vector unsigned short vec_sldw (vector unsigned short,
23369 vector unsigned short,
23371 vector signed int vec_sldw (vector signed int,
23374 vector unsigned int vec_sldw (vector unsigned int,
23375 vector unsigned int,
23377 vector signed long long vec_sldw (vector signed long long,
23378 vector signed long long,
23380 vector unsigned long long vec_sldw (vector unsigned long long,
23381 vector unsigned long long,
23383 * config/rs6000/rs6000-c: Add support for built-in functions
23384 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23385 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23386 * doc/extend.texi: Update the built-in documentation for the
23387 new built-in functions.
23389 2017-05-16 Marek Polacek <polacek@redhat.com>
23393 * tree.c (save_expr): Don't fold the expression.
23395 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23397 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23398 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23399 and (?*y,m). Update insn attributes.
23401 2017-05-16 Martin Liska <mliska@suse.cz>
23403 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23404 flags argument of print_gimple_stmt, print_gimple_expr,
23405 print_generic_stmt and print_generic_expr.
23406 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23407 * coretypes.h: Likewise.
23408 * except.c (dump_eh_tree): Likewise.
23409 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23410 * gimple-pretty-print.h: Likewise.
23411 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23412 (backprop::push_to_worklist): Likewise.
23413 (backprop::pop_from_worklist): Likewise.
23414 (backprop::process_use): Likewise.
23415 (backprop::intersect_uses): Likewise.
23416 (note_replacement): Likewise.
23417 * gimple-ssa-store-merging.c
23418 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23419 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23420 (pass_store_merging::execute): Likewise.
23421 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23422 (ssa_base_cand_dump_callback): Likewise.
23423 (dump_incr_vec): Likewise.
23424 (replace_refs): Likewise.
23425 (replace_mult_candidate): Likewise.
23426 (create_add_on_incoming_edge): Likewise.
23427 (create_phi_basis): Likewise.
23428 (insert_initializers): Likewise.
23429 (all_phi_incrs_profitable): Likewise.
23430 (introduce_cast_before_cand): Likewise.
23431 (replace_one_candidate): Likewise.
23432 * gimplify.c (gimplify_expr): Likewise.
23433 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23434 (set_rename): Likewise.
23435 (rename_uses): Likewise.
23436 (copy_loop_phi_nodes): Likewise.
23437 (add_close_phis_to_merge_points): Likewise.
23438 (copy_loop_close_phi_args): Likewise.
23439 (copy_cond_phi_args): Likewise.
23440 (graphite_copy_stmts_from_block): Likewise.
23441 (translate_pending_phi_nodes): Likewise.
23442 * graphite-poly.c (print_pdr): Likewise.
23443 (dump_gbb_cases): Likewise.
23444 (dump_gbb_conditions): Likewise.
23445 (print_scop_params): Likewise.
23446 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23447 (build_cross_bb_scalars_use): Likewise.
23448 (gather_bbs::before_dom_children): Likewise.
23449 * hsa-dump.c (dump_hsa_immed): Likewise.
23450 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23451 (get_replacement_map): Likewise.
23452 * ipa-inline-analysis.c (dump_condition): Likewise.
23453 (estimate_function_body_sizes): Likewise.
23454 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23455 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23456 * ipa-prop.c (ipa_dump_param): Likewise.
23457 (ipa_print_node_jump_functions_for_edge): Likewise.
23458 (ipa_modify_call_arguments): Likewise.
23459 (ipa_modify_expr): Likewise.
23460 (ipa_dump_param_adjustments): Likewise.
23461 (ipa_dump_agg_replacement_values): Likewise.
23462 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23463 * ipa-pure-const.c (check_stmt): Likewise.
23464 (pass_nothrow::execute): Likewise.
23465 * ipa-split.c (execute_split_functions): Likewise.
23466 * omp-offload.c (dump_oacc_loop_part): Likewise.
23467 (dump_oacc_loop): Likewise.
23468 * trans-mem.c (tm_log_emit): Likewise.
23469 (tm_memopt_accumulate_memops): Likewise.
23470 (dump_tm_memopt_set): Likewise.
23471 (dump_tm_memopt_transform): Likewise.
23472 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23473 (print_loop): Likewise.
23474 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23475 (chkp_gather_checks_info): Likewise.
23476 (chkp_get_check_result): Likewise.
23477 (chkp_remove_check_if_pass): Likewise.
23478 (chkp_use_outer_bounds_if_possible): Likewise.
23479 (chkp_reduce_bounds_lifetime): Likewise.
23480 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23481 (chkp_mark_completed_bounds): Likewise.
23482 (chkp_register_incomplete_bounds): Likewise.
23483 (chkp_mark_invalid_bounds): Likewise.
23484 (chkp_maybe_copy_and_register_bounds): Likewise.
23485 (chkp_build_returned_bound): Likewise.
23486 (chkp_get_bound_for_parm): Likewise.
23487 (chkp_build_bndldx): Likewise.
23488 (chkp_get_bounds_by_definition): Likewise.
23489 (chkp_generate_extern_var_bounds): Likewise.
23490 (chkp_get_bounds_for_decl_addr): Likewise.
23491 * tree-chrec.c (chrec_apply): Likewise.
23492 * tree-data-ref.c (dump_data_reference): Likewise.
23493 (dump_subscript): Likewise.
23494 (dump_data_dependence_relation): Likewise.
23495 (analyze_overlapping_iterations): Likewise.
23496 * tree-inline.c (expand_call_inline): Likewise.
23497 (tree_function_versioning): Likewise.
23498 * tree-into-ssa.c (dump_defs_stack): Likewise.
23499 (dump_currdefs): Likewise.
23500 (dump_names_replaced_by): Likewise.
23501 (dump_update_ssa): Likewise.
23502 (update_ssa): Likewise.
23503 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23504 * tree-parloops.c (build_new_reduction): Likewise.
23505 (try_create_reduction_list): Likewise.
23506 (ref_conflicts_with_region): Likewise.
23507 (oacc_entry_exit_ok_1): Likewise.
23508 (oacc_entry_exit_single_gang): Likewise.
23509 * tree-pretty-print.h: Likewise.
23510 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23511 (get_scalar_evolution): Likewise.
23512 (add_to_evolution): Likewise.
23513 (get_loop_exit_condition): Likewise.
23514 (analyze_evolution_in_loop): Likewise.
23515 (analyze_initial_condition): Likewise.
23516 (analyze_scalar_evolution): Likewise.
23517 (instantiate_scev): Likewise.
23518 (number_of_latch_executions): Likewise.
23519 (gather_chrec_stats): Likewise.
23520 (final_value_replacement_loop): Likewise.
23521 (scev_const_prop): Likewise.
23522 * tree-sra.c (dump_access): Likewise.
23523 (disqualify_candidate): Likewise.
23524 (create_access): Likewise.
23525 (reject): Likewise.
23526 (maybe_add_sra_candidate): Likewise.
23527 (create_access_replacement): Likewise.
23528 (analyze_access_subtree): Likewise.
23529 (analyze_all_variable_accesses): Likewise.
23530 (sra_modify_assign): Likewise.
23531 (initialize_constant_pool_replacements): Likewise.
23532 (find_param_candidates): Likewise.
23533 (decide_one_param_reduction): Likewise.
23534 (replace_removed_params_ssa_names): Likewise.
23535 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23536 * tree-ssa-copy.c (dump_copy_of): Likewise.
23537 (copy_prop_visit_cond_stmt): Likewise.
23538 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23539 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23540 (record_equivalences_from_stmt): Likewise.
23541 * tree-ssa-dse.c (compute_trims): Likewise.
23542 (delete_dead_call): Likewise.
23543 (delete_dead_assignment): Likewise.
23544 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23545 (forward_propagate_into_cond): Likewise.
23546 (pass_forwprop::execute): Likewise.
23547 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23548 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23550 (move_computations_worker): Likewise.
23551 (execute_sm): Likewise.
23552 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23553 (remove_exits_and_undefined_stmts): Likewise.
23554 (remove_redundant_iv_tests): Likewise.
23555 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23556 (adjust_iv_update_pos): Likewise.
23557 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23558 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23559 (value_replacement): Likewise.
23560 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23561 * tree-ssa-pre.c (print_pre_expr): Likewise.
23562 (get_representative_for): Likewise.
23563 (create_expression_by_pieces): Likewise.
23564 (insert_into_preds_of_block): Likewise.
23565 (eliminate_insert): Likewise.
23566 (eliminate_dom_walker::before_dom_children): Likewise.
23567 (eliminate): Likewise.
23568 (remove_dead_inserted_code): Likewise.
23569 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23570 * tree-ssa-reassoc.c (get_rank): Likewise.
23571 (eliminate_duplicate_pair): Likewise.
23572 (eliminate_plus_minus_pair): Likewise.
23573 (eliminate_not_pairs): Likewise.
23574 (undistribute_ops_list): Likewise.
23575 (eliminate_redundant_comparison): Likewise.
23576 (update_range_test): Likewise.
23577 (optimize_range_tests_var_bound): Likewise.
23578 (optimize_vec_cond_expr): Likewise.
23579 (rewrite_expr_tree): Likewise.
23580 (rewrite_expr_tree_parallel): Likewise.
23581 (linearize_expr): Likewise.
23582 (break_up_subtract): Likewise.
23583 (linearize_expr_tree): Likewise.
23584 (attempt_builtin_powi): Likewise.
23585 (attempt_builtin_copysign): Likewise.
23586 (transform_stmt_to_copy): Likewise.
23587 (transform_stmt_to_multiply): Likewise.
23588 (dump_ops_vector): Likewise.
23589 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23590 (print_scc): Likewise.
23591 (set_ssa_val_to): Likewise.
23592 (visit_reference_op_store): Likewise.
23593 (visit_use): Likewise.
23594 (sccvn_dom_walker::before_dom_children): Likewise.
23595 (run_scc_vn): Likewise.
23596 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23598 (expr_hash_elt::print): Likewise.
23599 (const_and_copies::pop_to_marker): Likewise.
23600 (const_and_copies::record_const_or_copy_raw): Likewise.
23601 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23602 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23603 (dump_predicates): Likewise.
23604 (find_uninit_use): Likewise.
23605 (warn_uninitialized_phi): Likewise.
23606 (pass_late_warn_uninitialized::execute): Likewise.
23607 * tree-ssa.c (verify_vssa): Likewise.
23608 (verify_ssa): Likewise.
23609 (maybe_optimize_var): Likewise.
23610 * tree-vrp.c (dump_value_range): Likewise.
23611 (dump_all_value_ranges): Likewise.
23612 (dump_asserts_for): Likewise.
23613 (register_edge_assert_for_2): Likewise.
23614 (vrp_visit_cond_stmt): Likewise.
23615 (vrp_visit_switch_stmt): Likewise.
23616 (vrp_visit_stmt): Likewise.
23617 (vrp_visit_phi_node): Likewise.
23618 (simplify_cond_using_ranges_1): Likewise.
23619 (fold_predicate_in): Likewise.
23620 (evrp_dom_walker::before_dom_children): Likewise.
23621 (evrp_dom_walker::push_value_range): Likewise.
23622 (evrp_dom_walker::pop_value_range): Likewise.
23623 (execute_early_vrp): Likewise.
23625 2017-05-16 Richard Biener <rguenther@suse.de>
23627 * dwarf2out.c (loc_list_from_tree_1): Do not create
23628 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23630 2017-05-16 Richard Biener <rguenther@suse.de>
23632 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23634 (note_variable_value_in_expr): If we resolved the decl ref
23635 do not push to the stack.
23637 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23639 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23640 operations in fast-math mode.
23641 (vaddq_f16): Likewise.
23642 (vmul_f16): Likewise.
23643 (vmulq_f16): Likewise.
23644 (vsub_f16): Likewise.
23645 (vsubq_f16): Likewise.
23646 * config/arm/neon.md (add<mode>3): New.
23648 (fma:<VH:mode>3): New. Also remove outdated comment.
23651 2017-05-16 Martin Liska <mliska@suse.cz>
23655 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23656 (odr_types_equivalent_p): Likewise.
23658 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23660 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23662 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23665 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23666 non-interunit SSE move alternatives with '?'.
23667 (zero-extendsidi peephole2): New peephole to skip intermediate
23668 general register in SSE zero-extend sequence.
23670 2017-05-15 Jeff Law <law@redhat.com>
23672 * reorg.c (relax_delay_slots): Create a new variable to hold
23673 the temporary target rather than clobbering TARGET_LABEL.
23675 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23676 missing argument to extract_bit_field call.
23677 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23679 2017-05-15 Martin Liska <mliska@suse.cz>
23682 * gcc.c (process_command): Do not allow empty argument of -o option.
23684 2017-05-15 Renlin Li <renlin.li@arm.com>
23686 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23687 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23688 * config/aarch64/constraints.md (Usf): Add long call check.
23689 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23690 (call_value): Likewise.
23691 (sibcall): Likewise.
23692 (sibcall_value): Likewise.
23694 (call_value_insn): New.
23695 (sibcall_insn): Update rtx pattern.
23696 (sibcall_value_insn): Likewise.
23697 (call_internal): Remove.
23698 (call_value_internal): Likewise.
23699 (sibcall_internal): Likewise.
23700 (sibcall_value_internal): Likewise.
23701 (call_reg): Likewise.
23702 (call_symbol): Likewise.
23703 (call_value_reg): Likewise.
23704 (call_value_symbol): Likewise.
23706 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23709 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23711 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23713 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23714 compatible with CCGOCmode and with CCZmode.
23716 2017-05-14 Martin Sebor <msebor@redhat.com>
23718 PR middle-end/77671
23719 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23720 (gimple_fold_builtin_snprintf): Same.
23721 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23722 (gimple_fold_builtin_snprintf): Same.
23723 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23724 of character types.
23725 (is_call_safe): New function.
23726 (try_substitute_return_value): Call it.
23727 (try_simplify_call): New function.
23728 (pass_sprintf_length::handle_gimple_call): Call it.
23730 2017-05-14 Martin Sebor <msebor@redhat.com>
23732 PR middle-end/80669
23733 * builtins.c (expand_builtin_stpncpy): Simplify.
23735 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23737 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23738 * config/i386/i386.h
23739 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23740 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23741 (struct machine_function): Add new members call_ms2sysv,
23742 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23743 (struct machine_frame_state): New fields sp_realigned and
23744 sp_realigned_offset.
23745 * config/i386/i386.c
23746 (enum xlogue_stub): New enum.
23747 (enum xlogue_stub_sets): New enum.
23748 (class xlogue_layout): New class.
23749 (struct ix86_frame): New fields stack_realign_allocate_offset,
23750 stack_realign_offset and outlined_save_offset. Modify comments to
23751 detail stack layout when using out-of-line stubs.
23752 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23753 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23754 -mcall-ms2sysv-xlogues.
23755 (stub_managed_regs): New static variable.
23756 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23757 registers managed by out-of-line stub.
23758 (disable_call_ms2sysv_xlogues): New function.
23759 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23760 m->call_ms2sysv when appropriate and compute frame layout for
23762 (sp_valid_at, fp_valid_at): New inline functions.
23763 (choose_basereg): New function.
23764 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23766 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23767 Use align parameter of choose_baseaddr to generated aligned SSE movs
23769 (pro_epilogue_adjust_stack): Modify to track
23770 machine_frame_state::sp_realigned.
23771 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23772 (ix86_nsaved_sseregs): Likewise.
23773 (ix86_emit_save_regs): Likewise.
23774 (ix86_emit_save_regs_using_mov): Likewise.
23775 (ix86_emit_save_sse_regs_using_mov): Likewise.
23776 (get_scratch_register_on_entry): Likewise.
23777 (gen_frame_set): New function.
23778 (gen_frame_load): Likewise.
23779 (gen_frame_store): Likewise.
23780 (emit_outlined_ms2sysv_save): Likewise.
23781 (emit_outlined_ms2sysv_restore): Likewise.
23782 (ix86_expand_prologue): Modify stack re-alignment code and call
23783 emit_outlined_ms2sysv_save when appropriate.
23784 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23785 parameter rtx_insn *insn, which allows the function to be used to only
23786 generate the notes.
23787 (ix86_expand_epilogue): Modify validity checks of frame and stack
23788 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23789 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23790 * config/i386/predicates.md
23791 (save_multiple): New predicate.
23792 (restore_multiple): Likewise.
23793 * config/i386/sse.md
23794 (save_multiple<mode>): New pattern.
23795 (save_multiple_realign<mode>): Likewise.
23796 (restore_multiple<mode>): Likewise.
23797 (restore_multiple_and_return<mode>): Likewise.
23798 (restore_multiple_leave_return<mode>): Likewise.
23799 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23801 2017-05-14 Julia Koval <julia.koval@intel.com>
23803 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23804 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23805 (__builtin_ia32_xsetbv): New builtins.
23806 * config/i386/i386.c (ix86_expand_special_args_builtin):
23808 (ix86_expand_builtin): Special expand for new intrinsics.
23809 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23810 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23811 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23813 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23815 * cfganal.c (inverted_post_order_compute): Change argument type
23817 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23818 * df-core.c (rest_of_handle_df_initialize): Adjust.
23819 (rest_of_handle_df_finish): Likewise.
23820 (df_analyze_1): Likewise.
23821 (df_analyze): Likewise.
23822 (loop_inverted_post_order_compute): Change argument to be a vec *.
23823 (df_analyze_loop): Adjust.
23824 (df_get_n_blocks): Likewise.
23825 (df_get_postorder): Likewise.
23826 * df.h (struct df_d): Change field to be a vec.
23827 * lcm.c (compute_laterin): Adjust.
23828 (compute_available): Likewise.
23829 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23830 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23831 * tree-ssa-pre.c (compute_antic): Likewise.
23833 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23835 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23836 (depth_first_search::depth_first_search): Change structure init
23837 function to this constructor.
23838 (depth_first_search::add_bb): Rename function to this member.
23839 (depth_first_search::execute): Likewise.
23840 (flow_dfs_compute_reverse_finish): Adjust.
23842 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23844 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23845 (longest_simple_path): Likewise.
23846 * shrink-wrap.c (spread_components): Likewise.
23847 (disqualify_problematic_components): Likewise.
23848 (emit_common_heads_for_components): Likewise.
23849 (emit_common_tails_for_components): Likewise.
23850 (insert_prologue_epilogue_for_components): Likewise.
23852 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23854 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23857 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23859 * df-core.c (df_set_blocks): Start using auto_bitmap.
23860 (df_compact_blocks): Likewise.
23861 * df-problems.c (df_rd_confluence_n): Likewise.
23862 * df-scan.c (df_insn_rescan_all): Likewise.
23863 (df_process_deferred_rescans): Likewise.
23864 (df_update_entry_block_defs): Likewise.
23865 (df_update_exit_block_uses): Likewise.
23866 (df_entry_block_bitmap_verify): Likewise.
23867 (df_exit_block_bitmap_verify): Likewise.
23868 (df_scan_verify): Likewise.
23869 * lra-constraints.c (lra_constraints): Likewise.
23870 (undo_optional_reloads): Likewise.
23871 (lra_undo_inheritance): Likewise.
23872 * lra-remat.c (calculate_gen_cands): Likewise.
23873 (do_remat): Likewise.
23874 * lra-spills.c (assign_spill_hard_regs): Likewise.
23875 (spill_pseudos): Likewise.
23876 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23877 (bitmap_set_subtract_values): Likewise.
23879 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23881 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23882 management with auto_bitmap.
23883 (fix_inter_tick): Likewise.
23884 (fix_recovery_deps): Likewise.
23885 * ira.c (add_store_equivs): Likewise.
23886 (find_moveable_pseudos): Likewise.
23887 (split_live_ranges_for_shrink_wrap): Likewise.
23888 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23889 (rtx_reuse_manager::seen_def_p): Likewise.
23890 (rtx_reuse_manager::set_seen_def): Likewise.
23891 * print-rtl.h (class rtx_reuse_manager): Likewise.
23893 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23895 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23897 (migrate_btr_def): Likewise.
23898 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23899 * df-core.c (loop_post_order_compute): Likewise.
23900 (loop_inverted_post_order_compute): Likewise.
23901 * hsa-common.h: Likewise.
23902 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23903 * init-regs.c (initialize_uninitialized_regs): Likewise.
23904 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23905 (inline_small_functions): Likewise.
23906 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23907 * ira.c (combine_and_move_insns): Likewise.
23908 (build_insn_chain): Likewise.
23909 * loop-invariant.c (find_invariants): Likewise.
23910 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23911 * predict.c (tree_predict_by_opcode): Likewise.
23912 (predict_paths_leading_to): Likewise.
23913 (predict_paths_leading_to_edge): Likewise.
23914 (estimate_loops_at_level): Likewise.
23915 (estimate_loops): Likewise.
23916 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23917 (spread_components): Likewise.
23918 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23919 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23920 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23921 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23922 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23923 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23924 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23925 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23926 (mark_threaded_blocks): Likewise.
23927 (thread_through_all_blocks): Likewise.
23928 * tree-ssa.c (verify_ssa): Likewise.
23929 (execute_update_addresses_taken): Likewise.
23930 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23932 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23934 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23936 (post_order_compute): Likewise.
23937 (inverted_post_order_compute): Likewise.
23938 (pre_and_rev_post_order_compute_fn): Likewise.
23940 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23942 * genrecog.c (int_set::int_set): Explicitly construct our
23943 auto_vec base class.
23944 * vec.h (auto_vec::auto_vec): New constructor.
23946 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23948 * bitmap.h (class auto_bitmap): New constructor taking
23949 bitmap_obstack * argument.
23951 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23953 * bitmap.h (class auto_bitmap): Change type of m_bits to
23954 bitmap_head, and adjust ctor / dtor and member operators.
23956 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23958 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23959 when returned register mode doesn't match original mode.
23961 2017-05-12 Jeff Law <law@redhat.com>
23962 Jakub Jelinek <jakub@redhat.com>
23964 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23965 we look for cc setter after the compare-elim changes.
23966 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23967 within the vector to match what compare-elim now expects.
23968 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23969 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23971 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23972 after the compare-elim changes.
23973 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23974 the vector to match what compare-elim now expects.
23975 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23976 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23977 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23978 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23979 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23981 * config/visium/visium.c (single_set_and_flags): Fix where
23982 we look for cc setter after the compare-elim changes.
23983 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23984 with the vector to match what compare-elim now expects.
23985 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23986 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23987 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23988 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23989 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23990 (neg<mode>2_insn_set_overflow): Likewise.
23992 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23994 PR middle-end/79794
23995 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23996 maybe_expand_insn call, set ops[0].target. If still set after call,
23997 set alt_rtl. Add extra arg to recursive calls.
23998 (extract_bit_field): Add alt_rtl argument. Pass to
24000 * expmed.h (extract_bit_field): Fix prototype.
24001 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24002 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24003 to extract_bit_field_calls.
24004 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24005 Pass alt_rtl to extract_bit_field calls.
24006 * calls.c (store_unaligned_arguments_into_psuedos)
24007 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24008 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24010 * optabs.h (struct expand_operand): Add target bitfield.
24012 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24014 * compare-elim.c (try_eliminate_compare): Canonicalize
24015 operation with embedded compare to
24016 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24017 (set (reg) (operation)].
24019 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24021 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24024 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24025 cost of adding a carry flag for ADC instruction.
24026 [case MINUS]: Ignore the cost of subtracting a carry flag
24027 for SBB instruction.
24029 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24031 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24033 * config/rs6000/bmiintrin.h: New file.
24034 * config/rs6000/bmi2intrin.h: New file.
24035 * config/rs6000/x86intrin.h: New file.
24037 2017-05-12 Jeff Law <law@redhat.com>
24039 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24042 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24044 PR middle-end/80707
24045 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24047 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24050 early expansion of vector divide builtins.
24051 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24052 builtins identified as having unsigned arguments.
24054 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24056 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24057 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24058 expansion of vector logical operations (and, andc, or, xor,
24061 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24063 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24064 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24066 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24068 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24069 early GIMPLE expansion of vector multiplies.
24071 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24073 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24074 TARGET_HAVE_MOVT conditional.
24075 (movt splitter): Likewise.
24077 2017-05-12 Richard Biener <rguenther@suse.de>
24079 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24080 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24081 Fold all stmts not inplace.
24083 2017-05-12 Richard Biener <rguenther@suse.de>
24085 PR tree-optimization/80713
24086 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24087 inserted_exprs bit for not removed stmts.
24089 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24091 PR middle-end/69921
24092 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24093 parallelized" attribute for parallelized OpenACC kernels.
24094 * omp-offload.c (execute_oacc_device_lower): Use it.
24096 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24097 Set "oacc kernels" attribute.
24098 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24099 parameter. Adjust all users.
24100 (oacc_fn_attrib_kernels_p): Remove function.
24101 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24102 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24103 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24104 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24105 assert "oacc kernels" attribute is set.
24107 2017-05-11 Carl Love <cel@us.ibm.com>
24109 * config/rs6000/rs6000-c: Add support for built-in functions
24110 vector unsigned char vec_popcnt (vector signed char)
24111 vector unsigned char vec_popcnt (vector unsigned char)
24112 vector unsigned short vec_popcnt (vector signed short)
24113 vector unsigned short vec_popcnt (vector unsigned short)
24114 vector unsigned int vec_popcnt (vector signed int)
24115 vector unsigned int vec_popcnt (vector unsigned int)
24116 vector unsigned long long vec_popcnt (vector signed long long)
24117 vector unsigned long long vec_popcnt (vector unsigned long long)
24118 vector signed long long vec_slo (vector signed long long,
24119 vector signed char)
24120 vector signed long long vec_slo (vector signed long long,
24121 vector unsigned char)
24122 vector unsigned long long vec_slo (vector unsigned long long,
24123 vector signed char)
24124 vector unsigned long long vec_slo (vector unsigned long long,
24125 vector unsigned char)
24126 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24127 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24128 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24129 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24130 * doc/extend.texi: Update the built-in documentation file for the
24131 new built-in functions.
24133 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24135 * attribs.h (sorted_attr_string): Move machine independent
24136 functions for target clone support from the i386 port to common
24137 code. Rename ix86_function_versions to common_function_versions.
24138 Rename make_name to make_unique_name.
24139 (common_function_versions): Likewise.
24140 (make_unique_name): Likewise.
24141 (make_dispatcher_decl): Likewise.
24142 (is_function_default_version): Likewise.
24143 * attribs.c (attr_strcmp): Likewise.
24144 (sorted_attr_string): Likewise.
24145 (common_function_versions): Likewise.
24146 (make_unique_name): Likewise.
24147 (make_dispatcher_decl): Likewise.
24148 (is_function_default_version): Likewise.
24149 * config/i386/i386.c (attr_strcmp): Likewise.
24150 (sorted_attr_string): Likewise.
24151 (ix86_function_versions): Likewise.
24152 (make_name): Likewise.
24153 (make_dispatcher_decl): Likewise.
24154 (is_function_default_version): Likewise.
24155 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24157 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24160 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24161 Account for direct move costs for vec_construct of integer
24164 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24167 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24168 (UNSPEC_STX_ATOMIC): Ditto.
24169 (loaddi_via_sse): New insn.
24170 (storedi_via_sse): Ditto.
24171 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24172 Update corresponding peephole2 patterns.
24173 (atomic_storedi_fpu): Ditto.
24175 2017-05-11 Julia Koval <julia.koval@intel.com>
24177 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24178 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24180 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24181 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24182 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24184 2017-05-11 Nathan Sidwell <nathan@acm.org>
24186 * graphite-poly.c: Include dumpfile.h.
24188 * dumpfle.h (dump_function): Declare here ...
24189 * tree-dump.h (dump_function): ... not here.
24190 * dumpfile.c: #include tree-cfg.h.
24191 (dump_function): Move here from ...
24192 * tree-dump.c (dump_function): ... here.
24193 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24194 * graphite-poly.c: Don't include tree-dump.h.
24195 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24196 * print-tree.c: Likewise.
24197 * stor-layout.c: Likewise.
24198 * tree-nested.c: Likewise.
24200 * dumpfile.c (dump_start): Use TDF_FLAGS.
24201 (dump_enable_all): Fix TDF_KIND check thinko.
24203 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24205 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24206 array entries to represent two legal parameterizations of the
24207 overloaded __builtin_cmpb function, as represented by the
24208 P6_OV_BUILTIN_CMPB constant.
24209 (altivec_resolve_overloaded_builtin): Add special case handling
24210 for the __builtin_cmpb function, as represented by the
24211 P6_OV_BUILTIN_CMPB constant.
24212 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24213 (BU_P6_64BIT_2): New macro.
24214 (BU_P6_OVERLOAD_2): New macro
24215 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24216 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24217 (CMPB): Add overload support to represent both 32-bit and 64-bit
24218 compare-bytes function.
24219 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24220 support for TARGET_CMPB.
24221 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24223 documentation of the __builtin_cmpb overloaded built-in function.
24225 2017-05-11 Richard Biener <rguenther@suse.de>
24227 PR tree-optimization/80705
24228 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24229 bases are not vectorizable.
24231 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24233 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24234 when counting register pressure.
24236 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24238 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24239 (iv_ca_more_deps): Renamed to ...
24240 (iv_ca_compare_deps): ... this.
24241 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24245 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24247 (determine_group_iv_costs): ... here.
24248 (find_inv_vars_cb): Record inv var if it's not recorded before.
24250 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24252 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24253 (get_shiftadd_cost): Ditto.
24255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24257 * tree-ssa-address.c: Include header file.
24258 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24260 (add_to_parts): Refactor.
24261 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24262 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24265 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24267 PR tree-optimization/53090
24268 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24270 (extract_cond_operands): Detect condition with IV on both sides
24271 and return COMP_IV_EXPR_2.
24272 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24273 (rewrite_use_compare): Simplify by removing call to function
24274 extract_cond_operands.
24276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24278 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24279 (extract_cond_operands): Detect condition comparing against non-
24280 invariant bound and return appropriate enum value.
24281 (find_interesting_uses_cond): Update use of extract_cond_operands.
24282 Handle its return value accordingly.
24283 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24285 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24287 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24288 nonlinear iv_use computation in loop invariant sensitive way.
24290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24292 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24293 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24297 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24298 (dump_cand): Support iv_cand.inv_exprs.
24299 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24301 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24304 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24306 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24308 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24309 as local function. Include necessary header files.
24310 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24312 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24314 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24318 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24319 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24320 RSHIFT_EXPR and BIT_NOT_EXPR.
24322 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24324 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24325 (adjust_setup_cost): New parameter supporting round up adjustment.
24326 (struct address_cost_data): Delete.
24327 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24328 (split_address_cost, ptr_difference_cost): Delete.
24329 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24330 (struct ainc_cost_data): New struct.
24331 (get_address_cost_ainc): New function.
24332 (get_address_cost, get_computation_cost): Reimplement.
24333 (determine_group_iv_cost_address): Record inv_expr for all uses of
24335 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24336 (iv_ca_has_deps): Reimplemented to ...
24337 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24339 (iv_ca_extend): Call iv_ca_more_deps.
24341 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24343 * tree-ssa-address.c (struct mem_address): Move to header file.
24344 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24345 * tree-ssa-address.h (struct mem_address): Move from C file.
24346 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24350 * tree-affine.h (aff_combination_type): New interface.
24351 (aff_combination_zero_p): Remove static.
24352 (aff_combination_const_p): New interface.
24353 (aff_combination_singleton_var_p): New interfaces.
24355 2017-05-11 Richard Biener <rguenther@suse.de>
24357 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24358 Skip unreachable blocks and destinations.
24359 (eliminate): Move stmt removal and fixup ...
24360 (fini_eliminate): ... here. Skip inserted exprs.
24361 (pass_pre::execute): Move fini_pre after fini_eliminate.
24362 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24363 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24364 PRE to get rid of dead code that has invalid SSA form and
24365 split critical edges again.
24367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24369 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24371 2017-05-11 Richard Biener <rguenther@suse.de>
24373 * passes.c (execute_function_todo): Verify loops if they are
24374 said to be up-to-date.
24375 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24376 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24378 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24381 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24382 handle calling assemble_external ourself.
24385 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24386 modes with zero size. Enhance comment.
24388 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24390 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24391 built-ins for vec_xl and vec_xst with short and char pointer
24394 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24396 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24397 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24398 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24399 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24400 (_mm_maskz_min_round_ss): New intrinsics.
24401 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24402 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24403 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24404 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24405 (__builtin_ia32_minss_mask_round): New builtins.
24406 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24407 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24408 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24410 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24411 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24413 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24416 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24418 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24419 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24420 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24421 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24422 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24423 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24424 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24425 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24426 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24427 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24428 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24429 (__builtin_ia32_mulss_mask_round): New builtins.
24430 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24431 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24432 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24434 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24435 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24437 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24440 2017-05-10 Julia Koval <julia.koval@intel.com>
24442 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24443 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24444 (_mm256_setr_m128i): New intrinsics.
24446 2017-05-10 Julia Koval <julia.koval@intel.com>
24448 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24449 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24450 (_mm_maskz_rcp14_ss): New intrinsics.
24451 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24452 (__builtin_ia32_rcp14ss_mask): New builtins.
24453 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24455 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24457 PR tree-optimization/51513
24458 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24459 (assert_unreachable_fallthru_edge_p): Use it.
24460 (group_case_labels_stmt): Likewise.
24461 * tree-cfg.h: Prototype it.
24462 * stmt.c: Include cfghooks.h and tree-cfg.h.
24463 (emit_case_dispatch_table) <gap_label>: New local variable.
24464 Use it to fill dispatch table gaps.
24465 Test for default_label before updating probabilities.
24466 (expand_case) <default_label>: Remove unneeded initialization.
24467 Test for unreachable default case statement and remove its edge.
24468 Set default_label accordingly.
24469 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24471 2017-05-10 Carl Love <cel@us.ibm.com>
24473 * config/rs6000/rs6000-c: Add support for built-in functions
24474 vector signed char vec_neg (vector signed char)
24475 vector signed short int vec_neg (vector short int)
24476 vector signed int vec_neg (vector signed int)
24477 vector signed long long vec_neg (vector signed long long)
24478 vector float vec_neg (vector float)
24479 vector double vec_neg (vector double)
24480 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24482 * config/rs6000/altivec.h: Add define for vec_neg
24483 * doc/extend.texi: Update the built-in documentation for the
24484 new built-in functions.
24486 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24488 PR tree-optimization/77644
24489 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24491 2017-05-10 Nathan Sidwell <nathan@acm.org>
24493 * dumpfile.h (TDI_lang_all): New.
24494 (TDF_KIND): New. Renumber others
24495 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24497 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24499 (get_dump_file_name): Adjust suffix generation.
24500 (dump_enable_all): Use TDF_KIND.
24501 * doc/invoke.texi (-fdump-lang-all): Document.
24503 * dumpfile.h: Tabify.
24505 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24508 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24509 Move member access before delete.
24511 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24513 * tree-inline.c (expand_call_inline): Split block at stmt
24516 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24519 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24520 are now unused after splitting mov{sf,sd}_hardfloat.
24521 (f32_lr2): Likewise.
24522 (f32_lm): Likewise.
24523 (f32_lm2): Likewise.
24524 (f32_li): Likewise.
24525 (f32_li2): Likewise.
24526 (f32_lv): Likewise.
24527 (f32_sr): Likewise.
24528 (f32_sr2): Likewise.
24529 (f32_sm): Likewise.
24530 (f32_sm2): Likewise.
24531 (f32_si): Likewise.
24532 (f32_si2): Likewise.
24533 (f32_sv): Likewise.
24534 (f32_dm): Likewise.
24535 (f32_vsx): Likewise.
24536 (f32_av): Likewise.
24537 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24538 For movsf, order stores so the VSX stores occur before the GPR
24539 store which encourages the register allocator to use a traditional
24540 FPR instead of a GPR. For movsd, order the stores so that the GPR
24541 store comes before the VSX stores to allow the power6 to work.
24542 This is due to the power6 not having a 32-bit integer store
24543 instruction from a FPR.
24544 (movsf_hardfloat): Likewise.
24545 (movsd_hardfloat): Likewise.
24547 2017-05-09 Martin Sebor <msebor@redhat.com>
24549 PR translation/80280
24550 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24553 PR translation/80280
24554 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24555 data member added in r247778.
24556 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24558 2017-05-09 Nathan Sidwell <nathan@acm.org>
24560 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24562 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24565 2017-05-09 Marek Polacek <polacek@redhat.com>
24567 * doc/invoke.texi: Fix typo.
24569 2017-05-09 Richard Biener <rguenther@suse.de>
24571 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24572 (vrp_val_is_min): Likewise.
24573 (set_value_range_to_value): Likewise.
24574 (set_value_range_to_nonnegative): Likewise.
24575 (gimple_assign_nonzero_p): Likewise.
24576 (gimple_stmt_nonzero_p): Likewise.
24577 (vrp_int_const_binop): Likewise. Remove unreachable case.
24578 (adjust_range_with_scev): Adjust comments.
24579 (compare_range_with_value): Likewise.
24580 (extract_range_from_phi_node): Likewise.
24581 (test_for_singularity): Likewise.
24583 2017-05-09 Richard Biener <rguenther@suse.de>
24585 * tree-vrp.c (get_single_symbol): Add assert that we don't
24586 get overflowed constants as invariant part.
24587 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24588 checks. Use wi::cmp instead of recursing for integer constants.
24589 (compare_values): Just ignore whether we assumed undefined
24590 overflow instead of failing the compare.
24591 (extract_range_for_var_from_comparison_expr): Add comment before the
24592 TREE_NO_WARNING sets.
24593 (test_for_singularity): Likewise.
24594 (extract_range_from_comparison): Do not disable optimization
24595 when we assumed undefined overflow.
24596 (extract_range_basic): Remove init of unused var.
24598 2017-05-09 Richard Biener <rguenther@suse.de>
24600 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24601 (extract_range_from_multiplicative_op_1): Adjust.
24602 (extract_range_from_binary_expr_1): Use int_const_binop.
24604 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24607 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24608 rs6000_store_data_bypass_p in seven define_bypass directives and
24609 in several comments.
24610 * config/rs6000/rs6000-protos.h: Add prototype for
24611 rs6000_store_data_bypass_p function.
24612 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24613 function implements slightly different (rs6000-specific) semantics
24614 than store_data_bypass_p, returning false rather than aborting
24615 with assertion error when arguments do not satisfy the
24616 requirements of store data bypass.
24617 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24618 rs6000_store_data_bypass_p.
24620 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24622 * config/xtensa/xtensa-protos.h
24623 (xtensa_initial_elimination_offset): New declaration.
24624 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24625 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24626 macro definition, add case for FRAME_POINTER_REGNUM when
24627 FRAME_GROWS_DOWNWARD.
24628 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24629 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24630 xtensa_initial_elimination_offset.
24632 2017-05-08 Nathan Sidwell <nathan@acm.org>
24634 * doc/invoke.texi: Alphabetize -fdump options.
24636 2017-05-08 Martin Sebor <msebor@redhat.com>
24638 PR translation/80280
24639 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24641 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24643 * target.def (compute_frame_layout): New optional target hook.
24644 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24645 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24646 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24648 * reload1.c (verify_initial_elim_offsets): Likewise.
24649 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24650 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24651 (arm_get_frame_offsets): Split up into this ...
24652 (arm_compute_frame_layout): ... and this function.
24654 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24656 * config/aarch64/constraints.md (Usa): New constraint.
24657 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24659 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24661 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24662 with_multilib_list after it has been checked.
24664 2017-05-08 Richard Biener <rguenther@suse.de>
24666 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24667 (bitmap_set_subtract_values): Likewise.
24669 2017-05-08 Richard Biener <rguenther@suse.de>
24671 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24672 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24674 (gimple_stmt_nonzero_warnv_p): Rename to ...
24675 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24677 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24678 (extract_range_basic): Adjust, do not disable propagation on
24679 strict overflow sensitive simplification.
24680 (vrp_visit_cond_stmt): Likewise.
24682 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24684 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24685 body size unconditionally.
24687 2017-05-07 Jeff Law <law@redhat.com>
24690 2017-05-06 Jeff Law <law@redhat.com>
24691 PR tree-optimization/78496
24692 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24695 PR tree-optimization/78496
24696 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24697 (simplify_stmt_using_ranges): Call it.
24698 (vrp_dom_walker::before_dom_children): Extract equivalences
24699 from an ASSERT_EXPR with an equality comparison against a
24702 2017-05-06 Jeff Law <law@redhat.com>
24704 PR tree-optimization/78496
24705 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24708 PR tree-optimization/78496
24709 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24710 (simplify_stmt_using_ranges): Call it.
24711 (vrp_dom_walker::before_dom_children): Extract equivalences
24712 from an ASSERT_EXPR with an equality comparison against a
24715 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24717 * lra-constraints.c (lra_copy_reg_equiv): New function.
24718 (split_reg): Use it to copy equivalence information from the
24719 original register to the spill register.
24721 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24723 PR rtl-optimization/75964
24724 * simplify-rtx.c (simplify_const_relational_operation): Remove
24725 invalid handling of comparisons of integer ABS.
24727 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24729 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24730 initialize to zero.
24731 (init_regs): Remove declaration.
24732 (function_arg_advance_32): Initialize error_p as boolean variable.
24734 2017-05-05 Nathan Sidwell <nathan@acm.org>
24736 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24737 lines. Use for (;;).
24739 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24741 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24742 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24743 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24744 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24745 VF=2 that require versioning.
24747 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24749 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24752 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24754 * diagnostic.h (diagnostic_override_option_index): Convert from
24755 macro to inline function.
24757 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24759 * diagnostic.c (last_module_changed_p): New function.
24760 (set_last_module): New function.
24761 (diagnostic_report_current_module): Convert macro usage to
24762 the above functions.
24763 * diagnostic.h (diagnostic_context::last_module): Strengthen
24764 from const line_map * to const line_map_ordinary *.
24765 (diagnostic_last_module_changed): Delete macro.
24766 (diagnostic_set_last_module): Delete macro.
24768 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24770 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24771 with diagnostic_report_diagnostic.
24772 (diagnostic_n_impl_richloc): Likewise.
24773 * diagnostic.h (report_diagnostic): Delete macro.
24774 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24775 with diagnostic_report_diagnostic.
24776 * substring-locations.c (format_warning_va): Likewise.
24778 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24780 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24781 save/restor of format_spec. Move option-printing code to...
24782 (print_option_information): ...this new function, and
24783 reimplement by simply printing to the pretty_printer,
24784 rather than appending to the format string.
24786 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24788 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24789 handling logic into...
24790 (update_effective_level_from_pragmas): ...this new function.
24792 2017-05-04 Andrew Waterman <andrew@sifive.com>
24794 * config/riscv/riscv.opt (mstrict-align): New option.
24795 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24796 (SLOW_UNALIGNED_ACCESS): Define.
24797 (riscv_slow_unaligned_access): Declare.
24798 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24800 (riscv_slow_unaligned_access): New variable.
24801 (rocket_tune_info): Set slow_unaligned_access to true.
24802 (optimize_size_tune_info): Set slow_unaligned_access to false.
24803 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24804 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24805 (riscv_option_override): Set riscv_slow_unaligned_access.
24806 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24808 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24810 * config/riscv/riscv.md: Unify indentation.
24812 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24817 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24819 (extendsi<mode>2): Add support for doing sign extension via
24820 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24821 don't have ISA 3.0 instructions.
24822 (extendsi<mode>2 splitter): Likewise.
24823 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24824 generate the normal insns since SImode can now go in vector
24825 registers. Disallow the special UNSPECs needed for previous
24826 machines to hide SImode being used. Add new insns
24827 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24828 (fix_trunc<mode>si2_stfiwx): Likewise.
24829 (fix_trunc<mode>si2_internal): Likewise.
24830 (fixuns_trunc<mode>si2): Likewise.
24831 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24832 (fctiw<u>z_<mode>_smallint): Likewise.
24833 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24834 of floating point to 32-bit integer from doing a direct move to
24835 the GPR registers to do a store.
24836 (fctiwz_<mode>): Break long line.
24838 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24840 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24841 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24842 (addr_list, addr_offset_valid_p): New.
24843 (split_address_groups): Check offset validity with above function.
24844 (gt-tree-ssa-loop-ivopts.h): Include header file.
24846 2017-05-05 Nathan Sidwell <nathan@acm.org>
24848 * config.gcc (arm*-*-*): Add missing 'fi'.
24850 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24852 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24853 included in -fopt-info does not matter.
24854 * doc/optinfo.texi (-fopt-info): Fix description of default
24855 behavour. Explicitly say order of options included in -fopt-info
24858 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24860 * config.gcc: Allow combinations of aprofile and rmprofile values for
24861 --with-multilib-list.
24862 * config/arm/t-multilib: New file.
24863 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24864 variables. Remove setting of ISA and floating-point ABI in
24865 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24866 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24867 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24868 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24870 * config/arm/t-rmprofile: Likewise except for the matches changes.
24871 * doc/install.texi (--with-multilib-list): Document the combination of
24872 aprofile and rmprofile values and warn about pitfalls in doing that.
24874 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24876 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24877 (movdi_aarch64): Likewise.
24879 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24881 PR tree-optimization/80632
24882 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24884 (build_arrays): Initialize it for virtual phis.
24885 (fix_phi_nodes): Use it for virtual phis.
24887 PR tree-optimization/80558
24888 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24889 [x, y] op z into [x op, y op z] for op & or | if conditions
24892 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24893 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24896 * config/arm/arm.md (use_literal_pool): Remove.
24897 (64-bit immediate split): No longer takes cost into consideration
24898 if arm_disable_literal_pool is enabled.
24899 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24900 used when arm_disable_literal_pool is enabled.
24901 (arm_max_const_double_inline_cost): Remove use of
24902 arm_disable_literal_pool.
24903 (push_minipool_fix): Add assert.
24904 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24905 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24906 (no_literal_pool_sf_immediate): New.
24908 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24910 PR tree-optimization/80613
24911 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24912 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24914 2017-05-05 Richard Biener <rguenther@suse.de>
24916 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24918 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24920 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24921 of this flag from insn conditions due to removal from r247495.
24923 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24925 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24927 (arm_early_store_addr_dep_ptr): Likewise.
24928 * config/arm/aarch-common-protos.h
24929 (arm_early_load_addr_dep_ptr): Add prototype.
24930 (arm_early_store_addr_dep_ptr): Likewise.
24931 * config/arm/cortex-a53.md: Add new bypasses.
24933 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24935 * tree.c (next_type_uid): Change type to unsigned.
24936 (type_hash_canon): Decrement back next_type_uid if
24937 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24938 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24941 2017-05-04 Martin Sebor <msebor@redhat.com>
24943 * builtins.c: Fix a trivial typo in a comment.
24945 PR middle-end/79234
24946 * builtins.c (check_sizes): Adjust to handle reading past the end.
24947 Avoid printing excessive upper bound of ranges. Use %E to print
24948 tree nodes instead of converting them to %wu.
24949 (expand_builtin_memchr): New function.
24950 (compute_dest_size): Rename...
24951 (compute_objsize): ...to this.
24952 (expand_builtin_memcpy): Adjust.
24953 (expand_builtin_mempcpy): Adjust.
24954 (expand_builtin_strcat): Adjust.
24955 (expand_builtin_strcpy): Adjust.
24956 (check_strncat_sizes): Adjust.
24957 (expand_builtin_strncat): Adjust.
24958 (expand_builtin_strncpy): Adjust and simplify.
24959 (expand_builtin_memset): Adjust.
24960 (expand_builtin_bzero): Adjust.
24961 (expand_builtin_memcmp): Adjust.
24962 (expand_builtin): Handle memcmp.
24963 (maybe_emit_chk_warning): Check strncat just once.
24965 2017-05-04 Martin Sebor <msebor@redhat.com>
24967 PR preprocessor/79214
24968 PR middle-end/79222
24969 PR middle-end/79223
24970 * builtins.c (check_sizes): Add inlining context and issue
24971 warnings even when -Wno-system-headers is set.
24972 (check_strncat_sizes): Same.
24973 (expand_builtin_strncat): Same.
24974 (expand_builtin_memmove): New function.
24975 (expand_builtin_stpncpy): Same.
24976 (expand_builtin): Handle memmove and stpncpy.
24978 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24980 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24981 which is not used any more.
24983 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24985 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24987 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24989 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24990 (cortexa53_tunings): Likewise.
24991 (cortexa57_tunings): Likewise.
24992 (cortexa72_tunings): Likewise.
24993 (cortexa73_tunings): Likewise.
24995 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24997 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24998 Set loop alignment to 8.
25000 2017-05-04 Martin Sebor <msebor@redhat.com>
25002 PR translation/80280
25003 * builtins.c (expand_builtin_object_size): Add missing quoting to
25004 %D and like directives.
25005 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25006 (hsa_type_for_tree_type): Same.
25007 (verify_function_arguments): Same.
25008 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25009 * varasm.c (get_section): Same.
25012 2017-05-04 Martin Sebor <msebor@redhat.com>
25014 PR translation/80280
25015 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25017 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25019 * config/aarch64/aarch64.c (generic_addrcost_table):
25020 Change HI/TI mode setting.
25022 2017-05-04 Martin Jambor <mjambor@suse.cz>
25024 PR tree-optimization/80622
25025 * tree-sra.c (comes_initialized_p): New function.
25026 (build_accesses_from_assign): Only set write lazily when
25027 comes_initialized_p is false.
25028 (analyze_access_subtree): Use comes_initialized_p.
25029 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25030 instead of testing for PARM_DECL.
25032 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25034 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25035 constraint on operand 0 to allow more general addressing modes.
25036 Adjust output template.
25037 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25039 * config/aarch64/aarch64-protos.h
25040 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25041 * config/aarch64/constraints.md (Dp): New address constraint.
25042 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25045 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25047 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25048 update use of estimate_ipcp_clone_size_and_time.
25049 (estimate_local_effects): Update use of
25050 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25051 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25052 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25053 Return nonspecialized time.
25055 2017-05-04 Richard Biener <rguenther@suse.de>
25057 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25058 for the last VUSE which def dominates the PHI. Directly call
25060 (get_continuation_for_phi_1): Remove.
25062 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25064 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25065 to explain the use of truncating division. Cap the number of
25066 iterations to the maximum given by nb_iterations_upper_bound,
25069 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25071 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25072 * configure: Regenerate.
25073 * config.in: Regenerate.
25074 * config/i386/driver-mingw32.c: new file.
25075 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25076 * config.host: Link driver-mingw32.o on MinGW host.
25077 * doc/install.texi: Document new --enable-mingw-wildcard configure
25080 2017-05-04 Marek Polacek <polacek@redhat.com>
25082 PR tree-optimization/80612
25083 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25085 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25086 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25088 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25089 (movt splitter): Likewise.
25090 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25091 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25092 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25093 block for Thumb-1 with MOVT.
25094 (thumb2_legitimate_address_p): Move code block ...
25095 (can_avoid_literal_pool_for_label_p): ... into this new function.
25096 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25098 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25099 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25100 "M-profile targets with the MOVT instruction".
25102 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25104 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25105 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25106 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25108 2017-05-04 Martin Liska <mliska@suse.cz>
25110 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25111 variable cond_code.
25113 2017-05-04 Richard Biener <rguenther@suse.de>
25115 * tree.c (array_at_struct_end_p): Handle arrays at struct
25116 end with flexarrays more conservatively. Refactor and treat
25117 arrays of arrays or aggregates more strict. Fix
25118 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25119 * tree.c (array_at_struct_end_p): Adjust prototype.
25120 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25121 * gimple-fold.c (get_range_strlen): Likewise.
25122 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25124 2017-05-04 Richard Biener <rguenther@suse.de>
25126 PR tree-optimization/31130
25127 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25129 (supports_overflow_infinity): Likewise.
25130 (is_negative_overflow_infinity): Likewise.
25131 (is_positive_overflow_infinity): Likewise.
25132 (is_overflow_infinity): Likewise.
25133 (stmt_overflow_infinity): Likewise.
25134 (overflow_infinity_range_p): Likewise.
25135 (usable_range_p): Remove as always returning true.
25136 (make_overflow_infinity): Remove.
25137 (negative_overflow_infinity): Likewise.
25138 (positive_overflow_infinity): Likewise.
25139 (avoid_overflow_infinity): Likewise.
25140 (set_value_range): Adjust accordingly.
25141 (set_value_range_to_nonnegative): Likewise, remove now unused
25142 overflow_infinity arg.
25143 (vrp_operand_equal_p): Adjust.
25144 (update_value_range): Likewise.
25145 (range_int_cst_singleton_p): Likewise.
25146 (operand_less_p): Likewise.
25147 (compare_values_warnv): Likewise.
25148 (extract_range_for_var_from_comparison_expr): Likewise.
25149 (vrp_int_const_binop): Likewise.
25150 (zero_nonzero_bits_from_vr): Likewise.
25151 (extract_range_from_multiplicative_op_1): Likewise.
25152 (extract_range_from_binary_expr_1): Likewise.
25153 (extract_range_from_unary_expr): Likewise.
25154 (extract_range_from_comparison): Likewise.
25155 (extract_range_basic): Likewise.
25156 (adjust_range_with_scev): Likewise.
25157 (compare_ranges): Likewise.
25158 (compare_range_with_value): Likewise.
25159 (dump_value_range): Likewise.
25160 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25162 (simplify_cond_using_ranges): Adjust.
25164 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25166 * brig-builtins.def: Added a builtin for class_f64.
25167 * builtin-types.def: Added a builtin type needed by class_f64.
25169 2017-05-03 Jason Merrill <jason@redhat.com>
25171 * timevar.def: Add TV_CONSTEXPR.
25173 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25175 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25177 2017-05-03 Martin Jambor <mjambor@suse.cz>
25179 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25180 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25181 * ipa-cp.c (ipcp_propagate_stage): Do not call
25182 ipa_update_after_lto_read.
25183 * ipa-inline.c (ipa_inline): Likewise.
25185 2017-05-03 Martin Jambor <mjambor@suse.cz>
25187 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25188 tag. Added a default constructor and a destructor.
25189 (ipa_edge_args_sum_t): New class;
25190 (ipa_edge_args_sum): Declare.
25191 (ipa_edge_args_vector): Remove declaration.
25192 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25193 (ipa_free_edge_args_substructures): Remove declaration.
25194 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25195 (ipa_edge_args_info_available_for_edge_p): Likewise.
25196 * ipa-prop.c (ipa_edge_args_vector): Removed.
25197 (edge_removal_hook_holder): Likewise.
25198 (edge_duplication_hook_holder): Likewise.
25199 (ipa_edge_args_sum): New variable.
25200 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25201 ipa_edge_args_vector.
25202 (ipa_free_edge_args_substructures): Likewise.
25203 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25204 ipa_edge_args_vector.
25205 (ipa_edge_removal_hook): Turned into method
25206 ipa_edge_args_sum_t::remove.
25207 (ipa_edge_duplication_hook): Turned into method
25208 ipa_edge_args_sum_t::duplicate.
25209 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25210 registering edge hooks.
25211 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25212 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25213 ipa_edge_args_sum instead of ipa_edge_args_vector.
25214 * ipa-profile.c (ipa_profile): Likewise.
25216 2017-05-03 Martin Jambor <mjambor@suse.cz>
25218 * symbol-summary.h (function_summary): New method exists.
25219 (function_summary::symtab_removal): Deallocate through release.
25220 (call_summary): New class.
25221 (gt_ggc_mx): New overload.
25222 (gt_pch_nx): Likewise.
25223 (gt_pch_nx): Likewise.
25225 2017-05-03 Jeff Law <law@redhat.com>
25227 PR tree-optimization/78496
25228 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25229 from simplify_cond_using_ranges. Split off code to walk
25230 backwards through casts into ...
25231 (simplify_cond_using_ranges_2): New function.
25232 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25233 (execute_vrp): After identifying jump threads, call
25234 simplify_cond_using_ranges_2.
25236 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25239 * ipa-inline.h (inline_summary): Add ctor.
25240 (create_ggc): Do not use ggc_cleared_alloc.
25242 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25243 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25245 * gcc.c (handle_braces): Support escaping in switch matching
25247 * doc/invoke.texi (Spec Files): Document it.
25248 Remove superfluous @code markup in items.
25250 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25252 * diagnostic-show-locus.c (struct column_range): New struct.
25253 (get_affected_columns): New function.
25254 (get_printed_columns): New function.
25255 (struct correction): New struct.
25256 (correction::ensure_capacity): New function.
25257 (correction::ensure_terminated): New function.
25258 (struct line_corrections): New struct.
25259 (line_corrections::~line_corrections): New dtor.
25260 (line_corrections::add_hint): New function.
25261 (layout::print_trailing_fixits): Reimplement in terms of the new
25263 (selftest::test_overlapped_fixit_printing): New function.
25264 (selftest::diagnostic_show_locus_c_tests): Call it.
25266 2017-05-03 Nathan Sidwell <nathan@acm.org>
25268 Canonicalize canonical type hashing
25269 * tree.h (type_hash_canon_hash): Declare.
25270 * tree.c (type_hash_list, attribute_hash_list): Move into
25271 type_hash_canon_hash.
25272 (build_type_attribute_qual_variant): Break out hash code calc into
25273 type_hash_canon_hash.
25274 (type_hash_canon_hash): New. Generic type hash computation.
25275 (build_range_type_1, build_array_type_1, build_function_type,
25276 build_method_type_directly, build_offset_type, build_complex_type,
25277 make_vector_type): Call it.
25279 2017-05-03 Richard Biener <rguenther@suse.de>
25281 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25282 When all DRs have unknown misaligned do not always peel
25283 when there is a store but apply the same costing model as if
25284 there were only loads.
25286 2017-05-03 Richard Biener <rguenther@suse.de>
25289 PR tree-optimization/80492
25290 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25291 compare_base_decls returning dont-know properly.
25293 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25295 * config/arm/iterators.md (CCSI): New mode iterator.
25296 (arch): New mode attribute.
25297 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25298 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25299 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25300 code iterator for success result mode.
25301 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25302 the corresponding new insn generators.
25304 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25307 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25308 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25310 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25312 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25313 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25314 (DDR_A): Wrap DDR argument in brackets.
25315 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25316 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25317 (DDR_REVERSED_P): Likewise.
25319 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25321 PR tree-optimization/79472
25322 * tree-switch-conversion.c (struct switch_conv_info): Add
25323 contiguous_range and default_case_nonstandard fields.
25324 (collect_switch_conv_info): Compute contiguous_range and
25325 default_case_nonstandard fields, don't clear final_bb if
25326 contiguous_range and only the default case doesn't have the required
25328 (check_all_empty_except_final): Set default_case_nonstandard instead
25329 of failing if contiguous_range and the default case doesn't have empty
25331 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25332 and only the default case doesn't have the required constants. Skip
25334 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25335 if default_case_nonstandard.
25336 (build_constructors): Build constant 1 just once. Assert that default
25337 values aren't inserted in between cases if contiguous_range. Skip
25339 (build_arrays): Skip virtual phis.
25340 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25341 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25342 Handle virtual phis.
25343 (gen_inbound_check): Handle default_case_nonstandard case.
25344 (process_switch): Adjust check_final_bb caller. Call
25345 gather_default_values with the first non-default case instead of
25346 default case if default_case_nonstandard.
25348 2017-05-02 Nathan Sidwell <nathan@acm.org>
25350 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25351 check. Fix formatting.
25353 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25355 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25356 errors when comparing specialized and unspecialized times.
25358 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25360 * diagnostic-show-locus.c
25361 (layout::should_print_annotation_line_p): Make private.
25362 (layout::print_annotation_line): Make private.
25363 (layout::annotation_line_showed_range_p): Make private.
25364 (layout::show_ruler): Make private.
25365 (layout::print_source_line): Make private. Pass in line and
25366 line_width, rather than calling location_get_source_line. Drop
25368 (layout::print_leading_fixits): New method.
25369 (layout::print_any_fixits): Rename to...
25370 (layout::print_trailing_fixits): ...this, and make private.
25371 Don't print newline fixits.
25372 (diagnostic_show_locus): Move logic for printing one row into...
25373 (layout::print_line): ...this new function. Move the
25374 location_get_source_line call and error-handling from
25375 print_source_line to here. Call print_leading_fixits, and rename
25376 print_any_fixits to print_trailing_fixits.
25377 (selftest::test_fixit_insert_containing_newline): Update now that
25378 newlines are partially supported.
25379 (selftest::test_fixit_insert_containing_newline_2): New test.
25380 (selftest::test_fixit_replace_containing_newline): Update comments.
25381 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25382 * edit-context.c (class added_line): New class.
25383 (class edited_line): Describe newline handling in comment.
25384 (edited_line::actually_edited_p): New method.
25385 (edited_line::print_content): Delete redundant decl.
25386 (edited_line::m_predecessors): New field.
25387 (edited_file::print_content): Call edited_line::print_content.
25388 (edited_file::print_diff): Update to support newlines.
25389 (edited_file::print_diff_hunk): Likewise.
25390 (edited_file::print_run_of_changed_lines): New function.
25391 (edited_file::print_diff_line): Convert to...
25392 (print_diff_line): ...this.
25393 (edited_file::get_effective_line_count): New function.
25394 (edited_line::edited_line): Initialize new field m_predecessors.
25395 (edited_line::~edited_line): Clean up m_predecessors.
25396 (edited_line::apply_fixit): Handle newlines.
25397 (edited_line::get_effective_line_count): New function.
25398 (edited_line::print_content): New function.
25399 (edited_line::print_diff_lines): New function.
25400 (selftest::test_applying_fixits_insert_containing_newline): New
25402 (selftest::test_applying_fixits_replace_containing_newline): New
25404 (selftest::insert_line): New function.
25405 (selftest::test_applying_fixits_multiple_lines): Add example of
25407 (selftest::edit_context_c_tests): Call the new tests.
25409 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25411 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25412 parameter cand. Update dump information.
25413 (get_computation_cost): Update uses.
25415 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25417 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25418 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25419 (get_computation_at, rewrite_use_address): Update use of
25420 get_computation_aff.
25422 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25424 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25425 (get_computation): Delete.
25426 (get_computation_cost): Implement like get_computation_cost_at.
25427 Use get_computation_at.
25428 (get_computation_cost_at): Delete.
25429 (rewrite_use_nonlinear_expr): Use get_computation_at.
25430 (rewrite_use_compare, remove_unused_ivs): Ditto.
25432 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25434 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25436 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25438 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25439 (ivopts_global_cost_for_size): Rename parameter and update uses.
25440 (iv_ca_recount_cost): Update uses.
25441 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25442 candidates seperately in n_invs and n_cands.
25443 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25445 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25447 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25448 (find_inv_vars_cb): New.
25449 (find_depends): Renamed to ...
25450 (find_inv_vars): ... this.
25451 (add_candidate_1, force_var_cost): Call find_inv_vars.
25452 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25454 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25456 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25457 inv_vars. Add inv_exprs.
25458 (struct iv_cand): Rename depends_on to inv_vars.
25459 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25460 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25461 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25462 (dump_cand): Dump inv_vars.
25463 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25464 (record_invariant, find_depends, add_candidate_1): Ditto.
25465 (set_group_iv_cost, force_var_cost): Ditto.
25466 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25467 (get_computation_cost_at, get_computation_cost): Ditto.
25468 (determine_group_iv_cost_generic): Ditto.
25469 (determine_group_iv_cost_address): Ditto.
25470 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25471 (determine_group_iv_costs): Ditto.
25472 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25473 (iv_ca_set_remove_invariants): Renamed to ...
25474 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25475 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25476 (iv_ca_set_add_invariants): Renamed to ...
25477 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25478 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25479 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25480 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25481 (create_new_ivs): Remove useless dump.
25483 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25485 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25487 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25488 (iv_ca_set_no_cp, create_new_iv): Ditto.
25490 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25492 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25494 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25496 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25497 function tree_check2.
25499 2017-05-02 Martin Liska <mliska@suse.cz>
25501 * doc/gcov.texi: Add missing preposition.
25502 * gcov.c (function_info::function_info): Properly fill up
25503 all member variables.
25505 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25507 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25509 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25511 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25513 2017-05-02 Martin Liska <mliska@suse.cz>
25516 * lto-streamer-in.c (lto_read_tree_1): Remove
25517 LTO_STREAMER_DEBUG.
25518 * lto-streamer.c (struct tree_hash_entry): Likewise.
25519 (struct tree_entry_hasher): Likewise.
25520 (tree_entry_hasher::hash): Likewise.
25521 (tree_entry_hasher::equal): Likewise.
25522 (lto_streamer_init): Likewise.
25523 (lto_orig_address_map): Likewise.
25524 (lto_orig_address_get): Likewise.
25525 (lto_orig_address_remove): Likewise.
25526 * lto-streamer.h: Likewise.
25527 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25528 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25530 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25532 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25533 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25534 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25535 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25536 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25537 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25538 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25539 (mm_maskz_sub_ss): New intrinsics.
25540 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25541 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25542 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25543 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25544 (__builtin_ia32_subss_mask_round): New builtins.
25545 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25546 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25547 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25549 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25550 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25552 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25555 2017-05-02 Martin Jambor <mjambor@suse.cz>
25557 PR tree-optimization/78687
25558 * tree-sra.c (access): New field parent.
25559 (process_subtree_disqualification): New function.
25560 (disqualify_candidate): Call it.
25561 (build_accesses_from_assign): Reset write flag if creating an
25563 (build_access_subtree): Fill in parent field and also prpagate
25564 down grp_write flag.
25565 (create_artificial_child_access): New parameter set_grp_write, set
25566 grp_write to its value.
25567 (propagate_subaccesses_across_link): Also propagate grp_write flag
25569 (propagate_all_subaccesses): Push the closest parent back to work
25570 queue if add_access_to_work_queue returned true.
25572 2017-05-02 Richard Biener <rguenther@suse.de>
25574 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25575 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25576 -fstrict-overflow documentation.
25577 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25578 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25579 flag_strict_overflow.
25580 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25581 * lto-opts.c (lto_write_options): Do not stream it.
25582 * lto-wrapper.c (merge_and_complain): Do not handle it.
25583 * opts.c (default_options_table): Do not set -fstrict-overflow.
25584 (finish_options): Likewise do not clear it when sanitizing.
25585 * simplify-rtx.c (simplify_const_relational_operation): Do not
25586 test flag_strict_overflow.
25588 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25590 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25591 using enabled attribute.
25592 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25593 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25594 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25595 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25596 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25597 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25598 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25599 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25600 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25601 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25603 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25605 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25607 2017-05-02 Richard Biener <rguenther@suse.de>
25609 PR tree-optimization/80591
25611 2017-04-10 Richard Biener <rguenther@suse.de>
25613 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25616 2017-05-02 Richard Biener <rguenther@suse.de>
25618 PR tree-optimization/80549
25619 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25620 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25621 headers if they do not have a preheader.
25623 2017-05-02 Martin Liska <mliska@suse.cz>
25626 * common.opt: Fix typo.
25627 * doc/invoke.texi: Likewise.
25629 2017-05-01 Jan Beulich <jbeulich@suse.com>
25631 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25632 swapping, add (x,x,m,x,n) alternative.
25634 2017-05-01 Nathan Sidwell <nathan@acm.org>
25636 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25637 unnecessary unadjusted_alignment check.
25639 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25642 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25644 * gimplify.c (gimplify_cilk_detach): New function.
25645 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25646 * tree-core.h: Document EXPR_CILK_SPAWN.
25647 * tree.h (EXPR_CILK_SPAWN): Define.
25649 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25651 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25652 to use new fixit_hint representation, using the "replace" logic.
25653 (get_line_span_for_fixit_hint): Likewise.
25654 (layout::print_any_fixits): Likewise.
25655 (selftest::test_one_liner_many_fixits): Rename to...
25656 (selftest::test_one_liner_many_fixits_1): ...this, and update
25657 comment and expected output to reflect that the multiple fix-it
25658 hints are now consolidated into one insertion.
25659 (selftest::test_one_liner_many_fixits_2): New test.
25660 (selftest::test_diagnostic_show_locus_one_liner): Update for
25662 (selftest::test_fixit_consolidation): Update for fix-it API
25664 * diagnostic.c (print_parseable_fixits): Likewise.
25665 * edit-context.c (edited_line::m_line_events): Convert from
25666 auto_vec <line_event *> to auto_vec <line_event>.
25667 (class line_event): Convert from abstract base class to a concrete
25668 class, taking over the role of replace_event.
25669 (class insert_event): Delete.
25670 (class replace_event): Rename to class line_event. Convert to
25672 (edit_context::add_fixits): Reimplement.
25673 (edit_context::apply_insert): Delete.
25674 (edit_context::apply_replace): Rename to...
25675 (edit_context::apply_fixit): ...this. Convert to half-open range.
25676 (edited_file::apply_insert): Delete.
25677 (edited_file::apply_replace): Rename to...
25678 (edited_file::apply_fixit): ...this.
25679 (edited_line::~edited_line): Drop deletion of events.
25680 (edited_line::apply_insert): Delete.
25681 (edited_line::apply_replace): Rename to...
25682 (edited_line::apply_fixit): ...this. Convert to half-open range.
25683 Update for change to type of m_line_events.
25684 * edit-context.h (edit_context::apply_insert): Delete.
25685 (edit_context::apply_replace): Rename to...
25686 (edit_context::apply_fixit): ...this.
25688 2017-05-01 Martin Sebor <msebor@redhat.com>
25690 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25693 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25696 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25697 __get_cpuid_max returns 0.
25698 (__get_cpuid_count): Ditto.
25700 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25702 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25703 replacement expression is another instance of one of its arguments.
25705 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25708 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25709 check for stack push/pop autoinc.
25710 * config/i386/i386.c (ix86_agi_dependent): Return false
25711 if the only reason why modified_in_p returned true is that
25712 addr is SP based and set_insn is a push or pop.
25714 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25716 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25719 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25722 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25723 (account_size_time): Use two predicates - exec_pred and
25725 (evaluate_conditions_for_known_args): Compute both clause and
25727 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25728 (inline_summary_t::duplicate): Update.
25729 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25731 (compute_inline_parameters): Likewise.
25732 (estimate_edge_size_and_time): Update caluclation of time.
25733 (estimate_node_size_and_time): Compute both time and nonspecialized
25735 (estimate_ipcp_clone_size_and_time): Update.
25736 (inline_merge_summary): Update.
25737 (do_estimate_edge_time): Update.
25738 (do_estimate_edge_size): Update.
25739 (do_estimate_edge_hints): Update.
25740 (inline_read_section, inline_write_summary): Stream both new predicates.
25741 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25743 (compute_inlined_call_time): Cleanup.
25744 (big_speedup_p): Update.
25745 (edge_badness): Update.
25746 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25747 (size_time_entry): Replace predicate by exec_predicate and
25748 nonconst_predicate.
25749 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25750 (estimate_edge_time): Return also nonspec_time.
25751 (reset_edge_growth_cache): Update.
25753 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25755 PR rtl-optimization/80491
25756 * ifcvt.c (noce_process_if_block): When looking for x setter
25757 with missing else_bb, don't check only the insn right before
25758 cond_earliest, but look for the last insn that x is modified in
25759 within the same bb.
25761 PR rtl-optimization/80491
25762 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25764 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25766 PR tree-optimization/80487
25767 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25769 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25771 PR tree-optimization/79697
25772 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25773 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25774 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25776 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25777 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25779 2017-04-28 Martin Sebor <msebor@redhat.com>
25781 PR tree-optimization/80523
25782 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25783 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25785 (maybe_warn, format_directive, parse_directive): Use new functions.
25786 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25788 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25790 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25792 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25794 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25795 target_header_dir): Set correctly.
25796 * configure: Regenerated.
25797 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25798 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25799 instead of SYSTEM_HEADER_DIR.
25801 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25803 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25804 (estimate_local_effects): Likewise.
25805 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25806 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25807 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25808 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25809 do_estimate_edge_time, estimate_edge_time): Likewise.
25810 * ipa-inline-analysis.c (estimate_node_size_and_time,
25811 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25812 (estimate_time_after_inlining): Remove.
25814 2017-04-28 Martin Liska <mliska@suse.cz>
25816 * doc/gcov.texi: Enhance documentation of gcov.
25818 2017-04-28 Martin Liska <mliska@suse.cz>
25820 * doc/gcov.texi: Sort options in alphabetic order.
25821 * doc/gcov-dump.texi: Likewise.
25822 * doc/gcov-tool.texi: Likewise.
25823 * gcov.c (print_usage): Likewise.
25824 * gcov-dump.c (print_usage): Likewise.
25825 * gcov-tool.c (print_merge_usage_message): Likewise.
25826 (print_rewrite_usage_message): Likewise.
25827 (print_overlap_usage_message): Likewise.
25829 2017-04-28 Martin Liska <mliska@suse.cz>
25831 PR gcov-profile/53915
25832 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25834 2017-04-28 Martin Liska <mliska@suse.cz>
25836 PR gcov-profile/79891
25837 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25838 is marked by compiler as living on a line.
25839 (get_cycles_count): Remove usage of the union.
25840 (output_intermediate_file): Likewise.
25841 (find_source): Fix GNU coding style.
25842 (accumulate_line_counts): Remove old non-all block mode.
25843 (output_lines): Remove usage of the union.
25844 * profile.c (output_location): Include all BBs, even if
25845 belonging to a same line (and file) as a previous BB.
25847 2017-04-28 Martin Liska <mliska@suse.cz>
25849 * gcov.c (process_args): Handle new argument 'w'.
25850 (read_graph_file): Assign ID to BBs.
25851 (output_branch_count): Display BB # if verbose flag is set.
25852 (output_lines): Likewise for arcs.
25853 (print_usage): Add '--verbose' option help.
25854 * doc/gcov.texi: Document --verbose (-w) option.
25856 2017-04-28 Martin Liska <mliska@suse.cz>
25858 * gcov.c (struct block_location_info): New struct.
25859 (process_file): Fill up the new structure.
25860 (read_graph_file): Replace usage of encoding by the newly added
25862 (add_line_counts): Likewise.
25863 (accumulate_line_counts): Remove usage of the union.
25864 (function_info::function_info): New function.
25865 (function_info::~function_info): Likewise.
25866 (process_file): Call delete instead of release_function.
25867 (release_function): Release the function.
25868 (release_structures): Call delete instead of release_function.
25869 (solve_flow_graph): Replace usage of num_blocks.
25870 (find_exception_blocks): Likewise.
25871 (output_lines): Fix GNU coding style.
25873 2017-04-28 Martin Liska <mliska@suse.cz>
25876 * coverage.c (coverage_remove_note_file): New function.
25877 * coverage.h: Declare the function.
25878 * toplev.c (finalize): Clean if an error has been seen.
25880 2017-04-28 Martin Liska <mliska@suse.cz>
25882 PR gcov-profile/80031
25883 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25884 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25885 * gcov.c (read_graph_file): Read just number of blocks.
25886 * profile.c (branch_prob): Do not stream 0 flags per a basic
25889 2017-04-28 Martin Liska <mliska@suse.cz>
25891 * gcov-dump.c (tag_*): Add new argument to declarations.
25892 (dump_gcov_file): Likewise.
25893 (tag_blocks): Add and use new argument depth.
25894 (tag_arcs): Likewise.
25895 (tag_lines): Likewise.
25896 (tag_counters): Likewise.
25897 (tag_summary): Likewise.
25898 (dump_working_sets): Use depth to do a proper indentation.
25900 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25903 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25904 * symtab.c (symtab_node::debug_symtab): Move definition here.
25906 2017-04-28 Richard Biener <rguenther@suse.de>
25908 * lto-streamer.h (LTO_major_version): Bump to 7.
25910 2017-04-28 Richard Biener <rguenther@suse.de>
25912 * tree-vrp.c (assert_info): New struct.
25913 (add_assert_info): New helper.
25914 (register_edge_assert_for_2): Refactor to add asserts to a vector
25916 (register_edge_assert_for_1): Likewise.
25917 (register_edge_assert_for): Likewise.
25918 (finish_register_edge_assert_for): New helper actually registering
25919 asserts where live on edge.
25920 (find_conditional_asserts): Adjust.
25921 (find_switch_asserts): Likewise.
25922 (evrp_dom_walker::try_find_new_range): Generalize.
25923 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25925 2017-04-27 Marek Polacek <polacek@redhat.com>
25928 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25929 arg10 and arg11 to itype.
25931 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25933 * doc/extend.texi (Object Size Checking): Improve grammar.
25935 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25938 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25939 that the logic for permitting reciprocal estimates matches that
25942 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25945 * tree.c (type_cache_hasher::equal): Only compare
25946 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25947 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25948 non-aggregate element types.
25949 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25950 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25952 2017-04-27 Richard Biener <rguenther@suse.de>
25954 PR middle-end/80533
25955 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25956 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25957 keeping a reference to a trailing array.
25959 2017-04-27 Richard Biener <rguenther@suse.de>
25961 PR middle-end/80539
25962 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25963 being in loop-closed SSA form conservatively.
25964 (chrec_fold_multiply_poly_poly): Likewise.
25966 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25968 PR middle-end/79665
25969 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25970 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25972 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25975 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25976 (aarch64_function_arg_alignment): Return unsigned int again, but still
25977 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25978 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25979 Don't emit -Wpsabi note.
25980 (aarch64_function_arg_boundary): Likewise.
25981 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25984 2017-04-26 Nathan Sidwell <nathan@acm.org>
25986 * tree.h (crc32_unsigned_n): Declare.
25987 (crc32_unsigned, crc32_unsigned): Make inline.
25988 * tree.c (crc32_unsigned_bits): Replace with ...
25989 (crc32_unsigned_n): ... this.
25990 (crc32_unsigned, crc32_byte): Remove.
25991 (crc32_string): Remove unnecessary braces.
25993 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25995 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25996 * ipa-inline-analysis.c (MAX_TIME): Remove.
25997 (account_size_time): Use sreal for time.
25998 (dump_inline_summary): Update.
25999 (estimate_function_body_sizes): Update.
26000 (estimate_edge_size_and_time): Update.
26001 (estimate_calls_size_and_time): Update.
26002 (estimate_node_size_and_time): Update.
26003 (inline_merge_summary): Update.
26004 (inline_update_overall_summary): Update.
26005 (estimate_time_after_inlining): Update.
26006 (inline_read_section): Update.
26007 (inline_write_summary): Update.
26008 * ipa-inline.c (compute_uninlined_call_time): Update.
26009 (compute_inlined_call_time): Update.
26010 (recursive_inlining): Update.
26011 (inline_small_functions): Update.
26012 (dump_overall_stats): Update.
26013 * ipa-inline.h: Include sreal.h.
26014 (size_time_entry): Turn time to sreal.
26015 (inline_summary): Turn self_time nad time to sreal.
26017 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26019 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26021 (sreal::stream_out, sreal::stream_in): New.
26022 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26024 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26026 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26029 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26032 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26034 (dimode_scalar_chain::compute_convert_gain): Ditto.
26035 (dimode_scalar_chain::make_vector_copies): Ditto.
26036 (dimode_scalar_chain::convert_reg): Ditto.
26037 (dimode_scalar_chain::convert_insn): Ditto.
26038 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26039 (VI248_AVX512BW_1): New mode iterator.
26040 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26041 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26044 2017-04-25 Martin Sebor <msebor@redhat.com>
26046 PR tree-optimization/80497
26047 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26048 constants are representable in HOST_WIDE_INT.
26049 (parse_directive): Ditto.
26051 2017-04-25 Martin Sebor <msebor@redhat.com>
26054 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26055 (new_zero_array): Adjust signature.
26056 (dom_info::dom_init): Used unsigned rather that size_t.
26057 (dom_info::dom_info): Same.
26059 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26060 Jakub Jelinek <jakub@redhat.com>
26063 * config/arm/arm.c: Include gimple.h.
26064 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26065 returns negative, increment ncrn only if it returned positive.
26066 (arm_needs_doubleword_align): Return int instead of bool,
26067 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26068 members, but if there is any such non-FIELD_DECL
26069 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26070 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26071 returns negative, increment nregs only if it returned positive.
26072 (arm_setup_incoming_varargs): Likewise.
26073 (arm_function_arg_boundary): Emit -Wpsabi note if
26074 arm_needs_doubleword_align returns negative, return
26075 DOUBLEWORD_ALIGNMENT only if it returned positive.
26077 2017-04-25 Marek Polacek <polacek@redhat.com>
26080 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26081 first argument to type.
26083 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26087 type checks to test for compatibility instead of equality.
26089 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26090 Jakub Jelinek <jakub@redhat.com>
26093 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26095 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26096 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26097 the alignment computation, but return their maximum in warn_alignment.
26098 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26099 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26101 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26102 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26105 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26107 * config/arc/simdext.md (dmpyh): Fix typo.
26109 2017-04-25 Richard Biener <rguenther@suse.de>
26111 PR tree-optimization/80492
26112 * alias.c (compare_base_decls): Handle registers with asm
26113 specification conservatively.
26114 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26115 compare_base_decls returning dont-know properly.
26117 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26119 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26120 (legitimate_offset_address_p): New function.
26121 (arc_legitimate_address_p): Use above function.
26123 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26125 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26127 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26129 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26130 ACCH registers whenever they are available.
26132 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26134 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26135 double regs fix when not used.
26137 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26139 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26141 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26142 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26144 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26146 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26147 when emitting short ADD instructions.
26149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26151 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26153 (cmpsi_cc_c_insn): Likewise.
26154 (cbranchsi4_scratch): Compute proper instruction length using
26155 compact_hreg_operand.
26156 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26158 2017-04-25 Richard Biener <rguenther@suse.de>
26160 PR middle-end/80509
26161 * passes.c (pass_manager::pass_manager): Initialize
26162 m_name_to_pass_map.
26164 2017-04-25 Richard Biener <rguenther@suse.de>
26166 PR tree-optimization/79201
26167 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26169 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26172 * config/s390/vector.md: Split MEM->GPR vector moves for
26173 non-s_operand addresses.
26175 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26178 * config/s390/predicates.md (reload_const_wide_int_operand): New
26180 * config/s390/s390.md ("movti"): Remove d/P alternative.
26181 ("movti_bigconst"): New pattern definition.
26183 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26186 * s390-protos.h (s390_expand_cs_hqi): Removed.
26187 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26188 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26189 modes as well as CCZ1mode and CCZmode.
26190 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26191 signature of s390_emit_compare_and_swap.
26192 (s390_expand_cs_hqi): Likewise, make static.
26193 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26194 compare-and-swap, in some cases.
26195 (s390_expand_cs): Wrapper function.
26196 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26198 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26199 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26200 patterns for small and large integers. Forbid symref memory operands.
26201 Move expander to s390.c. Require cc register.
26202 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26203 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26204 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26205 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26206 symref memory operands. Remove CC mode and call s390_match_ccmode
26208 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26210 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26212 * config/s390/s390.md (define_peephole2): New peephole to help
26213 combining the load-and-test pattern with volatile memory.
26215 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26217 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26218 with CCZmode for TARGET_Z196.
26220 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26222 PR rtl-optimization/80501
26223 * combine.c (make_compound_operation_int): Set subreg_code to SET
26224 even for AND with mask of the sign bit of mode.
26226 PR rtl-optimization/80500
26227 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26228 sum's initial value.
26230 2017-04-25 Julian Brown <julian@codesourcery.com>
26231 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26233 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26235 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26237 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26239 2017-04-25 Julian Brown <julian@codesourcery.com>
26240 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26242 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26243 (thunderx2t99_sha): New Reservation.
26245 2017-04-25 Julian Brown <julian@codesourcery.com>
26246 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26248 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26249 type for 1-element load.
26251 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26253 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26255 2017-04-24 Martin Jambor <mjambor@suse.cz>
26257 PR tree-optimization/80293
26258 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26259 char arrays not totally scalarizable if it is false.
26260 (analyze_all_variable_accesses): Pass correct value in the new
26261 parameter. Add a statistics counter.
26263 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26265 PR middle-end/79931
26266 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26268 2017-04-24 Richard Biener <rguenther@suse.de>
26270 PR tree-optimization/80494
26271 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26272 out for complex types.
26274 2017-04-24 Richard Biener <rguenther@suse.de>
26276 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26277 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26278 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26279 (DFS): Adjust and never fail.
26280 (sccvn_dom_walker::fail): Remove.
26281 (sccvn_dom_walker::before_dom_children): Adjust.
26282 (run_scc_vn): Likewise and never fail.
26283 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26284 (pass_fre::execute): Likewise.
26286 2017-04-24 Richard Biener <rguenther@suse.de>
26288 PR tree-optimization/79725
26289 * tree-ssa-sink.c (statement_sink_location): Return whether
26290 failure reason was zero uses. Move that check later.
26291 (sink_code_in_bb): Deal with zero uses by removing the stmt
26294 2017-04-24 Richard Biener <rguenther@suse.de>
26297 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26298 pointer-based references.
26300 2017-04-24 Richard Biener <rguenther@suse.de>
26303 * pass_manager.h (pass_manager::operator new): Remove.
26304 (pass_manager::operator delete): Likewise.
26305 * passes.c (pass_manager::operator new): Remove.
26306 (pass_manager::operator delete): Likewise.
26307 (pass_manager::pass_manager): Zero individual pass members.
26309 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26312 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26313 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26314 Check "XEXP (src, 1)" operand here.
26315 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26316 Check "XEXP (src, 1)" operand here.
26317 (dimode_scalar_chain::make_vector_copies): Detect count register
26318 of a shift instruction. Zero extend count register from QImode
26319 to DImode to satisfy vector shift pattern count operand predicate.
26320 Substitute vector shift count operand with a DImode copy.
26321 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26324 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26326 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26327 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26328 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26329 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26330 (UNSPEC_NOREX_MEM): Remove definition.
26332 2017-04-21 Richard Biener <rguenther@suse.de>
26334 PR tree-optimization/79547
26335 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26336 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26337 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26338 without any constraints.
26340 2017-04-21 Richard Biener <rguenther@suse.de>
26342 PR tree-optimization/78847
26343 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26345 2017-04-21 Richard Biener <rguenther@suse.de>
26347 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26348 (build_distinct_type_copy): Likewise.
26349 (build_variant_type_copy): Likewise.
26350 * tree.c (build_qualified_type): Pass down mem-stat info.
26351 (build_distinct_type_copy): Likewise.
26352 (build_variant_type_copy): Likewise.
26354 2017-04-21 Richard Biener <rguenther@suse.de>
26356 PR tree-optimization/80237
26357 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26359 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26360 for a simplified result.
26362 2016-04-21 Richard Biener <rguenther@suse.de>
26364 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26365 sth as strict as a simple_iv but a chrec without symbols and an
26366 operand defined in the loop we are peeling (and not some subloop).
26367 (propagate_constants_for_unrolling): Propagate all constants.
26369 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26372 * config/i386/i386.c (print_reg): Remove assert for disalowed
26373 regno values, call output_operand_lossage instead.
26375 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26378 * config/i386/constraints.md (Yc): New register constraint.
26379 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26380 Use Yc constraint for alternative 2 of operand 0. Remove
26381 preferred_for_speed attribute.
26383 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26385 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26386 lastprivate clauses in SIMT case.
26388 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26390 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26392 2017-04-20 Richard Biener <rguenther@suse.de>
26394 PR tree-optimization/57796
26395 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26396 as N scalar stores.
26397 (vect_model_load_cost): Cost gathers as N scalar loads.
26399 2017-04-20 Richard Biener <rguenther@suse.de>
26401 * ggc-page.c (ggc_allocated_p): Rename to ...
26402 (safe_lookup_page_table_entry): ... this and return the lookup
26404 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26406 2017-04-20 Richard Biener <rguenther@suse.de>
26408 PR tree-optimization/80453
26409 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26410 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26411 from the conditions.
26412 (vn_phi_eq): Pass them down.
26413 (vn_phi_lookup): Record them.
26414 (vn_phi_insert): Likewise.
26416 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26418 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26419 uninitialized variable warning to avoid buffer overrun.
26421 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26424 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26425 is suppressed for '{ 0 }' in C.
26427 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26429 * BASE-VER: Set to 8.0.0.
26431 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26433 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26434 priority .init_array and .fini_array section with SECTION_NOTYPE
26437 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26439 PR middle-end/80423
26440 * tree.h (build_array_type): Add typeless_storage default argument.
26441 * tree.c (type_cache_hasher::equal): Also compare
26442 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26443 (build_array_type): Add typeless_storage argument, set
26444 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26446 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26447 (build_array_type): Likewise. Add typeless_storage argument.
26449 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26450 Jakub Jelinek <jakub@redhat.com>
26452 PR tree-optimization/80426
26453 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26454 operation on symbolic operands, also compute the overflow for the
26455 invariant part when the operation degenerates into a negation.
26457 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26460 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26461 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26464 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26466 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26469 * config/avr/avr.c (tree.h): Include it.
26470 (cgraph.h): Include it.
26471 (avr_encode_section_info): Don't warn for uninitialized progmem
26472 variable if it's just an alias.
26474 2017-04-19 Richard Biener <rguenther@suse.de>
26477 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26478 when needed by AutoPGO.
26480 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26483 * doc/lto.texi: Remove an extra 'that'.
26485 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26487 PR rtl-optimization/80429
26488 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26489 are only used in debug insns.
26491 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26492 Vladimir Makarov <vmakarov@redhat.com>
26494 * config/sparc/predicates.md (input_operand): Add comment. Return
26495 true for any memory operand when LRA is in progress.
26496 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26498 2017-04-18 Jeff Law <law@redhat.com>
26501 * mips.md ({return,simple_return}_internal): Do not overwrite
26504 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26506 PR tree-optimization/80443
26507 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26508 instead of adding 1, subtract -1 and similarly instead of subtracting
26511 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26513 PR rtl-optimization/80357
26514 * haifa-sched.c (tmp_bitmap): New variable.
26515 (model_recompute): Handle duplicate use records.
26516 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26517 (free_global_sched_pressure_data): Free it.
26519 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26522 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26523 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26524 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26525 instead of SYSTEM_HEADER_DIR.
26527 2017-04-18 Jeff Law <law@redhat.com>
26529 PR middle-end/80422
26530 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26531 predecessors after walking up the insn chain.
26533 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26536 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26537 sizetype type into debug info.
26539 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26542 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26543 unneeded test for TARGET_UPPER_REGS_SF.
26544 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26546 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26549 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26550 instead of gsi_after_labels.
26552 2017-04-18 Jeff Law <law@redhat.com>
26554 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26558 2017-04-13 Jeff Law <law@redhat.com>
26559 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26560 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26562 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26565 * config/avr/avr.c (intl.h): Include it.
26566 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26568 2017-04-18 Martin Liska <mliska@suse.cz>
26570 PR gcov-profile/78783
26571 * gcov-tool.c (gcov_output_files): Validate that destination
26572 file is either removed by the tool or by a user.
26574 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26575 Guy Benyei <guybe@mellanox.com>
26577 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26578 block, and do not negate it, the stored id is already negative.
26580 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26582 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26584 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26587 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26588 masks of options that should be turned off if the VSX vector
26589 options are turned off.
26590 (OTHER_P8_VECTOR_MASKS): Likewise.
26591 (OTHER_VSX_VECTOR_MASKS): Likewise.
26592 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26593 rs6000_disable_incompatible_switches to validate no type switches
26595 (rs6000_incompatible_switch): New function to disallow turning on
26596 other vector options if -mno-vsx, -mno-power8-vector, or
26597 -mno-power9-vector are specified.
26599 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26601 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26603 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26605 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26606 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26607 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26608 (ARG_POINTER_CFA_OFFSET): Likewise.
26610 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26612 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26613 conditions to take advantage of various optimizations.
26615 2017-04-13 Jeff Law <law@redhat.com>
26617 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26618 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26619 (zero_extendsidi2_dext): Likewise.
26621 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26624 * fold-const.c (fold_ternary_loc): Revert
26625 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26628 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26630 PR rtl-optimization/80343
26631 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26632 new scratch pseudo.
26634 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26637 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26638 to ubsan_encode_value.
26640 2017-04-13 Jeff Law <law@redhat.com>
26642 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26643 appearing in DEBUG_INSNs.
26645 2017-04-13 Martin Liska <mliska@suse.cz>
26647 PR gcov-profile/80413
26648 * gcov-io.c (gcov_write_string): Copy to buffer just when
26649 allocated size is greater than zero.
26651 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26654 * dwarf2out.c (decls_for_scope): Ignore declarations of
26655 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26657 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26660 * ipa-visibility.c (non_local_p): Fix typos.
26661 (localize_node): When localizing symbol in same comdat group,
26662 dissolve the group only when we know external symbols are going
26664 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26666 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26668 PR tree-optimization/79390
26669 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26670 order does not result in usable sequence, retry with reversed operand
26676 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26677 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26678 op0 instead of fold_convert_loc (loc, type, arg0).
26680 2017-04-12 Jeff Law <law@redhat.com>
26682 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26683 has a delay slot in the generated code.
26685 * config/cris/cris.md (cris_preferred_reload_class): Return
26686 GENNONACR_REGS rather than GENERAL_REGS.
26688 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26691 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26692 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26693 signedness of the result type.
26695 2017-04-12 Richard Biener <rguenther@suse.de>
26696 Jeff Law <law@redhat.com>
26698 PR tree-optimization/80359
26699 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26700 trim stores to TARGET_MEM_REFs.
26702 2017-04-12 Richard Biener <rguenther@suse.de>
26704 PR tree-optimization/79390
26705 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26706 threading case even more.
26708 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26711 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26712 for quad_address_p for TImode, instead of just not indexed_address.
26714 2017-04-12 Richard Biener <rguenther@suse.de>
26715 Bernd Edlinger <bernd.edlinger@hotmail.de>
26717 PR middle-end/79671
26718 * alias.c (component_uses_parent_alias_set_from): Handle
26719 TYPE_TYPELESS_STORAGE.
26720 (get_alias_set): Likewise.
26721 * tree-core.h (tree_type_common): Add typeless_storage flag.
26722 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26723 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26724 for types containing members with TYPE_TYPELESS_STORAGE.
26725 (place_field): Likewise.
26726 (layout_type): Likewise for ARRAY_TYPE.
26727 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26728 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26729 TYPE_TYPELESS_STORAGE.
26730 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26732 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26735 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26736 first argument to type.
26738 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26742 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26743 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26744 (rs6000_expand_binop_builtin): Likewise.
26745 (rs6000_expand_ternop_builtin): Likewise; also add missing
26746 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26747 vshasigma built-ins.
26748 * doc/extend.texi: Document that vec_xxpermdi's third argument
26749 must be a constant.
26751 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26753 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26754 Use shift_const cost parameter when calculating gain of STV shifts.
26756 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26758 PR rtl-optimization/70478
26759 * lra-constraints.c (process_alt_operands): Check memory for
26760 disfavoring memory insn operand.
26762 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26764 PR middle-end/80100
26765 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26766 left shift in unsigned HOST_WIDE_INT type.
26768 PR rtl-optimization/80385
26769 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26770 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26773 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26774 if they have any depend clauses.
26776 2017-04-11 Martin Liska <mliska@suse.cz>
26779 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26780 * ipa-split.c (split_function): Create a local comdat symbol
26781 if caller is in a comdat group.
26783 2017-04-11 Martin Liska <mliska@suse.cz>
26786 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26789 2017-04-11 Martin Sebor <msebor@redhat.com>
26791 PR middle-end/80364
26792 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26793 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26795 (directive::set_width, directive::set_precision, format_character):
26797 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26800 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26803 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26804 conflict, set target->arch_name instead of target->cpu_name.
26806 2017-04-11 Richard Biener <rguenther@suse.de>
26808 PR tree-optimization/80374
26809 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26810 build_zero_cst, remove fold_convertible_p check again.
26812 2017-04-11 Martin Liska <mliska@suse.cz>
26815 * ubsan.c (instrument_object_size): Do not instrument register
26818 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26821 * config/i386/i386-builtin-types.def
26822 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26823 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26824 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26825 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26826 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26827 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26828 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26829 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26830 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26831 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26832 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26833 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26834 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26835 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26836 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26837 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26838 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26839 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26840 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26841 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26842 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26843 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26844 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26845 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26846 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26847 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26848 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26849 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26850 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26851 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26852 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26853 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26854 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26855 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26856 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26857 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26858 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26859 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26860 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26861 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26862 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26863 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26864 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26865 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26866 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26867 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26869 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26870 flag to second_arg_count, handle 4 argument function type _COUNT
26871 aliases, handle second_arg_count on second argument rather than last.
26873 2017-04-10 Jeff Law <law@redhat.com>
26875 PR tree-optimization/80374
26876 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26877 record anything if we can not convert integer_zero_node to the
26880 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26883 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26884 Enhance special handling given to the TARGET_P9_MINMAX option in
26885 relation to certain other options.
26887 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26889 PR tree-optimization/80153
26890 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26891 remove POINTER_PLUS_EXPR's base part directly, rather than through
26894 2017-04-10 Richard Biener <rguenther@suse.de>
26895 Bin Cheng <bin.cheng@arm.com>
26897 PR tree-optimization/80153
26898 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26899 the first element of pointer type aff_tree. Build result expr in
26901 (add_elt_to_tree): Convert to type unconditionally. Remove other
26902 fold_convert calls.
26903 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26904 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26906 2017-04-10 Richard Biener <rguenther@suse.de>
26908 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26911 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26913 PR rtl-optimization/70478
26914 * lra-constraints.c (curr_small_class_check): New.
26915 (update_and_check_small_class_inputs): New.
26916 (process_alt_operands): Update curr_small_class_check. Disfavor
26917 alternative insn memory operands. Check available regs for small
26920 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26923 * config/mips/mips.opt (-mvirt): Update description.
26924 * doc/invoke.texi (-mvirt): Likewise.
26926 2017-04-10 Richard Biener <rguenther@suse.de>
26928 PR middle-end/80362
26929 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26930 looking for NEGATE_EXPR in -A / -B to A / B folding.
26932 2017-04-10 Martin Liska <mliska@suse.cz>
26934 PR gcov-profile/80224
26935 * gcov.c (print_usage): Fix usage string.
26936 (get_gcov_intermediate_filename): Remove.
26937 (output_gcov_file): Use both for normal and intermediate format.
26938 (generate_results): Do not initialize special file for
26939 intermediate format.
26941 2017-04-10 Richard Biener <rguenther@suse.de>
26943 PR tree-optimization/80304
26944 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26947 2017-04-10 Nathan Sidwell <nathan@acm.org>
26950 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26951 (rs6000_init_builtins): Use it.
26953 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26955 * config/arm/arm.md (<mrc>): Add mode to SET source.
26956 (<mrrc>): Likewise.
26958 2017-04-10 Richard Biener <rguenther@suse.de>
26960 PR middle-end/80344
26961 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26963 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26966 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26967 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26968 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26969 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26970 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26971 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26972 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26973 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26974 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26975 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26976 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26977 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26978 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26979 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26980 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26981 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26982 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26983 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26984 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26985 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26986 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26987 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26988 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26990 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26992 PR rtl-optimization/70478
26993 * lra-constraints.c: Reverse the last patch.
26995 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26997 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26998 Add comment for WCHAR_T.
27000 2017-04-08 Martin Liska <mliska@suse.cz>
27003 2017-04-07 Martin Liska <mliska@suse.cz>
27006 * ipa-split.c (split_function): Add function part to a same comdat
27009 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27012 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27014 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27016 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27017 (vec_pairing): ...to this.
27018 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27019 (rs6000_sched_init): Adjust for name change.
27020 (struct rs6000_sched_context): Likewise.
27021 (rs6000_init_sched_context): Likewise.
27022 (rs6000_set_sched_context): Likewise.
27024 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27030 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27032 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27033 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27034 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27036 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27038 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27040 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27042 PR rtl-optimization/70703
27043 * ira-color.c (update_conflict_hard_regno_costs): Use
27044 int64_t instead of HOST_WIDE_INT.
27046 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27048 PR rtl-optimization/70478
27049 * lra-constraints.c (process_alt_operands): Disfavor alternative
27050 insn memory operands.
27052 2017-04-07 Jeff Law <law@redhat.com>
27054 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27055 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27057 2017-04-07 Martin Liska <mliska@suse.cz>
27060 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27061 Show error message instead of an ICE.
27063 2017-04-07 Martin Liska <mliska@suse.cz>
27066 * ipa-split.c (split_function): Add function part to a same comdat
27069 2017-04-07 Richard Biener <rguenther@suse.de>
27071 PR middle-end/80341
27072 * tree.c (get_unwidened): Also handle ! for_type case for
27074 * convert.c (do_narrow): Split out from ...
27075 (convert_to_integer_1): ... here. Do not pass final truncation
27076 type to get_unwidened for TRUNC_DIV_EXPR.
27078 2017-04-07 Richard Biener <rguenther@suse.de>
27080 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27082 (aff_combination_const): Adjust.
27083 (aff_combination_scale): Likewise.
27084 (aff_combination_add_elt): Likewise.
27085 (aff_combination_add_cst): Likewise.
27086 (aff_combination_convert): Likewise.
27087 (add_elt_to_tree): Likewise. Remove unused argument.
27088 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27090 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27092 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27094 * config/arm/arm.c (arm_default_short_enums): Use
27095 ARM_DEFAULT_SHORT_ENUMS.
27096 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27098 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27101 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27102 members with redundant out-of-class redeclaration.
27104 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27107 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27108 * config/i386/i386.md (*zero_extendsidi2):
27109 Add (?*x,*x) and (?*v,*v) alternatives.
27111 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27114 * config/i386/i386.c (ix86_expand_builtin)
27115 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27116 mode from insn data. Convert operands to insn operand mode.
27117 Copy operands that don't satisfy insn predicate to a register.
27119 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27121 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27124 2017-04-06 Richard Biener <rguenther@suse.de>
27126 PR tree-optimization/80334
27127 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27128 preserve alignment of accesses.
27130 2017-04-06 Richard Biener <rguenther@suse.de>
27132 PR tree-optimization/80262
27133 * tree-sra.c (build_ref_for_offset): Preserve address-space
27135 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27136 Drop useless address-space information on MEM_REF offsets.
27138 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27140 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27142 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27144 PR rtl-optimization/70703
27145 * ira-color.c (update_conflict_hard_regno_costs): Use
27146 HOST_WIDE_INT instead of long.
27148 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27151 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27152 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27154 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27155 for x86_64 target. Handle -m3dnowa option.
27157 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27159 PR rtl-optimization/70703
27160 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27161 (update_conflict_hard_regno_costs): Use long instead of unsigned
27162 arithmetic for cost calculation.
27164 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27165 Bernd Edlinger <bernd.edlinger@hotmail.de>
27168 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27171 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27174 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27175 ptr_mode with Pmode throughout.
27176 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27177 into probe_stack_range and use DImode.
27179 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27182 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27183 call_eh_return is true.
27185 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27187 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27188 Initialize last_match_fntype_index.
27190 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27193 * tree-nvr.c: Include internal-fn.h.
27194 (pass_return_slot::execute): Ignore internal calls without
27197 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27198 Richard Biener <rguenther@suse.de>
27201 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27202 captures used multiple times, except for the last use.
27203 * generic-match-head.c: Include gimplify.h.
27205 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27207 PR tree-optimization/79390
27208 * target.h (struct noce_if_info): Declare.
27209 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27210 * target.def (noce_conversion_profitable_p): New target hook.
27211 * ifcvt.h (struct noce_if_info): New type, moved from ...
27212 * ifcvt.c (struct noce_if_info): ... here.
27213 (noce_conversion_profitable_p): Renamed to ...
27214 (default_noce_conversion_profitable_p): ... this. No longer
27216 (noce_try_store_flag_constants, noce_try_addcc,
27217 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27218 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27219 instead of noce_conversion_profitable_p.
27220 * config/i386/i386.c: Include ifcvt.h.
27221 (ix86_option_override_internal): Don't override
27222 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27223 (ix86_noce_conversion_profitable_p): New function.
27224 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27225 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27226 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27227 * doc/tm.texi: Regenerated.
27229 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27231 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27234 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27237 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27238 instructions for small multiply cores.
27240 2017-04-04 Jeff Law <law@redhat.com>
27242 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27244 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27245 that are not set by the loop over the elements.
27247 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27250 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27251 int mode, convert_modes it to mode as unsigned, otherwise use
27252 lowpart_subreg to mode rather than SImode.
27253 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27254 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27255 Use DImode instead of SImode for the shift count operand.
27256 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27259 2017-04-04 Richard Biener <rguenther@suse.de>
27261 PR middle-end/80281
27262 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27263 arithmetic done for the negate or the plus. Simplify.
27264 (A - (-B) -> A + B): Likewise.
27265 * fold-const.c (split_tree): Make sure to not negate pointers.
27267 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27269 PR rtl-optimization/60818
27270 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27271 a compare of comparisons with the thing compared if this results
27272 in a different machine mode.
27274 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27276 * alias.c (base_alias_check): Fix typo in comment.
27277 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27278 * cgraphunit.c (symbol_table::compile): Likewise.
27279 * collect2.c (maybe_run_lto_and_relink): Likewise.
27280 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27281 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27282 * config/avr/avr.c (avr_map_op_t): Likewise.
27283 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27284 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27285 * config/epiphany/epiphany.md (movcc): Likewise.
27286 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27287 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27289 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27290 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27291 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27292 * config/sh/sh.c (sh_rtx_costs): Likewise.
27293 * fold-const.c (fold_truth_andor): Likewise.
27294 * genautomata.c (collapse_flag): Likewise.
27295 * gengtype.h (struct type::u::s): Likewise.
27296 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27297 * input.c (FORMAT_AMOUNT): Likewise.
27298 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27299 (known_aggs_to_agg_replacement_list): Likewise.
27300 * ipa-inline-analysis.c: Likewise.
27301 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27302 * ipa-polymorphic-call.c
27303 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27304 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27305 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27307 * modulo-sched.c (apply_reg_moves): Likewise.
27308 * omp-expand.c (build_omp_regions_1): Likewise.
27309 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27310 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27311 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27312 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27313 * value-prof.c: Likewise.
27314 * var-tracking.c (val_reset): Likewise.
27316 2017-04-03 Richard Biener <rguenther@suse.de>
27318 PR tree-optimization/80275
27319 * fold-const.c (split_address_to_core_and_offset): Handle
27322 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27324 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27325 descriptors is at least equal to that of functions.
27327 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27329 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27331 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27334 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27335 (mov<IMOD4:mode>): New expander.
27336 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27338 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27340 PR rtl-optimization/79405
27341 * fwprop.c (propagations_left): New variable.
27342 (forward_propagate_into): Decrement it.
27343 (fwprop_init): Initialize it.
27344 (fw_prop): If the variable has reached zero, stop propagating.
27345 (fwprop_addr): Ditto.
27347 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27350 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27351 a FUNCTION_DECL, pass it as decl instead of origin to
27354 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27356 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27359 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27362 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27363 TARGET_VSX_SMALL_INTEGER.
27365 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27367 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27368 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27370 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27372 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27373 extraction from odd-numbered MSA register.
27375 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27377 PR middle-end/80173
27378 * expmed.c (store_bit_field_1): Don't attempt to create
27379 a word subreg out of hard registers wider than word if they
27380 have HARD_REGNO_NREGS of 1 for their mode.
27382 PR middle-end/80163
27383 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27384 conversions to integer types wider than word and pointer.
27387 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27388 (rtx_equal_for_cselib_p): Pass 0 to it.
27389 * cselib.c (cselib_hasher::equal): Likewise.
27390 (rtx_equal_for_cselib_1): Add depth argument. If depth
27391 is 128, don't look up VALUE locs and punt. Increment
27392 depth in recursive calls when walking VALUE locs.
27394 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27396 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27397 (make_gcov_file_name): Use the canonical path name for generating
27399 (read_line): Fix handling of files with ascii null bytes.
27401 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27403 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27404 to initialise a vector register instead
27405 of using a const_int.
27407 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27409 PR translation/80189
27410 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27411 diagnostic messages.
27413 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27416 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27417 (dfp_diex_<mode>): Update mode of operand 1.
27418 * doc/extend.texi (dxex, dxexq): Document change to return type.
27419 (diex, diexq): Document change to argument type.
27421 2017-03-30 Martin Jambor <mjambor@suse.cz>
27424 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27425 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27426 it reflects the signature changes performed at the callee side.
27427 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27428 to cgraph_build_function_type_skip_args.
27429 (build_function_decl_skip_args): Adjust call to the above function.
27431 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27434 * config/i386/sse.md
27435 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27436 register as dest whenever it is a MEM not rtx_equal_p to the
27437 corresponding dup operand, and when forcing into reg move the
27438 reg into the memory afterwards.
27439 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27440 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27441 for the force_reg mode.
27442 (avx512vl_vextractf128<mode>): Use register as dest either
27443 always when a MEM, or when it is a MEM not rtx_equal_p to the
27444 corresponding dup operand, or even not when it is a CONST_VECTOR
27445 depending on the mode and lo vs. hi.
27446 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27448 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27449 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27450 Likewise. Require that operands[2] is even.
27451 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27452 Remove extraneous parens. Require that operands[2] is a multiple
27454 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27455 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27456 disallow memory then.
27458 2017-03-30 Richard Biener <rguenther@suse.de>
27460 PR tree-optimization/77498
27461 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27462 to non-constants over backedges.
27464 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27466 PR rtl-optimization/80233
27467 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27468 as last_combined_insn. Do not test for BARRIER_P separately.
27470 2017-03-29 Andreas Schwab <schwab@suse.de>
27473 * calls.c (prepare_call_address): Convert funexp to Pmode before
27474 copying to temp reg.
27476 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27478 PR tree-optimization/80158
27479 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27480 Handle possible future case of more than one alternate
27482 (replace_rhs_if_not_dup): Likewise.
27483 (replace_one_candidate): Likewise.
27485 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27487 PR rtl-optimization/80193
27488 * ira.c (ira): Do not check allocation for LRA.
27490 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27492 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27493 (nvptx_output_simt_exit): Declare.
27494 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27495 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27496 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27497 (nvptx_declare_function_name): ...here. Emit declaration of local
27498 memory space buffer for omp_simt_enter insn.
27499 (nvptx_output_unisimt_switch): New.
27500 (nvptx_output_softstack_switch): New.
27501 (nvptx_output_simt_enter): New.
27502 (nvptx_output_simt_exit): New.
27503 * config/nvptx/nvptx.h (struct machine_function): New fields
27504 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27505 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27506 (UNSPECV_SIMT_EXIT): Ditto.
27507 (omp_simt_enter_insn): New insn.
27508 (omp_simt_enter): New expansion.
27509 (omp_simt_exit): New insn.
27510 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27512 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27513 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27514 (expand_GOMP_SIMT_EXIT): New.
27515 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27516 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27517 (GOMP_SIMT_EXIT): Ditto.
27518 * target-insns.def (omp_simt_enter): New insn.
27519 (omp_simt_exit): Ditto.
27520 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27522 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27523 (lower_rec_input_clauses): Likewise.
27524 (lower_lastprivate_clauses): Handle SIMT privatization.
27526 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27527 (ompdevlow_adjust_simt_enter): New.
27528 (find_simtpriv_var_op): New.
27529 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27530 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27532 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27533 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27534 (copy_decl_for_dup_finish): Ditto.
27536 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27538 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27541 * config/i386/i386.c (ix86_option_override_internal): Always
27542 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27544 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27546 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27548 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27550 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27551 mark new edge's irreducible flag accordign to it.
27552 (vect_do_peeling): Check loop preheader edge's irreducible flag
27553 and pass it to function slpeel_add_loop_guard.
27555 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27557 PR tree-optimization/80218
27558 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27559 Update block frequencies and counts.
27561 2017-03-28 Richard Biener <rguenther@suse.de>
27563 PR tree-optimization/78644
27564 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27565 of a simplification result we may not use it at all.
27567 2017-03-28 Richard Biener <rguenther@suse.de>
27570 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27571 without arguments, generate default definition of a SSA name.
27573 2017-03-28 Richard Biener <rguenther@suse.de>
27575 PR middle-end/80222
27576 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27577 TYPE_REF_CAN_ALIAS_ALL references.
27578 * fold-const.c (fold_indirect_ref_1): Likewise.
27580 2017-03-28 Martin Liska <mliska@suse.cz>
27583 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27584 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27586 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27587 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27589 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27590 (EXTRA_SPECS): Define.
27591 (SUBTARGET_EXTRA_SPECS): Likewise.
27592 (SUBTARGET_CPP_SPEC): Likewise.
27593 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27594 SUBTARGET_EXTRA_SPECS.
27595 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27597 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27599 * config/arc/simdext.md (vst64_insn): Update pattern.
27600 (vld32wh_insn): Likewise.
27601 (vld32wl_insn): Likewise.
27602 (vld64_insn): Likewise.
27603 (vld32_insn): Likewise.
27605 2017-03-28 Marek Polacek <polacek@redhat.com>
27608 * fold-const.c (fold_comparison): Use protected_set_expr_location
27609 instead of SET_EXPR_LOCATION.
27611 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27613 * tree.c (add_expr): Avoid name lookup warning.
27615 2017-03-27 Jeff Law <law@redhat.com>
27617 PR tree-optimization/80216
27618 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27619 function name. Limit recursion depth.
27620 (record_temporary_equivalences): Corresponding changes.
27622 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27624 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27627 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27630 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27632 * cfgcleanup.c (reg_note_cfa_p): New array.
27633 (insns_have_identical_cfa_notes): New function.
27634 (old_insns_match_p): Don't cross-jump in between /f
27635 and non-/f instructions. If both i1 and i2 are frame related,
27636 verify all CFA notes, their order and content.
27638 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27641 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27642 HImode and SImode with zero extend to DImode to one insn.
27643 (bswap<mode>2_extenddi): Likewise.
27644 (bswapsi2_extenddi): Likewise.
27645 (bswaphi2_extendsi): Likewise.
27646 (bswaphi2): Combine bswap HImode and SImode into one insn.
27647 Separate memory insns from swapping register.
27648 (bswapsi2): Likewise.
27649 (bswap<mode>2): Likewise.
27650 (bswaphi2_internal): Delete, no longer used.
27651 (bswapsi2_internal): Likewise.
27652 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27653 store, and gpr<-gpr swap insns.
27654 (bswap<mode>2_store): Likewise.
27655 (bswaphi2_reg): Register only splitter, combine with the splitter.
27656 (bswaphi2 splitter): Likewise.
27657 (bswapsi2_reg): Likewise.
27658 (bswapsi2 splitter): Likewise.
27659 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27660 the insns into load, store, and register/register insns.
27661 (bswapdi2_ldbrx): Likewise.
27662 (bswapdi2_load): Likewise.
27663 (bswapdi2_store): Likewise.
27664 (bswapdi2_reg): Likewise.
27666 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27668 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27669 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27671 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27674 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27677 special handling for target option conflicts between dform
27678 options (-mpower9-dform, -mpower9-dform-vector,
27679 -mpower9-dform-scalar) and -mno-direct-move.
27681 2017-03-27 Richard Biener <rguenther@suse.de>
27683 PR tree-optimization/80181
27684 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27686 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27688 * config/arc/predicates.md (move_double_src_operand): Replace the
27689 call to move_double_src_operand with a call to address_operand.
27691 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27693 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27694 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27695 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27697 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27699 * config/arc/predicates.md (long_immediate_loadstore_operand):
27700 Consider scaled addresses cases.
27702 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27704 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27705 restored when in interrupt.
27706 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27707 doesn't have delay slot.
27709 2017-03-27 Richard Biener <rguenther@suse.de>
27712 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27713 inlined thunk clones.
27715 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27718 * asan.c (instrument_derefs): Copy over last operand from
27719 original COMPONENT_REF to the new COMPONENT_REF with
27720 DECL_BIT_FIELD_REPRESENTATIVE.
27721 * ubsan.c (instrument_object_size): Likewise.
27723 2017-03-27 Richard Biener <rguenther@suse.de>
27725 PR tree-optimization/80170
27726 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27727 sure DR/SCEV didnt fold in constants we do not see when looking
27728 at the reference base alignment.
27730 2017-03-27 Richard Biener <rguenther@suse.de>
27732 PR middle-end/80171
27733 * gimple-fold.c (fold_ctor_reference): Properly guard against
27734 NULL return value from canonicalize_constructor_val.
27736 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27739 * config/i386/i386.c (ix86_expand_builtin)
27740 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27741 flags reg setting and flags reg using instructions.
27742 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27743 clobbering instructions to zero extend op2.
27745 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27747 * doc/install.texi (Configuration) <--with-aix-soname>:
27748 Update link to AIX ld.
27750 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27752 PR rtl-optimization/80160
27753 PR rtl-optimization/80159
27754 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27755 reg_alternate_class into account.
27757 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27760 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27761 to consider in curr_insn_transform.
27763 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27765 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27766 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27767 and emit_mode_inner.
27769 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27771 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27772 argument to the overloaded builtin variants. Use the new flag to
27773 deprecate certain builtin variants.
27774 * config/s390/s390-builtin-types.def: Add new builtin types.
27775 * config/s390/s390-builtins.h: Support new flags field for
27776 overloaded builtins.
27777 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27778 (s390_macro_to_expand): Enable vector float data type.
27779 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27780 builtins by incrementing the __VEC__ version number.
27781 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27783 (s390_resolve_overloaded_builtin): Emit error messages depending
27784 on the builtin flags.
27785 * config/s390/s390.c (s390_expand_builtin): Support additional
27786 flags argument. Change error message to match the messages
27787 emitted in s390-c.c.
27788 * config/s390/s390.md: New UNSPEC_* constants.
27789 (op_type): Add new instruction types.
27790 * config/s390/vecintrin.h: Add new builtins and test data class
27792 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27793 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27794 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27795 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27796 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27797 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27799 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27800 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27801 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27802 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27804 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27805 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27806 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27807 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27809 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27810 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27811 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27812 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27813 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27814 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27815 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27817 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27818 ("vec_scatter_element<V_HW_4:mode>_DI")
27819 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27820 ("vec_fpint<mode>", "vflls")
27821 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27822 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27823 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27824 ("*vec_cmphe<mode>_cc"): ... these.
27826 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27827 mode constant instead of magic value.
27829 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27831 * config/s390/s390.c (s390_expand_vec_compare): Support other
27832 vector floating point modes than just V2DF.
27833 (s390_expand_vcond): Likewise.
27834 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27835 (s390_cannot_change_mode_class): Prevent mode changes between TF
27836 and V1TF in vector registers.
27837 * config/s390/s390.md (DF, SF): New mode attributes.
27838 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27839 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27840 SFmode support for VRs.
27841 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27843 (VFT, VF_HW): New mode iterators.
27844 (vw, sdx): New mode attributes.
27845 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27846 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27847 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27848 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27849 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27850 also the new vector floating point modes. Renaming to ...
27852 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27853 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27854 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27855 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27856 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27857 ("vec_unordered<mode>"): ... these.
27859 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27860 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27861 ("*vec_extendv2df"): New insn definitions.
27863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27865 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27866 ("mulditi3_2", "*muldi3_sign"): New patterns.
27867 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27868 rename the pattern definition.
27870 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27872 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27874 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27876 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27878 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27879 instruction if possible.
27880 * config/s390/vector.md (vec_halfnumelts): New mode
27882 ("*vec_vllezlf<mode>"): New pattern.
27884 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27886 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27887 ("popcountv4si2", "popcountv2di2"): Rename to ...
27888 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27889 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27891 ("popcount<mode>2_vxe"): New pattern.
27893 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27895 * common/config/s390/s390-common.c (processor_flags_table): Add
27897 * config.gcc: Add arch12.
27898 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27899 Default to arch12 for unknown CPU model numbers.
27900 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27901 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27902 PROCESSOR_max sanity check.
27903 * config/s390/s390-opts.h (enum processor_type): Add
27905 * config/s390/s390.c (processor_table): Add arch12.
27906 (s390_expand_builtin): Add check for B_VXE flag.
27907 (s390_issue_rate): Add PROCESSOR_ARCH12.
27908 (s390_get_sched_attrmask): Likewise.
27909 (s390_get_unit_mask): Likewise.
27910 (s390_sched_score): Enable z13 scheduling for arch12.
27911 (s390_sched_reorder): Likewise.
27912 (s390_sched_variable_issue): Likewise.
27913 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27915 (s390_tune_attr): Use z13 scheduling also for arch12.
27916 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27917 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27918 (TARGET_VXE_P): New macros.
27919 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27920 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27921 * config/s390/s390.opt: Add arch12 as processor_type.
27923 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27925 * config/s390/s390.md
27926 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27927 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27928 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27930 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27931 Rename expanders to ...
27933 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27934 ("fixuns_truncdddi2_emu"): ... these.
27936 ("fixuns_trunc<mode>si2_emu"): New expander.
27938 ("*fixuns_truncdfdi2_z13"): Rename to ...
27939 ("*fixuns_truncdfdi2_vx"): ... this.
27941 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27943 * config/s390/2964.md: Remove the single element vector compare
27944 instructions which are no longer used.
27945 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27947 (s390_canonicalize_comparison): Remove handling of DFmode
27949 (s390_expand_vec_compare_scalar): Remove function.
27950 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27951 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27953 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27955 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27957 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27959 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27960 will anyway by matched by mov<mode>_64dfp.
27962 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27964 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27965 vlef/vstef. Add missing operand to vleif.
27967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27969 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27970 pair for all vector types with 64 bit elements.
27971 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27972 * config/s390/vector.md (V_HW_64): ... here.
27973 (V_128_NOSINGLE): New mode iterator.
27974 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27975 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27976 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27977 ("*vec_load_pairv2di"): Change to ...
27978 ("*vec_load_pair<mode>"): ... this one.
27980 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27982 * config/s390/constraints.md: Add comments.
27983 (jKK): Reject element sizes > 8 bytes.
27984 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27986 * config/s390/s390.md: Add the s_operand checks formerly in
27987 s390_split_ok_p to various splitters where they are still
27989 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27990 for 128 bit vectors. Plus two splitters.
27992 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27994 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28000 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28001 error if the boundary argument is not constant.
28003 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28005 PR rtl-optimization/80112
28006 * loop-doloop.c (doloop_condition_get): Don't check condition
28007 if cmp isn't SET with IF_THEN_ELSE src.
28009 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28011 PR tree-optimization/80158
28012 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28013 replacing a candidate statement, also replace it for the
28014 candidate's alternate interpretation.
28015 (replace_rhs_if_not_dup): Likewise.
28016 (replace_one_candidate): Likewise.
28018 2017-03-24 Richard Biener <rguenther@suse.de>
28020 PR tree-optimization/80167
28021 * graphite-isl-ast-to-gimple.c
28022 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28024 (translate_isl_ast_to_gimple::get_rename): Likewise.
28026 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28028 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28029 handling of certain combinations of target options, including the
28030 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28031 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28033 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28036 * config/arm/arm.md (*load_multiple): Add reload_completed to
28037 matching condition.
28039 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28040 Richard Biener <rguenth@suse.de>
28042 PR tree-optimization/79908
28043 PR tree-optimization/80136
28044 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28045 been cast away, gimplify_and_add suffices.
28047 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28049 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28051 2017-03-23 Richard Biener <rguenther@suse.de>
28053 PR tree-optimization/80032
28054 * gimplify.c (gimple_push_cleanup): Forced unconditional
28055 cleanups still have to go to the conditional_cleanups
28058 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28060 PR tree-optimization/80072
28061 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28063 (next_operand_entry_id): Change type to unsigned int.
28064 (sort_by_operand_rank): Make sure to return the right return value
28065 even if unsigned fields are bigger than INT_MAX.
28066 (struct oecount): Change cnt and id type to unsigned int.
28067 (oecount_hasher::equal): Formatting fix.
28068 (oecount_cmp): Make sure to return the right return value
28069 even if unsigned fields are bigger than INT_MAX.
28070 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28073 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28074 TREE_READONLY on result if writing it more than once.
28077 * doc/invoke.texi (-fsanitize=thread): Document that with
28078 -fnon-call-exceptions atomics are not able to throw
28082 * tsan.c: Include tree-eh.h.
28083 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28084 maybe_clean_or_replace_eh_stmt where needed.
28085 (instrument_memory_accesses): Add cfg_changed argument.
28086 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28087 if it returned true.
28088 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28090 PR rtl-optimization/63191
28091 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28092 wrapper function, moved the whole old content into ...
28093 (ix86_delegitimize_address_1): ... this. New inline function.
28094 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28095 true as last argument instead of ix86_delegitimize_address.
28097 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28099 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28100 cortexa57_branch_cost.
28102 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28104 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28106 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28109 * doc/md.texi (Constraints): Document wA constraint.
28110 * config/rs6000/constraints.md (wA): New.
28111 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28112 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28113 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28114 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28116 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28119 * gimplify.c (is_oacc_declared): New function.
28120 (oacc_default_clause): Use it to set default flags for acc declared
28121 variables inside parallel regions.
28122 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28123 declared variables.
28124 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28125 declare attribute to any decl as necessary.
28127 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28130 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28131 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28132 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28133 (arm_arch_lpae): This.
28134 * config/arm/arm.c (arm_arch7ve): Rename into ...
28135 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28136 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28139 2017-03-22 Martin Liska <mliska@suse.cz>
28142 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28143 error message instead of an ICE.
28145 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28147 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28149 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28151 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28156 2017-03-21 Martin Sebor <msebor@redhat.com>
28158 * doc/extend.texi: Use "cannot" instead of "can't."
28159 * doc/hostconfig.texi: Same.
28160 * doc/install.texi: Same.
28161 * doc/invoke.texi: Same.
28162 * doc/loop.texi: Same.
28163 * doc/md.texi: Same.
28164 * doc/objc.texi: Same.
28165 * doc/rtl.texi: Same.
28166 * doc/tm.texi: Same.
28167 * doc/tm.texi.in: Same.
28168 * doc/trouble.texi: Same.
28170 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28173 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28174 (collect_checksum_attributes): Set it.
28175 (die_checksum_ordered): Use it.
28177 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28179 PR tree-optimization/79908
28180 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28181 change: For a VA_ARG whose LHS has been cast away, use
28182 force_gimple_operand to construct the side effects.
28184 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28186 PR translation/80001
28187 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28188 more amenable to translation.
28189 (oacc_loop_auto_partitions): Likewise.
28191 2017-03-21 Marek Polacek <polacek@redhat.com>
28192 Martin Sebor <msebor@redhat.com>
28194 PR tree-optimization/80109
28195 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28196 on INTEGRAL_TYPE_P.
28198 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28199 Segher Boessenkool <segher@kernel.crashing.org>
28202 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28203 check reg_used_between_p between insn and one of succ or succ2
28204 depending on if succ is artificial insn not inserted into insn
28207 2017-03-21 Martin Liska <mliska@suse.cz>
28209 PR gcov-profile/80081
28210 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28211 * doc/gcc.texi: Include gcov-dump stuff.
28212 * doc/gcov-dump.texi: New file.
28214 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28216 PR rtl-optimization/79150
28217 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28218 conditional jump, if the jump is the last insn of the loop.
28220 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28221 Richard Biener <rguenth@suse.de>
28223 PR tree-optimization/79908
28224 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28225 been cast away, use force_gimple_operand to construct the side
28228 2017-03-21 Martin Liska <mliska@suse.cz>
28230 PR libfortran/79956
28231 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28234 2017-03-21 Brad Spengler <spender@grsecurity.net>
28237 * plugin.c (htab_hash_plugin): New function.
28238 (add_new_plugin): Use it and adjust.
28239 (parse_plugin_arg_opt): Adjust.
28240 (init_one_plugin): Likewise.
28242 2017-03-21 Richard Biener <rguenther@suse.de>
28244 PR tree-optimization/80032
28245 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28246 if set force the cleanup to happen unconditionally.
28247 (gimplify_target_expr): Push inserted clobbers with force_uncond
28248 to avoid them being removed by control-dependent DCE.
28250 2017-03-21 Richard Biener <rguenther@suse.de>
28252 PR tree-optimization/80122
28253 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28254 va_arg_pack_len when the inlined call stmt requires pack
28256 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28258 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28261 * tsan.c (instrument_builtin_call): If the memory model argument
28262 is not a constant, assume it is valid.
28265 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28268 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28270 PR rtl-optimization/79910
28271 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28272 if its dest is used by an insn before I2 (other than the combined
28273 insns themselves, which are properly handled already).
28275 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28278 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28280 * combine.c (record_used_regs): New static function.
28281 (try_combine): Handle situations where there is an additional
28282 instruction between I2 and I3 which needs to have a LOG_LINK
28286 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28288 * combine.c (try_combine): Delete redundant i1 test. Call
28289 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28291 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28294 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28295 alternatives 13/14.
28297 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28299 PR tree-optimization/80054
28300 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28301 the optimization if a PHI or any of its arguments is not dominated
28302 by the candidate's basis. Use gphi* rather than gimple* as
28304 (replace_profitable_candidates): Clean up a gimple* variable that
28305 should be a gphi* variable.
28307 2017-03-20 Martin Sebor <msebor@redhat.com>
28310 * doc/extend.texi (attribute constructor): Document present limitation.
28312 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28315 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28316 __POWER9_VECTOR__ #ifdef control, change template definition to
28317 use Power9-specific built-in function.
28318 (vec_any_eq): Likewise.
28319 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28320 to control outcomes from this test.
28321 (vector_ae_<mode>p): For VEC_F modes, likewise.
28323 2017-03-20 Ian Lance Taylor <iant@google.com>
28325 * config/i386/i386.c (ix86_function_regparm): Save an extra
28326 register for -fsplit-stack with DECL_STATIC_CHAIN.
28328 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28331 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28332 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28334 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28336 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28338 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28341 2017-03-20 Marek Polacek <polacek@redhat.com>
28344 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28346 2017-03-20 Richard Biener <rguenther@suse.de>
28348 PR tree-optimization/80113
28349 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28350 allocate extra SSA name for PHI def.
28351 (add_close_phis_to_outer_loops): Likewise.
28352 (add_close_phis_to_merge_points): Likewise.
28353 (copy_loop_close_phi_args): Likewise.
28354 (copy_cond_phi_nodes): Likewise.
28356 2017-03-20 Martin Liska <mliska@suse.cz>
28358 PR middle-end/79753
28359 * tree-chkp.c (chkp_build_returned_bound): Do not build
28360 returned bounds for a LHS that's not a BOUNDED_P type.
28362 2017-03-20 Martin Liska <mliska@suse.cz>
28366 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28367 COMPLEX_CST and VECTOR_CST.
28369 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28372 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28373 target operand. A new splitter adds the clobber statement in case
28374 the target operand is dead anyway.
28376 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28378 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28379 to age-old versions of binutils and glibc.
28381 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28383 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28385 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28387 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28389 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28391 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28392 requirement for binutils 2.13.
28394 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28396 * combine.c (try_combine): Delete redundant i1 test. Call
28397 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28399 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28401 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28402 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28404 <riscv64-*-elf>: Re-arrange section
28405 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28406 <riscv32-*-linux>: Likewise.
28407 <riscv64-*-elf>: Likewise
28408 <riscv64-*-linux>: Likewise.
28410 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28413 * aarch64.opt(verbose-cost-dump): Fix typo.
28415 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28418 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28419 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28421 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28423 * reload.c (find_reloads): When reloading a nonoffsettable address,
28424 use RELOAD_OTHER for it and its address reloads.
28426 PR rtl-optimization/79910
28427 * combine.c (record_used_regs): New static function.
28428 (try_combine): Handle situations where there is an additional
28429 instruction between I2 and I3 which needs to have a LOG_LINK
28432 2017-03-17 Jeff Law <law@redhat.com>
28434 PR tree-optimization/71437
28435 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28436 conditional in the hash table first.
28437 (vrp_dom_walker::before_dom_children): Extract condition from
28438 ASSERT_EXPR. Record condition, its inverion and any implied
28439 conditions as well.
28441 2017-03-17 Marek Polacek <polacek@redhat.com>
28442 Markus Trippelsdorf <markus@trippelsdorf.de>
28444 PR tree-optimization/80079
28445 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28448 2017-03-17 Richard Biener <rguenther@suse.de>
28450 PR middle-end/80075
28451 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28452 Properly verify the LHS before the RHS possibly claims to be
28454 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28457 2017-03-17 Martin Jambor <mjambor@suse.cz>
28459 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28460 (List of -O2 options): Likewise.
28461 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28464 2017-03-17 Tom de Vries <tom@codesourcery.com>
28466 * gcov-dump.c (print_usage): Print bug_report_url.
28468 2017-03-17 Richard Biener <rguenther@suse.de>
28470 PR middle-end/80050
28471 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28472 (parser::peek): Likewise.
28474 2017-03-17 Richard Biener <rguenther@suse.de>
28476 PR tree-optimization/80048
28477 * sese.c (free_sese_info): Properly release rename_map and
28478 copied_bb_map elements.
28480 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28482 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28483 Add linked-list forward and backlinks. Insert on
28484 construction, remove on destruction.
28485 (class pass_store_merging): Add m_stores_head field.
28486 (pass_store_merging::terminate_and_process_all_chains):
28487 Iterate over m_stores_head list.
28488 (pass_store_merging::terminate_all_aliasing_chains):
28490 (pass_store_merging::execute): Check for debug stmts first.
28491 Push new chains onto the m_stores_head stack.
28493 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28496 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28497 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28498 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28500 2017-03-16 Jeff Law <law@redhat.com>
28502 PR tree-optimization/71437
28503 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28504 member function. Implementation moved into after_dom_children
28505 member function and into the threader's thread_outgoing_edges
28507 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28508 some code into new thread_outgoing_edges.
28509 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28510 definition. Simplify marker handling (do it here). Assume we always
28511 have the available expression and the const/copies tables.
28512 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28514 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28515 * tree-vrp.c (equiv_stack): No longer file scoped.
28516 (vrp_dom_walker): New class.
28517 (vrp_dom_walker::before_dom_children): New member function.
28518 (vrp_dom_walker::after_dom_children): Likewise.
28519 (identify_jump_threads): Setup domwalker. Use it rather than
28520 walking edges in a random order by hand. Simplify setup/finalization.
28521 (finalize_jump_threads): Remove.
28522 (vrp_finalize): Do not call identify_jump_threads here.
28523 (execute_vrp): Do it here instead and call thread_through_all_blocks
28526 PR tree-optimization/71437
28527 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28529 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28531 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28532 (dom_opt_dom_walker::thread_across_edge): Remove
28533 handle_dominating_asserts argument. All callers changed.
28534 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28535 changes. Remove calls to lhs_of_dominating_assert. Other
28536 uses of handle_dominating_asserts turn into unconditional code
28537 (simplify_control_stmt_condition_1): Likewise.
28538 (simplify_control_stmt_condition): Likewise.
28539 (thread_through_normal_block, thread_across_edge): Likewise.
28540 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28541 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28542 object if it is not an SSA_NAME.
28543 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28544 before calling into the VRP specific simplifiers.
28545 (identify_jump_threads): Remove handle_dominating_asserts
28548 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28551 * tree-diagnostic.c (default_tree_printer): No longer static.
28552 * tree-diagnostic.h (default_tree_printer): New prototype.
28554 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28556 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28557 Change ins into fmov.
28559 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28561 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28562 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28563 Use h_con constraint for operand 1.
28564 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28565 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28567 2017-03-15 Jeff Law <law@redhat.com>
28569 PR tree-optimization/71437
28570 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28571 (record_temporary_equivalences): Use it.
28573 PR tree-optimization/71437
28574 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28575 tree-ssa-scopedtables.
28576 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28577 (record_conditions, record_cond, vuse_eq): Likewise.
28578 (record_edge_info): Adjust to API tweak of record_conditions.
28579 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28580 (record_temporary_equivalences, optimize_stmt): Likewise.
28581 (eliminate_redundant_computations): Likewise.
28582 (record_equivalences_from_stmt): Likewise.
28583 * tree-ssa-scopedtables.c: Include options.h and params.h.
28584 (vuse_eq): New function, moved from tree-ssa-dom.c
28585 (build_and_record_new_cond): Likewise.
28586 (record_conditions): Likewise. Accept vector of conditions rather
28587 than edge_equivalence structure for first argument.
28588 for the first argument.
28589 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28590 from tree-ssa-dom.c.
28591 (avail_exprs_stack::record_cond): Likewise.
28592 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28593 from tree-ssa-dom.c.
28594 (avail_exprs_stack): Add new member functions lookup_avail_expr
28596 (record_conditions): Declare.
28598 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28601 * lra-constraints.c (process_alt_operands): Increase reject for
28602 reloading an input/output operand.
28604 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28607 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28608 insns to convert from signed/unsigned char/short to IEEE 128-bit
28610 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28612 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28615 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28616 subreg of inner mode for values already in registers.
28618 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28620 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28621 iteration reg is used after the loop.
28623 2017-03-14 Martin Sebor <msebor@redhat.com>
28625 PR tree-optimization/79800
28626 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28627 precision in negative-positive range.
28628 (format_floating): Call non-const overload with adjusted precision.
28630 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28633 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28636 2017-03-14 Martin Sebor <msebor@redhat.com>
28638 PR middle-end/80020
28639 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28640 * builtins.def (aligned_alloc): Use it.
28643 * Makefile.in (GTFILES): Add calls.c.
28644 * calls.c: Include "gt-calls.h".
28646 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28648 PR rtl-optimization/79728
28649 * regs.h (struct target_regs): New field
28650 x_contains_allocatable_regs_of_mode.
28651 (contains_allocatable_regs_of_mode): New macro.
28652 * reginfo.c (init_reg_sets_1): Initialize it, and change
28653 contains_reg_of_mode so it includes global regs as well.
28654 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28655 rather than contains_regs_of_mode.
28657 2017-03-14 Martin Liska <mliska@suse.cz>
28659 * doc/invoke.texi: Document options that can't be combined with
28660 -fcheck-pointer-bounds.
28662 2017-03-14 Martin Liska <mliska@suse.cz>
28664 PR middle-end/79831
28665 * doc/invoke.texi (-Wchkp): Document the option.
28667 2017-03-14 Martin Liska <mliska@suse.cz>
28669 * Makefile.in: Install gcov-dump.
28671 2017-03-14 Martin Liska <mliska@suse.cz>
28673 * multiple_target.c (expand_target_clones): Bail out for
28674 an invalid attribute.
28676 2017-03-14 Richard Biener <rguenther@suse.de>
28678 * alias.c (struct alias_set_entry): Pack properly.
28679 * cfgloop.h (struct loop): Likewise.
28680 * cse.c (struct set): Likewise.
28681 * ipa-utils.c (struct searchc_env): Likewise.
28682 * loop-invariant.c (struct invariant): Likewise.
28683 * lra-remat.c (struct cand): Likewise.
28684 * recog.c (struct change_t): Likewise.
28685 * rtl.h (struct address_info): Likewise.
28686 * symbol-summary.h (function_summary): Likewise.
28687 * tree-loop-distribution.c (struct partition): Likewise.
28688 * tree-object-size.c (struct object_size_info): Likewise.
28689 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28690 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28691 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28692 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28693 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28694 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28695 (struct _stmt_vec_info): Likewise.
28697 2017-03-14 Martin Liska <mliska@suse.cz>
28700 * multiple_target.c (create_dispatcher_calls): Check that
28701 a target can create a function dispatcher.
28703 2017-03-14 Martin Liska <mliska@suse.cz>
28706 * multiple_target.c (expand_target_clones): Drop local.local
28707 flag for default implementation.
28709 2017-03-14 Richard Biener <rguenther@suse.de>
28711 PR tree-optimization/80030
28712 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28714 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28716 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28717 gcc_fallthrough() instead of __attribute__((fallthrough));
28719 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28721 * doc/gcc.texi: Remove "up" link to (DIR).
28722 * doc/gccint.texi: Ditto.
28724 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28726 * doc/install.texi (Specific) <avr>: Remove reference to
28729 2017-03-13 Jeff Law <law@redhat.com>
28731 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28732 attribute rather than comments.
28734 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28735 match_scratch operand is highest.
28737 2017-03-13 Martin Liska <mliska@suse.cz>
28739 PR middle-end/78339
28740 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28741 is a CHKP clone, use original declaration.
28743 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28745 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28746 (arc_conditional_register_usage): Use a different allocation order
28747 when optimizing for size.
28748 * common/config/arc/arc-common.c (arc_option_optimization_table):
28749 Section anchors default on when optimizing for size.
28751 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28753 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28755 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28757 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28758 * config/arc/arc.md (cpu_facility): Add cd variant.
28759 (*movqi_insn): Add code density variant.
28760 (*movhi_insn): Likewise.
28761 (*movqi_insn): Likewise.
28762 (*addsi3_mixed): Likewise.
28763 (subsi3_insn): Likewise.
28765 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28767 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28769 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28771 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28772 expressions with MINUS and UNARY ops.
28774 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28777 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28779 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28780 between vec_select and vector argument.
28781 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28782 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28783 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28784 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28785 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28786 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28788 2017-03-13 Richard Biener <rguenther@suse.de>
28791 * params.def (vect-max-peeling-for-alignment): Fix typo.
28793 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28795 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28796 issue that only occurred with binutils below 2.18.
28798 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28800 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28801 refer to binutils 2.11/2.12 minimum.
28803 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28805 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28806 ftp.kernel.org and simplify binutils requirement.
28808 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28810 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28812 (Optimize Options): Ditto. Also remove redundancy.
28814 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28816 PR translation/79848
28817 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28819 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28820 to G_ to avoid double translation.
28822 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28824 PR translation/79923
28825 * auto-profile.c (get_combined_location): Convert leading
28826 character of diagnostics to lower case and remove trailing period.
28827 (read_profile): Likewise for various diagnostics.
28828 * config/arm/arm.c (arm_option_override): Remove trailing period
28829 from various diagnostics.
28830 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28831 (msp430_expand_delay_cycles): Likewise.
28833 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28836 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28837 full command-line argument, rather than just "str".
28838 (aarch64_validate_march): Likewise.
28839 (aarch64_validate_mtune): Likewise.
28841 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28843 PR rtl-optimization/78911
28844 * lra-assigns.c (must_not_spill_p): New function.
28845 (spill_for): Use it.
28847 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28849 PR tree-optimization/79981
28850 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28851 ATOMIC_COMPARE_EXCHANGE ifn result.
28852 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28853 IFN_ATOMIC_COMPARE_EXCHANGE.
28855 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28858 * opts.c (parse_sanitizer_options): Add missing question mark to
28859 "did you mean" message.
28861 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28863 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28865 (VMULEUH_UNS): Likewise.
28866 (VMULOUB_UNS): Likewise.
28867 (VMULOUH_UNS): Likewise.
28868 * config/rs6000/rs6000.c (builtin_function_type): Remove
28869 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28871 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28874 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28875 x with result of extra_parsing_for_operand_code_0.
28876 (function_reader::extra_parsing_for_operand_code_0): Convert
28877 return type from void to rtx, returning x. When reading
28878 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28879 larger size containing struct block_symbol.
28881 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28884 -mfloat128-hardware without -m64.
28886 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28889 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28890 entries to the case statement that marks unsigned arguments to
28891 overloaded functions.
28893 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28896 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28898 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28901 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28902 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28904 2017-03-10 Martin Liska <mliska@suse.cz>
28908 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28909 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28912 2017-03-10 Olivier Hainque <hainque@adacore.com>
28914 * tree-switch-conversion (array_value_type): Start by resetting
28915 candidate type to it's main variant.
28917 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28919 PR rtl-optimization/79909
28920 * combine.c (try_combine): Use simplify_replace_rtx on individual
28921 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28922 of the whole CALL_INSN_FUNCTION_USAGE.
28924 PR tree-optimization/79972
28925 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28926 get_range_info on SSA_NAMEs. Formatting fixes.
28928 2017-03-10 Richard Biener <rguenther@suse.de>
28929 Jakub Jelinek <jakub@redhat.com>
28931 PR tree-optimization/77975
28932 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28933 edge to be constant.
28934 (get_val_for): For constant x return it. Formatting fix.
28935 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28936 would use the same bases as the current one.
28938 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28940 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28941 instead of vec_select for V1TImode.
28942 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28944 (VSX_LE_128): Add V1TI to this mode iterator.
28945 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28946 (*vsx_le_perm_store_<mode>): Likewise.
28947 (pre-reload splitter for VSX stores): Likewise.
28948 (post-reload splitter for VSX stores): Likewise.
28949 (*vsx_xxpermdi2_le_<mode>): Likewise.
28950 (*vsx_lxvd2x2_le_<mode>): Likewise.
28951 (*vsx_stxvd2x2_le_<mode>): Likewise.
28953 2017-03-09 Michael Eager <eager@eagercon.com>
28955 Correct failures with --enable-checking=yes,rtl.
28957 * config/microblaze/microblaze.c (microblaze_expand_shift):
28958 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28959 test for const0_rtx.
28960 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28961 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28963 2017-03-09 Richard Biener <rguenther@suse.de>
28965 PR tree-optimization/79977
28966 * graphite-scop-detection.c (scop_detection::merge_sese):
28967 Handle the case of extra exits to blocks dominating the entry.
28969 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28971 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28974 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28976 PR rtl-optimization/79949
28977 * lra-constraints.c (process_alt_operands): Check memory when
28978 trying to predict a cycle. Print about the overall increase.
28980 2017-03-09 Richard Biener <rguenther@suse.de>
28982 PR middle-end/79971
28983 * gimple-expr.c (useless_type_conversion_p): Preserve
28984 TYPE_SATURATING for fixed-point types.
28986 2017-03-09 Richard Biener <rguenther@suse.de>
28989 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28990 alignment of BLKmode params.
28992 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28995 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28996 (VALL_NO_V2Q): Likewise.
28998 * config/aarch64/aarch64-simd.md
28999 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29001 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29002 VALL_NO_V2Q mode iterator.
29003 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29005 2017-03-09 Martin Liska <mliska@suse.cz>
29007 PR tree-optimization/79631
29008 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29009 tree_int_cst_sign_bit just for INTEGER constants.
29011 2017-03-09 Martin Liska <mliska@suse.cz>
29015 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29018 2017-03-09 Marek Polacek <polacek@redhat.com>
29021 * tree.c (inchash::add_expr): Handle TREE_VEC.
29023 2017-03-09 Martin Liska <mliska@suse.cz>
29026 (chkp_narrow_size_and_offset): New function.
29027 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29028 (void chkp_parse_bit_field_ref): New function.
29029 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29030 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29032 2017-03-09 Martin Liska <mliska@suse.cz>
29035 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29036 (chkp_find_bounds_1): Remove gcc_unreachable.
29038 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29041 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29042 BUILT_IN_SYNC*, determine the access type from the size suffix and
29043 always build a MEM_REF with that type. Handle forgotten
29044 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29047 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29048 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29049 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29050 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29051 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29052 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29053 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29054 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29055 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29056 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29057 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29058 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29059 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29060 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29061 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29062 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29063 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29064 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29065 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29066 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29067 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29068 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29069 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29070 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29071 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29072 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29073 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29074 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29075 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29076 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29077 definitions outside of __OPTIMIZE__ guarded section.
29080 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29081 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29082 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29083 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29086 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29088 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29089 ("vfenez<mode>"): Add missing constraints.
29091 2017-03-08 Martin Sebor <msebor@redhat.com>
29094 * config/nds32/nds32.c (nds32_option_override):
29095 Fix misspelled diagnostic.
29097 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29100 * gimplify.c (gimplify_omp_for): Replace index var in outer
29101 taskloop statement with an artificial variable and add
29102 OMP_CLAUSE_PRIVATE clause for it.
29104 2017-03-08 Richard Biener <rguenther@suse.de>
29106 PR tree-optimization/79955
29107 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29108 for accesses that are completely outside of the variable.
29110 2017-03-08 Andrew Haley <aph@redhat.com>
29112 PR tree-optimization/79943
29113 * tree-ssa-loop-split.c (compute_new_first_bound): When
29114 calculating the new upper bound, (END-BEG) should be added, not
29117 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29119 * config/avr/avr.md (setmemhi): Make sure match_dup
29120 operand number comes before match_scratch.
29122 2017-03-08 Richard Biener <rguenther@suse.de>
29124 PR tree-optimization/79920
29125 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29126 with ncopies == 1 to ...
29127 (vect_transform_slp_perm_load): ... here. Properly compute
29128 all element loads by iterating VF times over the group. Do
29129 not handle ncopies (computed in a broken way) in
29130 vect_create_mask_and_perm.
29132 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29135 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29136 is a uniform vector, use uniform_vector_p return value instead of
29137 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29139 2017-03-07 Marek Polacek <polacek@redhat.com>
29141 PR middle-end/79809
29142 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29143 (alloca_call_type): Likewise.
29145 2017-03-07 Martin Liska <mliska@suse.cz>
29147 * gcov.c (process_args): Put comment to correct location.
29149 2017-03-07 Martin Liska <mliska@suse.cz>
29151 PR middle-end/68270
29152 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29153 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29154 (chkp_narrow_bounds_for_field): Likewise.
29155 (chkp_parse_array_and_component_ref): Pass one more argument to
29158 2017-03-07 Richard Biener <rguenther@suse.de>
29160 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29163 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29165 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29166 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29168 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29171 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29172 to end of description.
29173 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29175 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29177 PR rtl-optimization/79901
29178 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29180 (*avx512f_<code><mode>3<mask_name>): ... this.
29181 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29182 iterator instead of VI8_AVX2_AVX512BW.
29184 PR rtl-optimization/79901
29185 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29186 min/max expander, expand it using expand_vec_cond_expr.
29189 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29192 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29195 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29196 to void * for PCH reasons.
29197 * dwarf2out.c (output_loc_operands, output_die): Cast
29198 v.val_vec.array to unsigned char *.
29200 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29203 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29206 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29208 PR rtl-optimization/79571
29209 * lra-constraints.c (process_alt_operands): Calculate static
29210 reject and subtract it from overall when only addresses will be
29213 2017-03-06 Julia Koval <julia.koval@intel.com>
29216 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29217 incoming stack boundary to 128 for 64-bit targets.
29219 2017-03-06 Richard Biener <rguenther@suse.de>
29221 PR tree-optimization/79894
29222 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29223 to NULL after folding it.
29225 2017-03-06 Richard Biener <rguenther@suse.de>
29227 PR tree-optimization/79824
29228 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29229 check disabling peeling for gaps.
29231 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29233 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29234 attributes): Document gettimeofday.
29236 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29238 * config/s390/s390.c (s390_option_override_internal): Set
29239 PARAM_MIN_VECT_LOOP_BOUND
29241 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29243 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29244 * config/s390/s390.md: Likewise.
29246 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29249 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29250 (<avx2_avx512>_perm<mode>): Rename to ...
29251 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29253 (<avx512>_perm<mode>_mask): Rename to ...
29254 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29256 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29257 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29258 instead of VI8F_256_512.
29259 (avx512f_perm<mode>): New define_expand.
29260 (avx512f_perm<mode>_mask): Likewise.
29261 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29262 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29264 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29266 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29267 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29269 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29271 2017-03-06 Martin Liska <mliska@suse.cz>
29274 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29275 when having a SSA NAME w/o VAR_DECL assigned to it.
29277 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29279 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29280 msa_dpsub_<su>_d): Fix MODE for vec_select.
29282 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29284 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29286 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29288 2017-03-06 Richard Biener <rguenther@suse.de>
29290 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29291 * plugin.c (register_plugin_info): Likewise.
29292 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29294 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29296 * config/i386/sse.md (sse_storehps, sse_storelps,
29297 avx_<castmode><avxsizesuffix>_<castmode>,
29298 avx512f_<castmode><avxsizesuffix>_<castmode>,
29299 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29300 in condition that at least one operand is not a MEM.
29302 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29304 PR middle-end/79805
29305 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29306 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29308 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29309 gimple_call_nothrow_p flag based on whether original builtin can throw.
29310 If it can, emit following stmts on the fallthrough edge.
29311 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29312 don't create new bb if inserting just debug stmts on the edge, try to
29313 insert them on the fallthru bb or just reset debug stmts.
29315 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29318 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29319 restore recog_data (including the operand rtxes inside it) around
29320 the call to get_insn_template.
29322 2017-03-03 Martin Sebor <msebor@redhat.com>
29324 PR tree-optimization/79699
29325 * context.c (context::~context): Free MPFR caches to avoid
29326 a memory leak on program exit.
29328 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29330 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29331 Use wide_int::ulow () instead of .elt (0).
29333 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29335 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29336 (*pushxf): Limit oF constraint to 32bit targets and add oC
29337 constraint for 64bit targets.
29338 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29339 (*pushdf): Change rmF constraint to rmC.
29341 2017-03-03 Martin Liska <mliska@suse.cz>
29343 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29344 Remove unused variable.
29346 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29349 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29350 is a memory operand, increase num_memory.
29351 (ix86_expand_args_builtin): Likewise.
29353 2017-03-03 Jan Hubicka <jh@suse.cz>
29356 * ipa-devirt.c (maybe_record_node): Properly handle
29357 __cxa_pure_virtual visibility.
29359 2017-03-03 Martin Liska <mliska@suse.cz>
29361 PR tree-optimization/79803
29362 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29364 (pass_loop_prefetch::execute): Disabled optimization if an
29365 assumption about L1 cache size is not met.
29367 2017-03-03 Martin Liska <mliska@suse.cz>
29369 PR rtl-optimization/79574
29370 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29371 (hash_scan_set): Likewise.
29372 (dump_hash_table): Likewise.
29373 (hoist_code): Likewise.
29375 2017-03-03 Richard Biener <rguenther@suse.de>
29377 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29378 in place of uhigh ().
29379 (fixed_convert_from_real): Likewise.
29381 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29384 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29386 2017-03-03 Richard Biener <rguenther@suse.de>
29388 PR middle-end/79818
29389 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29390 TYPE_OVERFLOW_UNDEFINED check.
29392 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29394 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29396 (vector_ae_<mode>_p): Likewise.
29397 (vector_nez_<mode>_p): Likewise.
29398 (vector_ne_v2di_p): Likewise.
29399 (vector_ae_v2di_p): Likewise.
29400 (vector_ne_<mode>_p): Likewise.
29401 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29403 (vsx_tsqrt<mode>2_fe): Likewise.
29405 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29408 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29410 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29412 PR rtl-optimization/79780
29413 * cprop.c (one_cprop_pass): When second and further conditional trap
29414 in a single basic block is turned into an unconditional trap, turn it
29415 into a deleted note to avoid RTL verification failures.
29417 2017-03-02 Richard Biener <rguenther@suse.de>
29419 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29421 2017-03-02 Richard Biener <rguenther@suse.de>
29423 PR tree-optimization/79345
29425 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29426 param and abort the walk, returning -1 if it is hit.
29427 (walk_aliased_vdefs): Take a limit param and pass it on.
29428 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29429 defaulting to 0 and return a signed int.
29430 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29431 (check_defs): New helper.
29432 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29433 about uninitialized memory.
29434 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29435 bogus uninitialized warning.
29436 (fixed_convert_from_real): Likewise.
29438 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29440 PR tree-optimization/66768
29441 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29442 iv_use if base object can't be determined.
29444 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29446 PR tree-optimization/79345
29447 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29448 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29449 (get_pattern_stats): Initialize it.
29450 * genemit.c (gen_expand): Verify match_scratch numbers come after
29451 match_operand/match_dup numbers.
29452 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29453 match_scratch numbers.
29454 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29456 * config/s390/s390.md (trunctdsd2): Likewise.
29458 2017-03-02 Richard Biener <rguenther@suse.de>
29460 * wide-int.h (wide_int_storage::operator=): Implement in terms
29463 2017-03-02 Richard Biener <rguenther@suse.de>
29465 PR tree-optimization/79777
29466 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29467 the to insert expression to sth existing.
29469 2017-03-01 Martin Sebor <msebor@redhat.com>
29471 PR middle-end/79692
29472 * gimple-ssa-sprintf.c
29473 (directive::known_width_and_precision): New function.
29474 (format_integer): Use it.
29475 (get_mpfr_format_length): Consider the full range of precision
29476 when computing %g output with the # flag. Set the likely byte
29477 count to 3 rather than 1 when precision is indeterminate.
29478 (format_floating): Correct the lower bound of precision.
29480 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29482 * doc/invoke.texi: Document default code model for 64-bit Linux.
29484 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29487 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29488 udiv rather than div since input pattern is unsigned.
29490 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29492 * config/i386/i386.c (print_reg): Warn for values of
29493 unsupported size in integer register.
29495 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29498 * config/rs6000/predicates.md (current_file_function_operand): Do
29499 not allow self calls to be local if the function is replaceable.
29501 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29504 * config/rs6000/altivec.h (vec_ctz and others): Change the
29505 preprocessor macro that controls conditional compilation from
29506 _ARCH_PWR9 to __POWER9_VECTOR__.
29507 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29508 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29509 control (instead of _ARCH_PWR9 control) so that template
29510 definition uses power9-specific function.
29511 (vec_any_eq): Likewise.
29512 (vec_all_ne): Change macro definition to use a power9-specific
29513 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29514 _ARCH_PWR9 control).
29515 (vec_any_eq) Likewise.
29516 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29517 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29518 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29519 support for xvcmpnedp instruction.
29520 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29521 macro expansion so that Power9 implementation of vec_all_ne does
29522 not use the AltiVec predicate framework.
29523 (VCMPNEH_P): Likewise.
29524 (VCMPNEW_P): Likewise.
29525 (VCMPNED_P): Likewise.
29526 (VCMPNEFP_P): Likewise.
29527 (VCMPNEDP_P): Likewise.
29528 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29529 implementation of vec_any_eq to not use AltiVec predicate
29531 (VCMPAEH_P): Likewise.
29532 (VCMPAEW_P): Likewise.
29533 (VCMPAED_P): Likewise.
29534 (VCMPAEFP_P): Likewise.
29535 (VCMPAEDP_P): Likewise.
29536 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29537 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29538 not use the AltiVec predicate framework.
29539 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29540 of vec_any_eq to not use AltiVec predicate framework.
29541 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29542 support for predefined __POWER9_VECTOR__ macro to indicate that
29543 Power9 instruction selection is enabled.
29544 (altivec_overloaded_builtins): Remove extraneous
29545 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29546 function argument types RS6000_BTI_bool_V16QI and
29547 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29548 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29549 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29550 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29551 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29552 Power9 for implementations of vec_cmpne. Change the signature for
29553 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29554 (representing vec_all_ne) to remove the previously described first
29555 argument of type RS6000_BTI_INTSI, as this was an artifact of
29556 reliance on the AltiVec predicate framework, which is no longer
29557 used in the implementation of these functions. Add
29558 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29559 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29560 since, unlike the AltiVec predicate framework implementation, we
29561 do not share function descriptors between vec_alle and vec_anyeq.
29562 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29563 set of modes that receive special treatment even when
29564 TARGET_P9_VECTOR is true. The special treatment emits code that
29565 does not depend on Power9 instructions.
29566 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29567 define_expand to not rely on AltiVec predicate framework.
29568 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29570 (vector_ne_v2di_p): Change this define_expand to not rely on
29571 AltiVec predicate framework.
29572 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29574 (vector_ne_<mode>_p): Change this define_expand to not rely on
29575 AltiVec predicate framework.
29576 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29578 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29579 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29580 define_insn pattern.
29581 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29582 define_insn pattern because the xvcmpne<VSs>. instruction is not
29584 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29585 instruction is not supported.
29587 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29589 * config/nvptx/nvptx.c: Include intl.h.
29591 2017-03-01 Martin Jambor <mjambor@suse.cz>
29594 * ipa-prop.h (ipa_bits): Removed field known.
29595 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29596 to pointers. Adjusted their comments to warn about their sharing.
29597 (ipcp_transformation_summary): Change bits to a vector of pointers.
29598 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29599 (ipa_get_ipa_bits_for_value): Declare.
29600 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29601 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29602 (ipa_bits_hash_table): Likewise.
29603 (ipa_vr_ggc_hash_traits): Likewise.
29604 (ipa_vr_hash_table): Likewise.
29605 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29606 being pointers and vr_known being removed.
29607 (ipa_set_jf_unknown): Likewise.
29608 (ipa_get_ipa_bits_for_value): New function.
29609 (ipa_set_jfunc_bits): Likewise.
29610 (ipa_get_value_range): New overloaded functions.
29611 (ipa_set_jfunc_vr): Likewise.
29612 (ipa_compute_jump_functions_for_edge): Use the above functions to
29613 construct bits and vr parts of jump functions.
29614 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29615 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29617 (ipcp_grow_transformations_if_necessary): Also allocate
29618 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29620 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29621 them. Fix too long lines.
29622 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29623 vr_known being removed.
29624 (ipa_read_jump_function): Use new setter functions to construct bits
29625 and vr parts of jump functions or set them to NULL.
29626 (write_ipcp_transformation_info): Adjust for bits being pointers.
29627 (read_ipcp_transformation_info): Likewise.
29628 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29630 Include gt-ipa-prop.h.
29631 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29633 (ipcp_store_bits_results): Likewise.
29634 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29635 Do not write to existing jump functions but use a temporary instead.
29637 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29640 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29641 attempt to use its first operand as BIT_FIELD_REF base.
29643 2017-03-01 Richard Biener <rguenther@suse.de>
29645 PR middle-end/79721
29646 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29647 interpolating formula in wrapping arithmetic.
29648 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29650 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29652 PR tree-optimization/79734
29653 * tree-vect-generic.c (expand_vector_condition): Optimize
29654 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29655 Handle VEC_COND_EXPR where comparison has different inner width from
29656 type's inner width.
29658 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29660 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29661 markup, and similar issues. Remove @opindex entries for things
29662 that aren't options. Add missing -mmpy-option entries.
29664 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29666 PR tree-optimization/79737
29667 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29668 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29669 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29670 instead of byte_size. Formatting fix.
29671 (shift_bytes_in_array_right): Formatting fix.
29673 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29676 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29677 condition on optimize for the leaf function test.
29679 2017-02-28 Martin Liska <mliska@suse.cz>
29682 * read-rtl-function.c (function_reader::handle_unknown_directive):
29683 Bail out when one uses -flto.
29685 2017-02-28 Martin Liska <mliska@suse.cz>
29687 * common.opt: Replace space with tabular for options of <number>
29689 * config/i386/i386.opt: Show <number> value for
29690 -mlarge-data-threshold.
29691 * opts.c (print_filtered_help): Do not display number in hexadecimal
29694 2017-02-28 Martin Liska <mliska@suse.cz>
29696 * common.opt: Fix --help=option -Q for options which are of
29699 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29701 * config/i386/i386.c (print_reg): Error out for values
29702 of 8-bit size in invalid integer register.
29704 2017-02-28 Martin Sebor <msebor@redhat.com>
29706 PR tree-optimization/79691
29707 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29709 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29712 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29713 gcc_unreachable with output_operand_lossage.
29715 2017-02-28 Richard Biener <rguenther@suse.de>
29717 PR tree-optimization/79740
29718 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29720 (visit_nary_op): Insert the nary into the hashtable if we
29721 pattern-matched sth.
29722 * tree-ssa-pre.c (eliminate_insert): Robustify.
29724 2017-02-28 Richard Biener <rguenther@suse.de>
29726 PR middle-end/79731
29727 * fold-const.c (decode_field_reference): Reject out-of-bound
29730 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29732 * config/i386/i386.c: Include intl.h.
29733 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29734 instead of just cond ? "..." : "...".
29735 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29736 * coverage.c (read_counts_file): Likewise.
29737 * omp-offload.c: Include intl.h.
29738 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29739 of just cond ? "..." : "...".
29740 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29741 of just cond ? "..." : "...".
29743 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29746 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29748 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29749 'tune for' CPU name.
29750 * config/arm/arm-cpu-data.h: Regenerated.
29752 2017-02-28 Richard Biener <rguenther@suse.de>
29754 PR tree-optimization/79732
29755 * tree-inline.c (expand_call_inline): Do not shadow var.
29757 2017-02-28 Richard Biener <rguenther@suse.de>
29759 PR tree-optimization/79723
29760 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29761 address-space properly.
29763 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29765 * doc/optinfo.texi (Optimization groups): Fix option used for
29767 * doc/invoke.texi (-fopt-info): Document "omp".
29768 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29769 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29770 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29771 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29772 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29774 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29776 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29777 OPTGROUP_OMP with "omp".
29779 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29782 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29783 for arithmetic shift of unsigned V2DI.
29785 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29787 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29788 arc/linux.h headers.
29789 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29790 (LINK_SPEC): Likewise.
29791 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29792 (EXTRA_SPECS): Likewise.
29793 (STARTFILE_SPEC): Likewise.
29794 (ENDFILE_SPEC): Likewise.
29795 (LIB_SPEC): Likewise.
29796 (TARGET_SDATA_DEFAULT): Likewise.
29797 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29798 (MULTILIB_DEFAULTS): Likewise.
29799 (DWARF2_UNWIND_INFO): Likewise.
29800 * config/arc/big.h: New file.
29801 * config/arc/elf.h: Likewise.
29802 * config/arc/linux.h: Likewise.
29803 * config/arc/t-uClibc: Remove.
29805 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29807 PR tree-optimization/77536
29808 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29809 (tree_transform_and_unroll_loop): Use above function to compute the
29810 estimated niter of unrolled loop and use it when scaling profile.
29811 Also use count info rather than frequency if it's non-zero.
29812 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29813 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29814 (vect_transform_loop): Call above function.
29816 2017-02-27 Richard Biener <rguenther@suse.de>
29818 PR tree-optimization/45397
29819 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29820 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29821 (visit_nary_op): Add pattern matching for CSEing sign-changed
29822 or truncated operations with wider ones.
29824 2017-02-27 Richard Biener <rguenther@suse.de>
29826 PR tree-optimization/79690
29827 * tree-vect-stmts.c (vectorizable_store): Use vector type
29828 built from the DR with address-space.
29830 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29832 * doc/invoke.texi (Optimize Options): Refine the description
29833 of asan-use-after-return.
29835 2017-02-25 Alan Modra <amodra@gmail.com>
29837 PR rtl-optimization/79584
29838 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29839 base, not ad->base_term, the reg within base. Remove assertion
29840 that ad->base == ad->base_term. Replace gen_int_mode using
29841 bogus mode with const0_rtx.
29843 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29845 PR middle-end/79396
29846 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29847 FMA_EXPR like tcc_binary or tcc_unary.
29849 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29852 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29854 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29855 (output_loc_operands): Handle DW_OP_call_ref and
29856 DW_OP_GNU_variable_value.
29857 (struct variable_value_struct): New type.
29858 (struct variable_value_hasher): Likewise.
29859 (variable_value_hash): New variable.
29860 (string_types): Remove.
29861 (copy_loc_descr): New function.
29862 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29863 (prepend_loc_descr_to_each): New function.
29864 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29865 instead of add_loc_descr_to_each if the first argument is single
29866 location list and the second has multiple.
29867 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29868 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29869 when looking for variable value which doesn't have other location info.
29870 (loc_list_from_tree): Formatting fix.
29871 (gen_array_type_die): Simplify DW_AT_string_length handling.
29872 (adjust_string_types): Remove.
29873 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29874 string_types. Call resolve_variable_values.
29875 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29876 (resolve_addr_in_expr): Likewise. Add A argument.
29877 (copy_deref_exprloc): Remove deref argument. Adjust for the
29878 original expression being DW_OP_GNU_variable_value with optionally
29879 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29880 optionally after it.
29881 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29882 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29883 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29884 (variable_value_hasher::hash, variable_value_hasher::equal): New
29886 (resolve_variable_value_in_expr, resolve_variable_value,
29887 resolve_variable_values, note_variable_value_in_expr,
29888 note_variable_value): New functions.
29889 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29892 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29895 * opts.h (handle_generated_option): Add GENERATED_P argument.
29896 * opts-common.c (handle_option): Adjust function comment.
29897 (handle_generated_option): Add GENERATED_P argument, pass it to
29899 (control_warning_option): Pass false to handle_generated_option
29901 * opts.c (maybe_default_option): Pass true to handle_generated_option
29903 * optc-gen.awk: Likewise.
29905 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29907 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29908 a REG, look at the REG it is a SUBREG of.
29909 (splitter for cmpeqsi_t): Ditto.
29911 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29913 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29914 the special USEs with the pattern of the insn, not the insn itself.
29916 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29919 * doc/invoke.texi: Document -mload-store-pairs.
29921 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29922 Sandra Loosemore <sandra@codesourcery.com>
29924 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29925 argument isn't a CONST_INT.
29926 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29927 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29928 (nios2_validate_compare): Bypass alternate compare logic if *op2
29929 is not a CONST_INT.
29930 (ldstwm_operation_p): Return false if first_base is not a REG or
29931 if first_offset is not a CONST_INT.
29933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29935 * config/cris/cris.md: Use correct operand in a define_peephole2.
29937 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29939 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29941 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29943 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29944 this_insn if it is an INSN or JUMP_INSN.
29945 (force_offsettable): Look at base, not at addr.
29946 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29947 on things that aren't necessarily CONST_INTs.
29949 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29951 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29952 -mfpmath=sse is the default also for x86-32 targets with SSE2
29953 instruction set when @option{-ffast-math} is enabled
29955 2017-02-24 Jeff Law <law@redhat.com>
29957 PR rtl-optimizatoin/79286
29958 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29961 2017-02-24 Richard Biener <rguenther@suse.de>
29963 PR tree-optimization/79389
29964 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29967 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29969 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29970 function comment to reflect reality.
29971 (loop_exits_before_overflow): Fix typo in function description.
29973 2017-02-24 Richard Biener <rguenther@suse.de>
29975 PR tree-optimization/79389
29976 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29977 properly that a threading opportunity exists. Detect conditional
29978 copy/constant propagation opportunities.
29980 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29982 * config/visium/visium.md (type): Add trap.
29983 (b): New mode attribute.
29984 (*btst): Rename into...
29985 (*btst<mode>): ...this and adjust.
29986 (*cbranchsi4_btst_insn): Rename into...
29987 (*cbranch<mode>4_btst_insn): ...this and adjust.
29988 (trap): New define_insn.
29990 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29992 PR tree-optimization/79389
29993 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29994 (noce_reversed_cond_code): New function.
29995 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29996 reversed_comparison_code. Formatting fix.
29997 (noce_try_store_flag): Test rev_cond != NULL in addition to
29998 reversed_comparison_code.
29999 (noce_try_store_flag_constants): Likewise.
30000 (noce_try_store_flag_mask): Likewise.
30001 (noce_try_addcc): Use rev_cond if non-NULL instead of
30002 reversed_comparison_code.
30003 (noce_try_cmove_arith): Likewise. Formatting fixes.
30004 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30005 (noce_find_if_block): Initialize rev_cond.
30006 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30007 instead of false as last argument never attempt to reverse it
30010 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30012 PR tree-optimization/79663
30013 * tree-predcom.c (combine_chains): Process refs in reverse order
30014 only for ZERO length chains, and add explaining comment.
30016 2017-02-23 Jeff Law <law@redhat.com>
30018 PR tree-optimization/79578
30019 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30020 in call to operand_equal_p.
30022 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30025 * config/i386/cpuid.h: Fix another undefined behavior.
30027 2017-02-23 Richard Biener <rguenther@suse.de>
30029 PR tree-optimization/79683
30030 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30031 vector types for data-refs.
30033 2017-02-23 Martin Liska <mliska@suse.cz>
30035 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30037 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30039 PR middle-end/79665
30040 * internal-fn.c (get_range_pos_neg): Moved to ...
30041 * tree.c (get_range_pos_neg): ... here. No longer static.
30042 * tree.h (get_range_pos_neg): New prototype.
30043 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30044 are known to be in between 0 and signed maximum inclusive, try to
30045 expand both unsigned and signed divmod and use the cheaper one from
30048 2017-02-22 Jeff Law <law@redhat.com>
30050 PR tree-optimization/79578
30051 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30052 to compare base operands.
30054 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30057 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30058 gpc_reg_operand instead of fpr_reg_operand.
30060 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30062 * config/mips/mips.c (mips_return_in_memory): Force FP
30063 vector types to be returned in memory for o32 ABI.
30065 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30067 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30068 instead of DW_TAG_member for static data member declarations and don't
30069 set no_linkage_name for static inline data members.
30070 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30073 2017-02-22 Martin Liska <mliska@suse.cz>
30075 * doc/invoke.texi: Replace inequality signs with square brackets
30078 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30081 * lra-constraints.c (simplify_operand_subreg): Handle
30082 WORD_REGISTER_OPERATIONS targets.
30084 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30087 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30088 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30089 elimination by swapping fld*.
30091 2017-02-22 Richard Biener <rguenther@suse.de>
30093 PR tree-optimization/79673
30094 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30095 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30096 irrelevant address-space qualifiers and avoiding a
30097 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30099 2017-02-22 Richard Biener <rguenther@suse.de>
30101 PR tree-optimization/79666
30102 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30103 to not symbolically negate if that may introduce undefined
30106 2017-02-22 Martin Liska <mliska@suse.cz>
30109 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30110 * data-streamer-out.c (streamer_write_gcov_count_stream):
30112 * value-prof.c (stream_out_histogram_value): Make assert more
30113 precise based on type of counter.
30115 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30118 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30119 Use nonimmediate_operand instead of memory_operand for operand 1.
30120 (float-extend standard_x87sse_constant_load splitter): Ditto.
30122 2017-02-21 Jeff Law <law@redhat.com>
30124 PR tree-optimization/79621
30125 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30126 blocks with edges to themselves.
30128 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30131 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30132 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30133 Use gimple_call_builtin_p.
30136 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30137 on temporarily removed DEBUG_INSNs.
30139 PR tree-optimization/79649
30140 * tree-loop-distribution.c (classify_partition): Give up on
30141 non-generic address space loads/stores.
30143 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30145 * doc/loop.texi (Loop manipulation): Remove nonexistent
30146 tree_ssa_loop_version from the documentation.
30147 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30149 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30152 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30153 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30154 * config/rs6000/rs6000.c: Include except.h.
30155 (rs6000_expand_split_stack_prologue): Call
30156 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30158 2017-02-21 Martin Jambor <mjambor@suse.cz>
30161 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30162 have been analyzed.
30164 2017-02-21 Martin Jambor <mjambor@suse.cz>
30166 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30167 for backward compatibility only.
30168 * doc/invoke.texi (Option Summary): Remove all references to
30169 -fipa-cp-alignment.
30171 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30175 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30177 * lra-constraints.c (curr_insn_transform): Handle
30178 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30180 2017-02-21 Martin Liska <mliska@suse.cz>
30182 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30184 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30186 PR translation/79638
30187 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30189 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30192 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30193 (arm_function_ok_for_sibcall): Return false for an indirect call by
30194 descriptor if all the argument registers are used.
30195 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30196 alignment of the function.
30198 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30200 PR tree-optimization/61441
30201 * simplify-rtx.c (simplify_const_unary_operation): For
30202 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30203 the sNaN unmodified.
30205 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30207 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30208 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30209 instead of SYSTEM_HEADER_DIR.
30211 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30212 Martin Liška <mliska@suse.cz>
30214 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30215 Fix typos and grammar, use active voice, and clarify.
30217 2017-02-20 Marek Polacek <polacek@redhat.com>
30219 PR middle-end/79537
30220 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30223 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30225 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30228 * config/i386/i386.c (ix86_expand_builtin): Handle
30229 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30230 ix86_builtins_isa[fcode].isa as a requirement of those
30231 flags and any other flag in the bitmask.
30232 (ix86_init_mmx_sse_builtins): Use 0 instead of
30233 ~OPTION_MASK_ISA_64BIT as mask.
30234 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30235 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30236 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30237 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30239 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30242 * lra-constraints.c (split_reg): Check requested split mode
30243 is supported by the register.
30245 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30247 * lra-constraints.c (simplify_operand_subreg): Remove early
30250 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30253 * lra-constraints.c (curr_insn_transform): Tighten condition
30254 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30256 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30259 * lra-constraints.c (curr_insn_transform): Handle
30260 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30262 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30265 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30267 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30269 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30272 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30275 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30277 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30280 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30282 * final.c (last_columnnum, override_columnnum): New variables.
30283 (final_start_function): Set last_columnnum, pass it to begin_prologue
30284 hook and pass 0 to dwarf2out_begin_prologue.
30285 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30286 to source_line debug hook.
30287 (notice_source_line): Compute last_columnnum and for debug_column_info
30288 return true on column changes.
30289 * debug.h (struct gcc_debug_hooks): Add column argument to
30290 source_line and begin_prologue hooks.
30291 (debug_nothing_int_charstar_int_bool): Remove prototype.
30292 (debug_nothing_int_int_charstar,
30293 debug_nothing_int_int_charstar_int_bool): New prototypes.
30294 (dwarf2out_begin_prologue): Add column argument.
30295 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30296 begin_prologue hooks.
30297 (debug_nothing_int_charstar_int_bool): Remove.
30298 (debug_nothing_int_int_charstar,
30299 debug_nothing_int_int_charstar_int_bool): New functions.
30300 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30301 through to dwarf2out_source_line.
30302 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30303 (dwarf2out_source_line): Add column argument, emit it if requested.
30304 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30306 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30307 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30308 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30309 through to dwarf2out_begin_prologue.
30310 (vmsdbgout_source_line): Add column argument, pass it through to
30311 dwarf2out_source_line.
30312 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30313 dbxout_source_line caller.
30314 (dbxout_source_line): Add column argument.
30316 * common.opt (gno-column-info, gcolumn-info): New options.
30317 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30318 (check_die): Also test for multiple DW_AT_decl_column attributes.
30319 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30320 DW_AT_decl_column if requested.
30321 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30323 (gen_variable_die): Likewise.
30324 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30325 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30328 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30329 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30330 (ix86_handle_option): Handle OPT_m3dnowa.
30331 * doc/invoke.texi (-m3dnowa): Document.
30332 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30333 -m3dnowa instead of -m3dnow -march=athlon.
30336 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30337 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30339 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30342 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30343 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30344 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30345 generator for vsx_xxpermdi_<mode>_be.
30346 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30347 force big-endian semantics.
30348 (vsx_xxpermdi_<mode>_be): New define_expand with same
30349 implementation as previous version of vsx_xxpermdi_<mode>.
30351 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30353 PR tree-optimization/79327
30354 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30355 variable, its initialization and use.
30357 2017-02-17 Julia Koval <julia.koval@intel.com>
30359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30360 (OPTION_MASK_ISA_PKU_UNSET): New.
30361 (ix86_handle_option): Handle -mrdpid.
30362 * config/i386/cpuid.h (bit_RDPID): New.
30363 * config/i386/driver-i386.c (host_detect_local_cpu):
30364 Detect RDPID feature.
30365 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30366 * config/i386/i386-c.c (ix86_target_macros_internal):
30368 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30369 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30370 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30371 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30372 * config/i386/i386.md (define_insn "rdpid"): New.
30373 * config/i386/i386.opt Add -mrdpid.
30374 * config/i386/immintrin.h (_rdpid_u32): New.
30376 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30378 PR rtl-optimization/79541
30379 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30380 instead of transforming it into USE.
30382 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30384 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30385 If HONOR_SNANS (SFmode) force the input to a register.
30386 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30387 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30388 an frsp or similar insn.
30390 2017-02-17 Martin Liska <mliska@suse.cz>
30392 PR rtl-optimization/79577
30393 * params.def (selsched-max-sched-times): Increase minimum to 1.
30395 2017-02-17 Martin Liska <mliska@suse.cz>
30397 PR rtl-optimization/79574
30398 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30400 2017-02-17 Martin Liska <mliska@suse.cz>
30402 PR tree-optimization/79529
30403 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30404 ssa_defined_default_def_p to handle cases which are implicitly
30406 * tree-ssa.c (ssa_defined_default_def_p): New function.
30407 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30408 which are implicitly defined.
30409 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30411 2017-02-17 Richard Biener <rguenther@suse.de>
30413 PR middle-end/79576
30414 * params.def (max-ssa-name-query-depth): Limit to 10.
30416 2017-02-17 Richard Biener <rguenther@suse.de>
30418 PR tree-optimization/79552
30419 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30422 2017-02-17 Richard Biener <rguenther@suse.de>
30425 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30427 2017-02-17 Marek Polacek <polacek@redhat.com>
30429 PR middle-end/79536
30430 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30431 (fold_negate_expr): New wrapper.
30433 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30435 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30436 Correct terminology and de-emphasize pre-standard behavior.
30438 2017-02-16 Alan Modra <amodra@gmail.com>
30440 PR rtl-optimization/79286
30441 * ira.c (def_dominates_uses): New function.
30442 (update_equiv_regs): Don't create an equivalence for insns that
30443 may trap where the register def does not dominate the use.
30445 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30447 PR rtl-optimization/78127
30448 * lra.c (lra): Call lra_eliminate before finish the loop after
30451 2017-02-16 Richard Biener <rguenther@suse.de>
30453 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30455 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30456 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30457 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30458 (isl_val_int_from_wi): New function.
30459 (extract_affine_gmp): Rename to ...
30460 (extract_affine_wi): ... this, take a widest_int.
30461 (extract_affine_int): Just wrap extract_affine_wi.
30462 (add_param_constraints): Use isl_val_int_from_wi.
30463 (add_loop_constraints): Likewise, and extract_affine_wi.
30465 2017-02-15 Jeff Law <law@redhat.com>
30467 PR middle-end/79521
30468 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30469 ira_init_register_move_cost_if_necessary.
30471 2017-02-15 Martin Sebor <msebor@redhat.com>
30473 PR middle-end/32003
30474 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30475 removed in a prior commit.
30477 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30479 PR tree-optimization/79347
30480 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30481 counters during peeling.
30483 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30485 * Makefile.in (site.exp): Remove "set ISLVER".
30487 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30490 * real.c (real_from_integer): Call real_convert even for decimal.
30492 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30495 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30497 2017-02-14 Andrew Pinski <apinski@cavium.com>
30499 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30500 cores and change the partno/implementer to be correct.
30501 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30502 the 'B" as the implementer.
30503 * config/aarch64/aarch64-tune.md: Regenerate.
30505 2017-02-14 Carl Love <cel@us.ibm.com>
30507 * config/rs6000/rs6000.c: Add case statement entry to make the
30508 xvcvuxdsp built-in argument unsigned.
30509 * config/rs6000/vsx.md: Fix the source and return operand types so they
30510 match the instruction definitions from the ISA document. Fix typo
30511 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30514 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30517 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30518 member early_clobber_alts.
30519 * lra-lives.c (reg_early_clobber_p): New.
30520 (process_bb_lives): Use it.
30521 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30522 (debug_operand_data): Initialize early_clobber_alts.
30523 (setup_operand_alternative): Set up early_clobber_alts.
30524 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30525 alternatives to new_insn_reg.
30526 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30528 (lra_update_insn_regno_info): Pass the new arg.
30530 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30532 PR middle-end/79505
30533 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30534 (new_oacc_loop_raw): Don't clear already cleared fields.
30537 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30538 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30539 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30541 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30544 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30546 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30549 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30550 the extra instruction to the right place to store 128-bit constant
30553 2017-02-14 Martin Sebor <msebor@redhat.com>
30555 PR middle-end/79448
30556 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30557 warning for strings of unknown length.
30559 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30561 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30563 2017-02-14 Jeff Law <law@redhat.com>
30566 * ira-costs.c (scan_one_insn): Initialize register move costs
30567 for pseudos seen in USE/CLOBBER insns.
30569 PR tree-optimization/79095
30570 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30571 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30572 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30573 if the operands are known to be not equal, then the resulting range
30575 (intersect_ranges): If the new range is ~[0,0] and the old range is
30576 wide, then prefer ~[0,0].
30577 * tree-vrp.c (overflow_comparison_p_1): New function.
30578 (overflow_comparison_p): New function.
30579 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30580 if NAME is used in an overflow test.
30581 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30582 overflow check that can be expressed as an equality test, then adjust
30583 ops to be that equality test.
30585 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30587 * config/s390/s390-builtin-types.def: Remove flags argument.
30588 * config/s390/s390.c (s390_init_builtins): Likewise.
30590 2017-02-14 Martin Liska <mliska@suse.cz>
30592 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30593 vector. Fix trailing white spaces.
30595 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30597 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30600 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30602 PR rtl-optimization/68664
30603 * config/arm/arm.c (arm_sched_can_speculate_insn):
30604 New function. Declare prototype.
30605 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30607 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30609 PR rtl-optimization/68664
30610 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30612 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30614 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30616 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30617 max skip bytes for function, loop and jump.
30619 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30621 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30622 ABS_EXPR for gimple dump.
30624 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30627 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30629 PR tree-optimization/79408
30630 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30631 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30632 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30633 also if rhs1 is INTEGER_CST.
30635 2017-02-14 Richard Biener <rguenther@suse.de>
30637 PR middle-end/79432
30638 * tree-into-ssa.c (insert_phi_nodes): When the function can
30639 have abnormal edges rewrite SSA names with broken use-def
30640 dominance out of SSA and register them for PHI insertion.
30642 2017-02-13 Martin Sebor <msebor@redhat.com>
30644 PR middle-end/79496
30645 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30646 clearing info.nowrite flag when snprintf size argument is a range.
30648 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30650 * cprop.c (cprop_jump): Add missing space in string literal.
30651 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30652 (get_constraint_for_component_ref): Likewise.
30653 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30654 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30655 * lra-constraints.c (process_alt_operands): Likewise.
30656 * ipa-inline.c (inline_small_functions): Likewise.
30657 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30658 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30659 * trans-mem.c (diagnose_tm_1_op): Likewise.
30660 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30661 (grid_parallel_clauses_gridifiable): Likewise.
30663 * config/nvptx/mkoffload.c (process): Add space in between
30666 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30667 "MOD4_SSE_REGS" and "ALL_REGS".
30669 * spellcheck.c (test_data): Add , in between "foo" and "food".
30671 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30674 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30675 boundary crossing check and subsequent code generation agree.
30677 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30679 * config/aarch64/aarch64.c (has_memory_op): Delete.
30680 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30683 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30685 PR rtl-optimization/79388
30686 PR rtl-optimization/79450
30687 * combine.c (distribute_notes): When removing TEM_INSN for which
30688 corresponding dest has last value recorded, invalidate that last
30691 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30693 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30694 of explicit '@'. Add missing assembly comment marker on branch costs
30697 2017-02-13 Nathan Sidwell <nathan@acm.org>
30699 * gengtype-lex.l (<in_struct>): Add '/'.
30701 2017-02-13 Martin Liska <mliska@suse.cz>
30704 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30706 2017-02-13 Richard Biener <rguenther@suse.de>
30708 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30710 * configure: Re-generate.
30711 * config.in: Likewise.
30712 * graphite-dependences.c: Simplify as if
30713 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30714 * graphite-isl-ast-to-gimple.c: Likewise.
30715 * graphite-optimize-isl.c: Likewise.
30716 * graphite-poly.c: Likewise.
30717 * graphite-sese-to-poly.c: Likewise.
30718 * graphite.h: Likewise.
30719 * toplev.c: Include isl/version.h and use isl_version () for
30720 printing the ISL version.
30721 * doc/install.texi: Update ISL requirement.
30723 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30725 * doc/standards.texi (Standards): Update reference to
30728 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30730 * doc/extend.texi (Named Address Spaces): sourceware.org now
30732 * doc/install.texi (Binaries): Ditto.
30735 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30737 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30738 standard terminology "stringize"/"stringizing" throughout.
30739 * doc/cppinternals.texi: Likewise.
30741 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30743 * doc/extend.texi: Fix some spelling mistakes and typos.
30744 * doc/invoke.texi: Likewise.
30746 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30749 * params.def (inline-min-speedup) Change from 10 to 8.
30751 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30753 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30756 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30759 * ipa-inline-analysis.c (get_minimal_bb): New function.
30760 (record_modified): Use it.
30761 (remap_edge_change_prob): Handle also ancestor functions.
30763 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30765 * doc/contrib.texi (Contributors): Remove broken link into
30766 the Mauve CVS repository.
30768 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30770 PR middle-end/79454
30771 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30772 result computation whenever lhs doesn't have vector mode, not
30773 just when it has BLKmode.
30775 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30777 * doc/makefile.texi (profiledbootstrap): Refer to the
30778 installation instructions only in textual form.
30780 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30783 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30785 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30787 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30788 (Specific): Update mingw-w64 reference.
30790 (Specific): Remove broken link to Renesas RX processor.
30792 2017-02-10 Richard Biener <rguenther@suse.de>
30794 * toplev.c (process_options): Do not mention obsolete graphite
30795 options when printing sorry message about missing graphite support.
30796 Mention -floop-nest-optimize.
30798 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30800 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30801 (vtst_p16): Likewise.
30802 (vtstq_p8): Likewise.
30803 (vtstq_p16): Likewise.
30805 (vtstq_p64): Likewise.
30806 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30807 (vset_lane_p64): New.
30808 (vsetq_lane_p64): New.
30810 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30812 PR tree-optimization/79411
30813 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30814 stmt operands are SSA_NAMEs used in abnormal phis.
30815 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30818 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30821 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30824 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30826 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30828 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30830 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30834 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30835 not arbitrary TREE_CONSTANT.
30838 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30839 "omp declare target link" attribute unless is_global_var.
30840 * omp-offload.c (find_link_var_op): Likewise.
30842 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30843 Chung-Lin Tang <cltang@codesourcery.com>
30845 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30847 (gimplify_adjust_omp_clauses): Don't delete TILE.
30848 (gimplify_omp_for): Deal with TILE.
30849 * internal-fn.c (expand_GOACC_TILE): New function.
30850 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30852 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30853 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30855 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30856 avoid DIV for outermost collapse var.
30857 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30858 Remove out of date comments, fix whitespace.
30859 * omp-general.c (omp_extract_for_data): Deal with tiling.
30860 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30861 adjust OLF_DIM_BASE value.
30862 (struct omp_for_data): Add tiling field.
30863 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30864 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30865 for auto loops. Remove default auto determining, moved to
30866 oacc_loop_fixed_partitions.
30867 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30868 stmts, add e_mask field.
30869 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30870 (oacc_thread_numbers): Use oacc_dim_call.
30871 (oacc_xform_tile): New.
30872 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30873 (finish_oacc_loop): Adjust for ifns vector.
30874 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30875 add case for GOACC_TILE fns.
30876 (oacc_loop_xform_loop): Delete.
30877 (oacc_loop_process): Iterate over call list directly, and add
30878 handling for GOACC_TILE fns.
30879 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30881 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30882 vector partitioning to outer loops. Assign 2 partitions to loops
30883 when available. Add TILE handling.
30884 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30885 (execite_oacc_device_lower): Process GOACC_TILE fns,
30886 ignore unknown specs.
30887 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30888 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30889 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30891 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30893 * configure.ac (ACX_BUGURL): Update.
30894 * configure: Regenerate.
30896 2017-02-09 Richard Biener <rguenther@suse.de>
30898 PR tree-optimization/69823
30899 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30900 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30902 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30904 * config/arc/arc-c.def: Add __NPS400__ definition.
30905 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30906 (TARGET_NPS400): Define.
30908 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30910 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30912 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30913 pointer, arch_info.
30914 (arc_cpu_types): Fill the arch_info field with a pointer into the
30915 arc_arch_types table.
30916 (arc_selected_cpu): Declare.
30917 * config/arc/arc.c (arc_selected_cpu): Make global.
30918 (arc_selected_arch): Delete.
30919 (arc_base_cpu): Delete.
30920 (arc_override_options): Remove references to deleted variables,
30921 update access to arch information.
30922 (ARC_OPT): Update access to arch information.
30923 (ARC_OPTX): Likewise.
30924 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30925 (TARGET_ARC600): Update access to arch information.
30926 (TARGET_ARC601): Likewise.
30927 (TARGET_ARC700): Likewise.
30928 (TARGET_EM): Likewise.
30929 (TARGET_HS): Likewise.
30930 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30933 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30936 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30937 condition/operands for integer GE/LE/GEU/LEU operations.
30939 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30941 PR translation/79397
30942 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30945 2017-02-08 Martin Jambor <mjambor@suse.cz>
30948 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30949 whether allocation happened.
30950 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30951 nothing was allocated.
30953 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30955 PR tree-optimization/79408
30956 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30957 constant, but SSA_NAME with a known integer range, use the minimum
30958 of that range instead of op1 to determine if modulo can be replaced
30959 with its first operand.
30961 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30963 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30965 2017-02-08 Richard Biener <rguenther@suse.de>
30967 PR tree-optimization/71824
30968 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30969 Check all loops contained in the merged region.
30971 2017-02-07 Andrew Pinski <apinski@cavium.com>
30973 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30975 2017-02-07 Andrew Pinski <apinski@cavium.com>
30977 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30978 (thunderxt88): Likewise.
30979 (thunderxt81): Disable LSE and change v8.1 to v8.
30980 (thunderxt83): Likewise.
30982 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30983 Richard Biener <rguenther@suse.de>
30985 PR middle-end/79399
30986 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30987 type from int to size_t.
30988 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30990 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30992 PR rtl-optimization/79386
30993 * cprop.c (bypass_conditional_jumps): Initialize
30994 bypass_last_basic_block already before splitting bbs after
30995 unconditional traps...
30996 (bypass_conditional_jumps): ... rather than here.
30999 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31000 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31001 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31002 fix -masm=intel patterns.
31004 2017-02-07 Richard Biener <rguenther@suse.de>
31006 PR tree-optimization/79256
31007 PR middle-end/79278
31008 * builtins.c (get_object_alignment_2): Use min_align_of_type
31009 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31010 and ADJUST_FIELD_ALIGN.
31012 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31014 * doc/tm.texi: Regenerate.
31015 * stor-layout.c (layout_decl): Adjust.
31016 (update_alignment_for_field): Likewise.
31017 (place_field): Likewise.
31018 (min_align_of_type): Likewise.
31019 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31020 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31021 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31022 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31023 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31024 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31025 * config/i386/i386.c (x86_field_alignment): Likewise.
31026 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31027 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31028 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31029 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31030 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31031 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31035 2017-01-30 Richard Biener <rguenther@suse.de>
31037 PR tree-optimization/79256
31038 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31039 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31042 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31044 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31045 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31046 builtins to SImode and emit a zero-extend, if necessary.
31048 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31050 * docs/invoke.texi (RISC-V Options): Alphabetize.
31052 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31054 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31057 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31059 * config/riscv/riscv.c: New file.
31060 * common/config/riscv/riscv-common.c: Likewise.
31061 * config.gcc: Likewise.
31062 * config/riscv/constraints.md: Likewise.
31063 * config/riscv/elf.h: Likewise.
31064 * config/riscv/generic.md: Likewise.
31065 * config/riscv/linux.h: Likewise.
31066 * config/riscv/multilib-generator: Likewise.
31067 * config/riscv/peephole.md: Likewise.
31068 * config/riscv/pic.md: Likewise.
31069 * config/riscv/predicates.md: Likewise.
31070 * config/riscv/riscv-builtins.c: Likewise.
31071 * config/riscv/riscv-c.c: Likewise.
31072 * config/riscv/riscv-ftypes.def: Likewise.
31073 * config/riscv/riscv-modes.def: Likewise.
31074 * config/riscv/riscv-opts.h: Likewise.
31075 * config/riscv/riscv-protos.h: Likewise.
31076 * config/riscv/riscv.h: Likewise.
31077 * config/riscv/riscv.md: Likewise.
31078 * config/riscv/riscv.opt: Likewise.
31079 * config/riscv/sync.md: Likewise.
31080 * config/riscv/t-elf-multilib: Likewise.
31081 * config/riscv/t-linux: Likewise.
31082 * config/riscv/t-linux-multilib: Likewise.
31083 * config/riscv/t-riscv: Likewise.
31084 * configure.ac: Likewise.
31085 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31086 Waterman as RISC-V maintainers.
31087 * doc/install.texi: Add RISC-V entries.
31088 * doc/invoke.texi: Add RISC-V options section.
31089 * doc/md.texi: Add RISC-V constraints section.
31090 * configure: Regenerated.
31092 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31095 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31096 false values to be constant vectors with all 0 or all 1 bits set.
31097 (vcondu<mode><mode>): Likewise.
31098 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31100 (fpmask_comparison_operator): Update comment.
31101 (vecint_comparison_operator): New predicate.
31102 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31103 vector conditionals when the true and false values are constant
31104 vectors with all 0 bits or all 1 bits set.
31106 2017-02-06 Martin Sebor <msebor@redhat.com>
31108 PR tree-optimization/79376
31109 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31111 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31113 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31114 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31115 to simplify split condition.
31117 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31119 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31120 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31123 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31125 PR rtl-optimization/68664
31126 * target.def (can_speculate_insn): New hook.
31127 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31128 * doc/tm.texi: Regenerate.
31129 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31130 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31131 (rs6000_sched_can_speculate_insn): New function.
31133 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31135 PR tree-optimization/79284
31136 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31137 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31138 vectorizable_mask_load_store, vectorizable_operation,
31139 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31140 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31141 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31142 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31143 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31144 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31145 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31146 is_gimple_assign (stmt). Replace another such test with
31147 is_gimple_assign (stmt).
31149 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31152 * config/avr/avr.c (rtl-iter.h): Include it.
31153 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31154 (avr_legitimate_combined_insn): ...and implementation.
31156 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31158 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31159 * config/s390/s390.c (s390_const_operand_ok)
31160 (s390_canonicalize_comparison, s390_extract_part)
31161 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31162 (s390_contiguous_bitmask_p, s390_rtx_costs)
31163 (legitimize_pic_address): Likewise.
31164 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31165 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31166 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31167 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31168 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31170 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31172 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31173 REGNO($0) == REGNO($1).
31175 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31177 * config/s390/linux.h(SIZE_TYPE): Add comment.
31179 2017-02-06 Julian Brown <julian@codesourcery.com>
31180 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31181 Virendra Pathak <virendra.pathak@broadcom.com>
31183 * config/aarch64/aarch64-cores.def: Change the scheduler
31185 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31186 * config/aarch64/thunderx2t99.md: New file.
31188 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31190 * doc/standards.texi (Go Language): Update link to language
31193 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31195 * tree-eh.c (lower_resx): Sanitize profile.
31196 (cleanup_empty_eh_move_lp): Likewise.
31198 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31201 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31203 * cfgloopmanip.h (loop_version): Update prototype.
31204 * modulo-sched.c (sms_schedule): Update call of loop_version.
31205 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31206 * tree-parloops.c (gen_parallel_loop): Likewise.
31207 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31208 * tree-ssa-loop-split.c (split_loop): Likewise.
31209 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31210 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31212 2017-02-05 Martin Liska <mliska@suse.cz>
31215 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31217 (print_operand_address): Initialize a struct to zero.
31219 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31221 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31222 garbage collector only in textual form.
31224 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31226 * doc/extend.texi (x86 specific memory model extensions for
31227 transactional memory): Simplify a phrase.
31229 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31232 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31233 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31234 (atomic_storedi_1): Likewise.
31236 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31238 PR tree-optimization/79338
31239 * tree-parloops.c (gather_scalar_reductions): Don't call
31240 vect_analyze_loop_form for loop->inner before destroying loop's
31243 2017-02-03 Martin Sebor <msebor@redhat.com>
31245 PR tree-optimization/79327
31246 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31247 when precision has resulted in leading zeros.
31248 (format_integer): Adjust the likely counter to assume an unknown
31249 argument that may be zero is non-zero.
31251 2017-02-03 Jason Merrill <jason@redhat.com>
31254 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31255 avoid copying non-taken branch.
31257 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31259 PR tree-optimization/79340
31260 * tree-vect-loop.c (vectorizable_reduction): Release
31261 vec_defs elements after safe_splicing them into other vectors.
31264 PR tree-optimization/79327
31265 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31266 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31268 (format_integer): Use wide_int_to_tree instead of build_int_cst
31269 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31270 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31271 of shortest and longest sequence.
31273 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31275 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31276 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31278 2017-02-03 Walter Lee <walt@tilera.com>
31281 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31282 after initial stackframe link reg save.
31283 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31285 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31288 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31289 wu for stxssp alternative.
31291 2017-02-03 Martin Sebor <msebor@redhat.com>
31293 PR tree-optimization/79352
31294 * gimple-fold.c (get_range_strlen): Add argument.
31295 (get_range_strlen): Change return type to bool.
31296 (get_maxval_strlen): Pass in a dummy argument.
31297 * gimple-fold.h (get_range_strlen): Change return type to bool.
31298 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31299 * tree.h (array_at_struct_end_p): Add argument.
31300 * tree.c (array_at_struct_end_p): Handle it.
31302 2017-02-03 Martin Liska <mliska@suse.cz>
31305 * multiple_target.c (create_dispatcher_calls): Redirect edge
31306 from a caller of a dispatcher.
31307 (expand_target_clones): Make the clones local.
31308 (ipa_target_clone): Do both target clones and resolvers.
31309 (ipa_dispatcher_calls): Remove the pass.
31310 (pass_dispatcher_calls::gate): Likewise.
31311 (make_pass_dispatcher_calls): Likewise.
31312 * passes.def (pass_target_clone): Put as very first IPA early
31315 2017-02-03 Martin Liska <mliska@suse.cz>
31317 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31318 in case of a function with ifunc attribute.
31320 2017-02-03 Martin Liska <mliska@suse.cz>
31322 * cgraph.c (cgraph_node::dump): Dump function version info.
31323 * symtab.c (symtab_node::dump_base): Add missing new line.
31325 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31327 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31328 (ifcombine_ifandif): Use it.
31330 2017-02-03 Martin Liska <mliska@suse.cz>
31332 * doc/invoke.texi: Document default value for
31333 use-after-scope-direct-emission-threshold.
31335 2017-02-03 Martin Liska <mliska@suse.cz>
31337 PR tree-optimization/79339
31338 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31339 (format_floating): Likewise.
31341 2017-02-03 Martin Liska <mliska@suse.cz>
31344 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31346 (ipa_node_params_t::remove): Likewise.
31347 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31348 initialization from removed ipa_node_params_t::insert.
31349 (ipa_node_params::~ipa_node_params): Move from removed
31350 ipa_node_params_t::release.
31351 * symbol-summary.h (symbol_summary::m_released): New member.
31352 Do not release a summary twice. Do not allow to call finalizer
31353 for types of a summary that live in GGC memory.
31355 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31357 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31360 2017-02-02 Martin Sebor <msebor@redhat.com>
31362 PR middle-end/79275
31363 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31364 (format_string): Tighten up the range of output for non-constant
31365 strings and correct the expected range for wide non-constant strings.
31367 2017-02-02 Martin Sebor <msebor@redhat.com>
31369 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31371 PR middle-end/32003
31372 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31374 (-fdump-tree-@var): Add to index and document how to come up
31375 with pass-specific option and dump file names.
31376 (-fdump-passes): Clarify where to look for output.
31378 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31380 PR middle-end/77445
31381 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31382 statistics of the analyzed path; allow threading for speed when
31383 any of BBs along the path are optimized for speed.
31385 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31387 PR middle-end/78468
31388 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31389 settings of the virtual registers.
31392 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31394 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31395 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31398 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31400 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31401 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31403 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31405 * config/s390/s390.md: Add missing comments with the expanded
31407 * config/s390/vector.md: Likewise.
31408 * config/s390/vx-builtins.md: Likewise.
31410 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31413 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31414 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31415 conditions on a single line.
31417 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31419 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31420 __S390_VX__ to __VX__.
31422 2017-02-01 Andrew Pinski <apinski@cavium.com>
31424 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31425 stmt_info to record_stmt_cost.
31426 (vect_get_known_peeling_cost): Pass stmt_info if known to
31428 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31429 cpu_vector_cost field into
31430 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31431 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31432 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31433 splitting of scalar_stmt_cost and vec_stmt_cost.
31434 (thunderx_vector_cost): Likewise.
31435 (cortexa57_vector_cost): LIkewise.
31436 (exynosm1_vector_cost): Likewise.
31437 (xgene1_vector_cost): Likewise.
31438 (thunderx2t99_vector_cost): Improve after the splitting of the two
31440 (aarch64_builtin_vectorization_cost): Update for the splitting of
31441 scalar_stmt_cost and vec_stmt_cost.
31443 2017-02-01 Torvald Riegel <triegel@redhat.com>
31444 Richard Henderson <rth@redhat.com>
31446 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31447 conditional on existance of a fast atomic load.
31448 * optabs-query.c (can_atomic_load_p): New function.
31449 * optabs-query.h (can_atomic_load_p): Declare it.
31450 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31451 no fast atomic load is available for the particular size of access.
31452 (expand_atomic_compare_and_swap): Likewise.
31453 (expand_atomic_load): Likewise.
31454 (expand_atomic_store): Likewise.
31455 (expand_atomic_fetch_op): Likewise.
31456 * testsuite/lib/target-supports.exp
31457 (check_effective_target_sync_int_128): Remove x86 because it provides
31458 no fast atomic load.
31459 (check_effective_target_sync_int_128_runtime): Likewise.
31461 2017-02-01 Richard Biener <rguenther@suse.de>
31463 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31464 (graphite_transform_loops): Provide opt-info for optimized nests.
31465 * tree-parloop.c (parallelize_loops): Provide opt-info for
31466 parallelized loops.
31468 2017-02-01 Richard Biener <rguenther@suse.de>
31470 PR middle-end/79315
31471 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31472 was not set before.
31474 2017-02-01 Richard Biener <rguenther@suse.de>
31476 PR tree-optimization/71824
31477 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31478 Verify the loops are valid in the merged SESE region.
31479 (scop_detection::can_represent_loop_1): Check analyzing the
31480 evolution of the number of iterations in the region succeeds.
31482 2017-01-31 Ian Lance Taylor <iant@golang.org>
31484 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31485 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31487 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31489 PR preprocessor/79210
31490 * input.c (get_substring_ranges_for_loc): Replace line_width
31491 assertion with error-handling.
31493 2017-01-31 Richard Biener <rguenther@suse.de>
31495 PR tree-optimization/77318
31496 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31497 (create_pw_aff_from_tree): Take loop parameter.
31498 (add_condition_to_pbb): Pass loop of the condition to
31499 create_pw_aff_from_tree.
31501 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31503 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31504 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31506 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31510 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31512 (convert_int_to_float128): Likewise.
31513 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31514 (convert_int_to_float128): Likewise.
31515 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31516 (UNSPEC_IEEE128_CONVERT): Likewise.
31517 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31518 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31519 Use local variables for IBM extended format.
31520 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31521 (fix_trunc<mode>si2_fprs): Likewise.
31522 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31523 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31524 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31525 to know that we can now have integers of all sizes in vector
31527 (fix<uns>_<mode>di2_hw): Likewise.
31528 (float<uns>_<mode>si2_hw): Likewise.
31529 (fix_<mode>si2_hw): Likewise.
31530 (fixuns_<mode>si2_hw): Likewise.
31531 (float<uns>_<mode>di2_hw): Likewise.
31532 (float_<mode>di2_hw): Likewise.
31533 (float_<mode>si2_hw): Likewise.
31534 (floatuns_<mode>di2_hw): Likewise.
31535 (floatuns_<mode>si2_hw): Likewise.
31536 (xscvqp<su>wz_<mode>): Delete, no longer used.
31537 (xscvqp<su>dz_<mode>): Likewise.
31538 (xscv<su>dqp_<mode>): Likewise.
31539 (ieee128_mfvsrd_64bit): Likewise.
31540 (ieee128_mfvsrd_32bit): Likewise.
31541 (ieee128_mfvsrwz): Likewise.
31542 (ieee128_mtvsrw): Likewise.
31543 (ieee128_mtvsrd_64bit): Likewise.
31544 (ieee128_mtvsrd_32bit): Likewise.
31546 2017-01-31 Martin Liska <mliska@suse.cz>
31549 * ipa-prop.c (ipa_free_all_node_params): Call release method
31550 instead of ~sumbol_summary to not to trigger double times
31553 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31555 PR tree-optimization/71691
31556 * bitmap.h (class auto_bitmap): New.
31557 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31558 is_maybe_undefined instead of ssa_undefined_value_p.
31560 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31562 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31563 __S390_ARCH_LEVEL__ to __ARCH__.
31565 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31567 PR tree-optimization/79267
31568 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31569 if should_remove_lhs_p is true.
31571 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31574 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31575 (add_alignment_attribute): New.
31576 (base_type_die): Add alignment attribute.
31577 (subrange_type_die): Likewise.
31578 (modified_type_die): Likewise.
31579 (gen_array_type_die): Likewise.
31580 (gen_descr_array_type_die: Likewise.
31581 (gen_enumeration_type_die): Likewise.
31582 (gen_subprogram_die): Likewise.
31583 (gen_variable_die): Likewise.
31584 (gen_field_die): Likewise.
31585 (gen_ptr_to_mbr_type_die): Likewise.
31586 (gen_struct_or_union_type_die): Likewise.
31587 (gen_subroutine_type_die): Likewise.
31588 (gen_typedef_die): Likewise.
31589 (base_type_cmp): Compare alignment attribute.
31591 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31594 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31595 (setb_unsigned) New pattern for setb with CCUNS.
31596 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31597 subfc./subfe sequence to avoid overflow problems. Generate a
31598 shorter sequence with cmpld/setb for power9.
31599 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31600 for generating subfc. instruction.
31601 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31602 now uses this instruction.
31604 2017-01-30 Ian Lance Taylor <iant@google.com>
31607 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31608 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31610 2017-01-30 Martin Sebor <msebor@redhat.com>
31612 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31613 Move constant to the right of a relational operator.
31614 (get_mpfr_format_length, format_character, format_string): Ditto.
31615 (should_warn_p, maybe_warn): Same.
31617 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31619 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31622 * asan.c (get_translation_unit_decl): Remove function.
31623 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31625 2017-01-30 Martin Liska <mliska@suse.cz>
31627 PR gcov-profile/79259
31628 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31629 -fprofile-generate.
31631 2017-01-30 Martin Liska <mliska@suse.cz>
31634 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31635 Initialize variables with NULL value.
31637 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31640 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31642 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31644 2017-01-30 Richard Biener <rguenther@suse.de>
31646 PR tree-optimization/79276
31647 * tree-vrp.c (process_assert_insertions): Properly adjust common
31648 when removing a duplicate.
31650 2017-01-30 Richard Biener <rguenther@suse.de>
31652 PR tree-optimization/79256
31653 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31654 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31656 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31658 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31661 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31662 ("*r<noxa>sbg_<mode>_sll_bitmask")
31663 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31664 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31665 Use contiguous_bitmask_nowrap_operand.
31667 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31670 * config/rs6000/altivec.h (vec_xl): Revise #define.
31671 (vec_xst): Likewise.
31673 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31675 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31677 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31679 PR rtl-optimization/79194
31680 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31681 traps before call to bypass_conditional_jumps.
31683 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31685 PR tree-optimization/71374
31686 * lra-constraints.c (check_conflict_input_operands): New.
31687 (match_reload): Use it.
31689 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31692 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31693 account to calculate conflict_set.
31695 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31697 PR rtl-optimization/78559
31698 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31699 other_insn in combine.
31701 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31703 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31704 uint16_type_node for BT_UINT16.
31706 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31708 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31709 "RTL Tests" to menu.
31710 (GIMPLE Tests): New node.
31711 (RTL Tests): New node.
31713 2017-01-27 Richard Biener <rguenther@suse.de>
31715 PR tree-optimization/79245
31716 * tree-loop-distribution.c (distribute_loop): Apply cost
31717 modeling also to detected patterns.
31719 2017-01-27 Richard Biener <rguenther@suse.de>
31721 PR tree-optimization/71433
31722 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31723 (compare_assert_loc): New function.
31724 (process_assert_insertions): Sort and optimize assert locations
31725 to remove duplicates and push down identical assertions on
31726 edges to their destination block.
31728 2017-01-27 Richard Biener <rguenther@suse.de>
31730 PR tree-optimization/79244
31731 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31732 out SSA names even if abnormal.
31734 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31736 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31737 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31738 instead of MPFR_RNDN.
31740 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31743 * arm.c (arm_option_override): Don't call build_target_option_node
31744 until after doing all option overrides.
31745 (arm_valid_target_attribute_tree): Likewise.
31747 2017-01-27 Martin Liska <mliska@suse.cz>
31749 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31750 for {cd}tors and C++ {cd}tors.
31752 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31754 * config/s390/s390.md ("*setmem_long_and")
31755 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31757 2017-01-26 Martin Sebor <msebor@redhat.com>
31759 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31762 2017-01-26 Martin Sebor <msebor@redhat.com>
31764 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31765 HAVE_DFmode before using XFmode or DFmode.
31766 (parse_directive): Avoid using the z length modifier to avoid
31767 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31769 PR middle-end/78703
31770 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31771 to accept adjustment as an array.
31772 (get_int_range): New function.
31773 (struct directive): Make width and prec arrays.
31774 (directive::set_width, directive::set_precision): Call get_int_range.
31775 (format_integer, format_floating): Handle width and precision ranges.
31776 (format_string, parse_directive): Same.
31778 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31781 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31782 comdat_type_p set, just clone them, but keep the children in the
31786 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31787 which have direct callers with -fvar-tracking-assignments enabled
31789 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31790 inside of type units.
31792 2017-01-26 Martin Sebor <msebor@redhat.com>
31794 PR middle-end/78703
31795 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31797 (struct format_result): Replace number_chars, number_chars_min,
31798 and number_chars_max with a single member of struct result_range.
31800 (format_result::operator+=): Adjust.
31801 (struct fmtresult): Remove bounded. Handle likely and unlikely
31803 (fmtresult::adjust_for_width_or_precision): New function.
31804 (fmtresult:type_max_digits): New function.
31805 (bytes_remaining): Handle likely and unlikely counters.
31806 (min_bytes_remaining): Remove.
31807 (format_percent): Simplify.
31808 (format_integer, format_floating): Set likely and unlikely counters.
31809 (get_string_length, format_character, format_string): Same.
31810 (format_plain, should_warn_p): New function.
31811 (maybe_warn): Call should_warn_p. Update diagnostic messages
31812 and handle those for all directives, including plain strings.
31813 (format_directive): Handle likely and unlikely counters.
31814 Remove unnecessary quoting from diagnostics. Add an informational
31816 (add_bytes): Remove.
31817 (pass_sprintf_length::compute_format_length): Simplify.
31818 (try_substitute_return_value): Handle likely and unlikely counters.
31820 2017-01-26 Carl Love <cel@us.ibm.com>
31822 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31823 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31825 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31828 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31829 endianess for subregs into account.
31830 * lra-constraints.c (lra_constraints): Do risky transformations
31831 always on the first iteration.
31832 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31833 last_call_used_reg_set.
31834 (process_bb_lives): Define and use last_call_used_reg_set.
31835 * lra.c (lra): Always continue after lra_constraints on the first
31838 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31840 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31842 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31844 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31846 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31847 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31848 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31849 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31850 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31851 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31852 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31853 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31854 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31856 2017-01-26 Marek Polacek <polacek@redhat.com>
31859 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31860 for the third operand.
31862 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31864 PR middle-end/79236
31865 * omp-low.c (struct omp_context): Add simt_stmt field.
31866 (scan_omp_for): Return omp_context *.
31867 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31868 context to the _simt_ SIMD stmt.
31869 (lower_omp_for): For combined SIMD with sibling _simt_
31870 SIMD, make sure to use the same decls in _looptemp_
31871 clauses as in the sibling.
31873 2017-01-26 David Sherwood <david.sherwood@arm.com>
31875 PR middle-end/79212
31876 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31879 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31882 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31883 emit fld b; fld a; if possible.
31885 * brig-builtins.def: Update copyright years.
31886 * config/arm/arm_acle_builtins.def: Update copyright years.
31888 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31891 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31892 constraint instead of o for the stxsd instruction.
31894 2017-01-25 Carl Love <cel@us.ibm.com>
31896 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31897 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31899 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31901 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31903 2017-01-25 Richard Biener <rguenther@suse.de>
31905 PR tree-optimization/69264
31906 * target.def (vector_alignment_reachable): Improve documentation.
31907 * doc/tm.texi: Regenerate.
31908 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31910 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31911 earlier changes with respect to TYPE_USER_ALIGN.
31912 (vector_alignment_reachable_p): Likewise. Improve dumping.
31914 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31917 * config/arm/arm.md (xordi3): Force constant operand into a register
31920 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31922 * doc/invoke.texi (-fstore-merging): Correct default optimization
31923 levels at which it is enabled.
31924 (-O): Move -fstore-merging from list to...
31927 2017-01-25 Richard Biener <rguenther@suse.de>
31930 * omp-expand.c: Include debug.h.
31931 (expand_omp_taskreg): Make sure to generate early debug before
31932 outlining anything from a function.
31933 (expand_omp_target): Likewise.
31934 (grid_expand_target_grid_body): Likewise.
31936 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31939 * asan.c (get_translation_unit_decl): New function.
31940 (asan_add_global): Extract modules file name from globals
31941 TRANSLATION_UNIT_DECL name.
31943 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31946 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31947 for long calls with APCS frame and VFP.
31949 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31951 * cfg.c (original_copy_tables_initialized_p): New function.
31952 * cfg.h (original_copy_tables_initialized_p): New decl.
31953 * cfgrtl.c (relink_block_chain): Guard the call to
31954 free_original_copy_tables with a call to
31955 original_copy_tables_initialized_p.
31956 * cgraph.h (symtab_node::native_rtl_p): New decl.
31957 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31958 (symtab_node::needed_p): Don't assert for early assembly output
31959 for __RTL functions.
31960 (cgraph_node::finalize_function): Set "force_output" for __RTL
31962 (cgraph_node::analyze): Bail out early for __RTL functions.
31963 (analyze_functions): Update assertion to support __RTL functions.
31964 (cgraph_node::expand): Bail out early for __RTL functions.
31965 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31967 * function.h (struct function): Update comment for field
31969 * gimple-expr.c: Include "tree-pass.h".
31970 (gimple_has_body_p): Return false for __RTL functions.
31971 * Makefile.in (OBJS): Add run-rtl-passes.o.
31972 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31974 (gcc::pass_manager::get_clean_slate): New accessor.
31975 * passes.c: Include "insn-addr.h".
31976 (should_skip_pass_p): Add logging. Update logic for running
31977 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31978 property-provider override so it is only done for gimple passes.
31980 (skip_pass): New function.
31981 (execute_one_pass): Call skip_pass when skipping passes.
31982 * read-md.c (md_reader::read_char): Support filtering
31983 the input to a subset of line numbers.
31984 (md_reader::md_reader): Initialize fields
31985 m_first_line and m_last_line.
31986 (md_reader::read_file_fragment): New function.
31987 * read-md.h (md_reader::read_file_fragment): New decl.
31988 (md_reader::m_first_line): New field.
31989 (md_reader::m_last_line): New field.
31990 * read-rtl-function.c (function_reader::create_function): Only
31991 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31992 curr_properties. Set DECL_INITIAL to a dummy block.
31993 (read_rtl_function_body_from_file_range): New function.
31994 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31996 * run-rtl-passes.c: New file.
31997 * run-rtl-passes.h: New file.
31999 2017-01-24 Jeff Law <law@redhat.com>
32001 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32004 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32006 PR tree-optimization/79159
32007 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32008 (record_nonwrapping_iv): Improve boundary using above function if no
32009 value range information.
32011 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32012 Martin Jambor <mjambor@suse.cz>
32014 * brig-builtins.def: New file.
32015 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32016 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32017 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32018 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32019 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32020 * builtin-types.def (BT_INT8): New.
32021 (BT_INT16): Likewise.
32022 (BT_UINT8): Likewise.
32023 (BT_UINT16): Likewise.
32024 (BT_FN_ULONG): Likewise.
32025 (BT_FN_UINT_INT): Likewise.
32026 (BT_FN_UINT_ULONG): Likewise.
32027 (BT_FN_UINT_LONG): Likewise.
32028 (BT_FN_UINT_PTR): Likewise.
32029 (BT_FN_ULONG_PTR): Likewise.
32030 (BT_FN_INT8_FLOAT): Likewise.
32031 (BT_FN_INT16_FLOAT): Likewise.
32032 (BT_FN_UINT32_FLOAT): Likewise.
32033 (BT_FN_UINT16_FLOAT): Likewise.
32034 (BT_FN_UINT8_FLOAT): Likewise.
32035 (BT_FN_UINT64_FLOAT): Likewise.
32036 (BT_FN_UINT16_UINT32): Likewise.
32037 (BT_FN_UINT32_UINT16): Likewise.
32038 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32039 (BT_FN_INT_PTR_INT): Likewise.
32040 (BT_FN_UINT_PTR_UINT): Likewise.
32041 (BT_FN_LONG_PTR_LONG): Likewise.
32042 (BT_FN_ULONG_PTR_ULONG): Likewise.
32043 (BT_FN_VOID_UINT64_UINT64): Likewise.
32044 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32045 (BT_FN_INT8_INT8_INT8): Likewise.
32046 (BT_FN_INT16_INT16_INT16): Likewise.
32047 (BT_FN_INT_INT_INT): Likewise.
32048 (BT_FN_UINT_FLOAT_UINT): Likewise.
32049 (BT_FN_FLOAT_UINT_UINT): Likewise.
32050 (BT_FN_ULONG_UINT_UINT): Likewise.
32051 (BT_FN_ULONG_UINT_PTR): Likewise.
32052 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32053 (BT_FN_UINT_UINT_UINT): Likewise.
32054 (BT_FN_VOID_UINT_PTR): Likewise.
32055 (BT_FN_UINT_UINT_PTR: Likewise.
32056 (BT_FN_UINT32_UINT64_PTR): Likewise.
32057 (BT_FN_INT_INT_UINT_UINT): Likewise.
32058 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32059 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32060 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32061 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32062 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32063 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32064 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32065 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32066 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32067 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32068 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32069 * doc/frontends.texi: List BRIG FE.
32070 * doc/install.texi (Testing): Add BRIG tesring requirements.
32071 * doc/invoke.texi (Overall Options): Mention BRIG.
32072 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32074 2017-01-24 Richard Biener <rguenther@suse.de>
32076 PR translation/79208
32077 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32079 2017-01-24 Martin Jambor <mjambor@suse.cz>
32082 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32083 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32084 and known_contexts.
32086 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32088 PR middle-end/79123
32089 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32090 casts from signed to unsigned really don't have a range.
32092 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32094 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32095 GMP_RNDx for compatiblity.
32097 2017-01-24 Martin Liska <mliska@suse.cz>
32100 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32101 that would prevent us to call alloca with -1 as argument.
32103 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32105 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32106 Avoid -Wformat-security warning.
32108 2017-01-23 Andrew Pinski <apinski@cavium.com>
32110 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32113 2017-01-23 Martin Sebor <msebor@redhat.com>
32115 PR middle-end/78703
32116 * gimple-ssa-sprintf.c (warn_level): New global.
32117 (format_integer): Use it here and throughout the rest of the file.
32118 Use the same switch to compute sign as base.
32119 (maybe_warn): New function.
32120 (format_directive): Factor out warnings into maybe_warn.
32121 Add debugging output. Use warn_level.
32122 (add_bytes): Use warn_level.
32123 (pass_sprintf_length::compute_format_length): Add debugging output.
32124 (try_substitute_return_value): Same.
32125 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32127 PR middle-end/78703
32128 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32129 (struct fmtresult, format_integer, format_floating): Adjust.
32130 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32131 (get_string_length, format_string,format_directive): Same.
32132 (pass_sprintf_length::compute_format_length): Same.
32133 (try_substitute_return_value): Simplify slightly.
32135 PR middle-end/78703
32136 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32137 (fmtresult::operator+=): Outlined.
32138 (struct fmtresult): Add ctors.
32139 (struct conversion_spec): Rename...
32140 (struct directive): ...to this. Add and remove data members.
32141 (directive::set_width, directive::set_precision): New functions.
32142 (format_percent): Use fmtresult ctor.
32143 (get_width_and_precision): Remove.
32144 (format_integer): Make naming changes. Avoid computing width and
32146 (format_floating): Same. Adjust indentation.
32147 (format_character, format_none): New functions.
32148 (format_string): Moved character handling to format_character.
32149 (format_directive): Remove arguments, change return type.
32150 (parse_directive): New function.
32151 (pass_sprintf_length::compute_format_length): Move directive
32152 parsing to parse_directive.
32154 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32156 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32157 (assign_assembler_name_if_needed): ... this.
32158 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32159 (assign_assembler_name_if_needed): ... this.
32160 (free_lang_data_in_cgraph): Adjust callers.
32161 * cgraphunit.c (cgraph_node::analyze): Likewise.
32162 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32165 2017-01-23 Richard Biener <rguenther@suse.de>
32167 PR tree-optimization/79088
32168 PR tree-optimization/79188
32169 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32170 resetting loop bounds after last path deletion. Reset loop
32171 bounds of the target loop, make code match the comments.
32172 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32173 Make sure loops need no fixups.
32175 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32177 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32178 exponent support with double type for first argument.
32179 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32180 type returned by __builtin_vec_extract_sig,
32181 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32182 functions from "vector int" to "vector unsigned int" or from
32183 "vector long long int" to "vector unsigned long long int".
32184 Changed type returned by __builtin_vec_extract_exp,
32185 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32186 functions from "vector int" to "vector unsigned int" or from
32187 "vector long long int" to "vector unsigned long long int".
32188 Changed return type of __builtin_vec_test_data_class,
32189 __builtin_vec_test_data_class_sp, and
32190 __builtin_vec_test_data_class_dp from "vector int" to
32191 "vector bool int" or from "vector long long int" to "vector bool
32192 long long int" and changed second argument type from "unsigned
32193 int" to "int". Added new overloaded function forms "vector float
32194 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32195 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32196 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32197 double, vector unsigned long long int)" and "vector double
32198 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32199 long int)". Changed return type of
32200 __builtin_scalar_test_data_class and
32201 __builtin_scalar_test_data_class_sp and
32202 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32203 int" and changed second argument from "unsigned int" to "int".
32204 Changed type returned by __builtin_scalar_test_neg,
32205 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32206 from "int" to "bool int". Added new overloaded function form
32207 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32208 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32209 exponent double-precision with floating point first argument.
32210 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32211 documentation of scalar_test_data_class, scalar_test_neg,
32212 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32213 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32214 vec_test_data_class built-in functions to reflect refinements in
32215 their type signatures.
32217 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32219 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32221 (aarch64_elf_asm_destructor): Likewise.
32223 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32225 PR rtl-optimization/78634
32226 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32227 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32228 * ifcvt.c (noce_try_cmove): Add missing cost check.
32230 PR rtl-optimization/71724
32231 * combine.c (if_then_else_cond): Look for situations where it is
32232 beneficial to undo the work of one of the recursive calls.
32234 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32236 PR tree-optimization/70754
32237 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32238 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32239 combined stmt before it if not NULL.
32240 (combine_chains): Process refs reversely and compute dominance point
32243 2017-01-23 Martin Liska <mliska@suse.cz>
32245 PR tree-optimization/79196
32246 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32247 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32249 (strlen_optimize_stmt): Call the renamed function.
32251 2017-01-23 Michael Matz <matz@suse.de>
32253 PR tree-optimization/78384
32254 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32256 2017-01-23 Richard Biener <rguenther@suse.de>
32258 PR tree-optimization/79186
32259 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32260 both incoming edges before moving an assert.
32262 2017-01-23 Martin Jambor <mjambor@suse.cz>
32264 * ipa-prop.c (load_from_param_1): Removed.
32265 (load_from_unmodified_param): Bits from load_from_param_1 put back
32267 (load_from_param): Removed.
32268 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32269 with stmt. Reverted back to use of load_from_unmodified_param.
32271 2017-01-23 Martin Jambor <mjambor@suse.cz>
32274 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32275 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32276 field a pointer to garbage collected vector, mark lattices and
32277 ipcp_orig_node with GTY((skip)).
32278 (ipa_get_param_count): Adjust to descriptors being a pointer.
32279 (ipa_get_param): Likewise.
32280 (ipa_get_type): Likewise.
32281 (ipa_get_param_move_cost): Likewise.
32282 (ipa_set_param_used): Likewise.
32283 (ipa_get_controlled_uses): Likewise.
32284 (ipa_set_controlled_uses): Likewise.
32285 (ipa_is_param_used): Likewise.
32286 (ipa_node_params_t): Move into garbage collector. New methods insert
32288 (ipa_node_params_sum): Annotate wth GTY(()).
32289 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32291 (ipa_load_from_parm_agg): Adjust declaration.
32292 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32293 * ipa-profile.c (ipa_profile): Likewise.
32294 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32295 (ipa_populate_param_decls): Make descriptors parameter garbage
32297 (ipa_dump_param): Adjust to descriptors being a pointer.
32298 (ipa_alloc_node_params): Likewise.
32299 (ipa_initialize_node_params): Likewise.
32300 (load_from_param_1): Make descriptors parameter garbage collected.
32301 (load_from_unmodified_param): Likewise.
32302 (load_from_param): Likewise.
32303 (ipa_load_from_parm_agg): Likewise.
32304 (ipa_node_params::~ipa_node_params): Removed.
32305 (ipa_free_all_node_params): Remove call to delete operator.
32306 (ipa_node_params_t::insert): New.
32307 (ipa_node_params_t::remove): Likewise.
32308 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32309 copy known_csts and known_contexts vectors.
32310 (ipa_read_node_info): Adjust to descriptors being a pointer.
32311 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32313 (ipcp_transform_function): Make descriptors variable garbage
32316 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32318 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32319 * config/i386/avx512dqintrin.h: Ditto.
32320 * config/i386/avx512fintrin.h: Ditto.
32321 * config/i386/i386.c: Handle new builtins.
32322 * config/i386/i386-builtin.def: Add new builtins.
32323 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32324 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32326 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32327 Martin Liska <mliska@suse.cz>
32329 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32330 * asan.c (asan_expand_poison_ifn): Support stores and use
32331 appropriate ASAN report function.
32332 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32333 * internal-fn.def (ASAN_POISON_USE): Declare.
32334 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32335 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32336 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32337 ASAN_POISON calls w/o LHS.
32338 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32339 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32340 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32341 * gimplify.c (asan_poison_variables): Add attribute
32342 use_after_scope_memory to variables that really needs to live
32344 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32345 having the attribute.
32347 2017-01-23 Martin Liska <mliska@suse.cz>
32349 * asan.c (create_asan_shadow_var): New function.
32350 (asan_expand_poison_ifn): Likewise.
32351 * asan.h (asan_expand_poison_ifn): New declaration.
32352 * internal-fn.c (expand_ASAN_POISON): Likewise.
32353 * internal-fn.def (ASAN_POISON): New builtin.
32354 * sanopt.c (pass_sanopt::execute): Expand
32355 asan_expand_poison_ifn.
32356 * tree-inline.c (copy_decl_for_dup_finish): Make function
32358 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32359 * tree-ssa.c (is_asan_mark_p): New function.
32360 (execute_update_addresses_taken): Rewrite local variables
32361 (identified just by use-after-scope as addressable) into SSA.
32363 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32365 * doc/install.texi (Specific): opensource.apple.com uses https
32366 now. Remove trailing slash.
32368 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32370 * README.Portability: Remove note on an Irix compatibility issue.
32372 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32374 * gcov.c (INCLUDE_ALGORITHM): Define.
32375 (INCLUDE_VECTOR): Define.
32376 No longer include <vector> and <algorithm> directly.
32378 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32380 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32382 * doc/invoke.texi (Code Gen Options): Ditto.
32384 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32387 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32389 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32391 rtl-optimization/79125
32392 * cprop.c (local_cprop_pass): Handle cases where we make an
32393 unconditional trap.
32395 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32399 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32400 read from, for big endian.
32402 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32404 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32405 register pauth builtins for LP64 only.
32407 2017-01-20 Marek Polacek <polacek@redhat.com>
32410 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32413 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32415 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32417 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32418 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32419 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32421 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32424 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32425 in big-endian mode when they are not a single duplicated value.
32427 2017-01-20 Richard Biener <rguenther@suse.de>
32429 * BASE-VER: Bump to 7.0.1.
32431 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32433 * omp-low.c (omplow_simd_context): New struct. Use it...
32434 (lower_rec_simd_input_clauses): ...here and...
32435 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32436 references to idx, lane, max_vf, is_simt.
32438 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32440 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32443 2017-01-20 Martin Jambor <mjambor@suse.cz>
32445 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32446 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32448 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32449 (GTFILES): Rename hsa.c to hsa-common.c.
32450 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32451 * hsa-dump.c: Likewise.
32452 * hsa-gen.c: Likewise.
32453 * hsa-regalloc.c: Likewise.
32454 * ipa-hsa.c: Likewise.
32455 * omp-expand.c: Likewise.
32456 * omp-low.c: Likewise.
32457 * toplev.c: Likewise.
32459 2017-01-20 Marek Polacek <polacek@redhat.com>
32462 * doc/invoke.texi: Document -Wduplicated-branches.
32463 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32464 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32465 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32466 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32467 return 0 only when not OEP_LEXICOGRAPHIC.
32468 (fold_build_cleanup_point_expr): Use the expression
32469 location when building CLEANUP_POINT_EXPR.
32470 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32471 * tree.c (add_expr): Handle error_mark_node.
32473 2017-01-20 Martin Liska <mliska@suse.cz>
32476 * tree-profile.c (init_ic_make_global_vars): Do not call
32478 (gimple_init_gcov_profiler): Likewise.
32480 2017-01-20 Martin Liska <mliska@suse.cz>
32483 * cgraph.h (maybe_create_reference): Remove argument and
32485 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32487 * ipa-cp.c (create_specialized_node): Likewise.
32488 * symtab.c (symtab_node::maybe_create_reference): Handle
32489 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32491 2017-01-20 Martin Liska <mliska@suse.cz>
32493 * read-rtl-function.c (function_reader::create_function): Use
32494 build_decl instread of build_decl_stat.
32496 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32498 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32499 * config/i386/avx512dqintrin.h: Ditto.
32500 * config/i386/avx512fintrin.h: Ditto.
32501 * config/i386/i386-builtin-types.def: Add new types.
32502 * config/i386/i386.c: Handle new types.
32503 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32504 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32505 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32506 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32507 (__builtin_ia32_kshiftridi): New.
32508 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32510 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32514 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32515 define to rs6000_init_stack_protect_guard.
32516 (rs6000_init_stack_protect_guard): New function.
32518 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32519 Yunqiang Su <yunqiang.su@imgtec.com>
32521 * config.gcc (supported_defaults): Add madd4.
32522 (with_madd4): Add validation.
32523 (all_defaults): Add madd4.
32524 * config/mips/mips.opt (mmadd4): New option.
32525 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32527 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32529 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32530 (ISA_HAS_FUSED_MADD4): Likewise.
32531 * doc/invoke.texi (-mmadd4): Document the new option.
32532 * doc/install.texi (--with-madd4): Document the new option.
32534 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32536 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32537 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32538 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32539 (aarch64_init_pauth_hint_builtins): New.
32540 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32541 (aarch64_expand_builtin): Expand new builtins.
32543 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32545 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32546 * combine-stack-adj.c (no_unhandled_cfa): Handle
32547 REG_CFA_TOGGLE_RA_MANGLE.
32548 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32549 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32550 info for return address signing.
32551 (aarch64_expand_epilogue): Likewise.
32553 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32555 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32556 * config/aarch64/aarch64-protos.h
32557 (aarch64_return_address_signing_enabled): New declaration.
32558 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32560 (aarch64_expand_prologue): Sign return address before it's pushed onto
32562 (aarch64_expand_epilogue): Authenticate return address fetched from
32564 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32565 (aarch64_attributes): New function attributes for "sign-return-address".
32566 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32567 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32568 ("*do_return"): Generate combined instructions according to key index.
32569 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32570 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32572 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32573 * config/aarch64/aarch64.opt (msign-return-address=): New.
32574 * doc/extend.texi (AArch64 Function Attributes): Documents
32575 "sign-return-address=".
32576 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32578 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32580 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32581 overall option summary.
32583 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32585 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32586 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32587 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32588 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32590 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32592 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32593 -mpower9-minmax by default for -mcpu=power9.
32594 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32595 128-bit floating point.
32597 2017-01-20 Alan Modra <amodra@gmail.com>
32599 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32600 optimizing for size.
32602 2017-01-20 Alan Modra <amodra@gmail.com>
32605 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32606 for strcmp and strncmp from corresponding builtin decl.
32608 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32610 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32611 instead of i386/rtems-64.h.
32612 * config/i386/rtems-64.h: Remove.
32614 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32618 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32620 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32622 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32624 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32625 Change int to HOST_WIDE_INT.
32626 * config/aarch64/aarch64-protos.h
32627 (aarch64_simd_gen_const_vector_dup): Likewise.
32628 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32630 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32632 * langhooks-def.h (lhd_type_for_size): New decl.
32633 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32634 * langhooks.c (lhd_type_for_size): New function, taken from
32637 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32639 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32640 define_bypass for CR latency.
32641 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32642 (power9-alu2): Add define_bypass for CR latency.
32644 (power9-mul): Update insn latency.
32645 (power9-mul-compare): Update insn latency, bypass latency and remove
32648 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32650 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32652 * config/aarch64/aarch64.md
32653 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32654 aarch64_nopcrelative_literal_loads.
32655 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32657 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32659 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32660 TARGET_LOONGSON_3A.
32661 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32663 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32666 * config.gcc (supported_defaults): Add lxc1-sxc1.
32667 (with_lxc1_sxc1): Add validation.
32668 (all_defaults): Add lxc1-sxc1.
32669 * config/mips/mips.opt (mlxc1-sxc1): New option.
32670 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32672 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32673 __mips_no_lxc1_sxc1.
32674 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32675 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32676 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32678 2017-01-19 Richard Biener <rguenther@suse.de>
32680 PR tree-optimization/72488
32681 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32682 sure to restore SSA info.
32683 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32685 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32687 PR rtl-optimization/79121
32688 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32689 of the inner type when shifting an extended value.
32691 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32694 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32695 interposable aliases.
32697 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32700 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32701 Use the evmergelohi instruction.
32702 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32703 (mov_sitf_e500_subreg8_2_be): Likewise.
32704 (mov_sitf_e500_subreg12_2_le): Likewise.
32705 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32706 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32707 (mov_sitf_e500_subreg8_2_le): Likewise.
32708 (mov_sitf_e500_subreg12_2_be): Likewise.
32710 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32712 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32713 attribute from vecsimple to vecperm.
32714 (altivec_vbpermq2): Likewise.
32716 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32719 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32721 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32722 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32723 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32724 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32725 case where N arg is SIZE_MAX.
32726 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32727 (cmpstrsi): Add pattern.
32729 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32731 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32732 __builtin_vec_revb builtins.
32733 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32734 built-in functions to support generation of the ISA 3.0 XXBR<x>
32735 vector byte reverse instructions.
32736 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32737 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32738 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32739 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32740 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32741 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32742 (P9V_BUILTIN_VEC_REVB): Likewise.
32743 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32744 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32745 (p9_xxbrq_v16qi): Likewise.
32746 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32747 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32748 (p9_xxbrh_v8hi): Likewise.
32749 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32750 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32751 vec_revb built-in functions.
32753 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32755 PR rtl-optimization/78952
32756 * config/i386/i386.md (any_extract): New code iterator.
32757 (*insvqi_2): Use any_extract for source operand.
32758 (*insvqi_3): Use any_shiftrt for source operand.
32760 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32762 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32764 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32766 2017-01-18 Matthias Klose <doko@ubuntu.com>
32768 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32770 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32772 * config/rs6000/altivec.h (vec_bperm): Change #define.
32773 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32774 (altivec_vbpermq2): New define_insn.
32775 (altivec_vbpermd): Likewise.
32776 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32777 function interface.
32778 (VBPERMD): Likewise.
32779 (VBPERM): New polymorphic function interface.
32780 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32781 Add entries for P9V_BUILTIN_VEC_VBPERM.
32782 * doc/extend.texi: Add interfaces for vec_bperm.
32784 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32786 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32787 first letter of error messages.
32788 (s390_resolve_overloaded_builtin): Likewise.
32789 * config/s390/s390.c (s390_expand_builtin): Likewise.
32790 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32791 (s390_valid_target_attribute_inner_p): Likewise.
32792 * config/s390/s390.md ("tabort"): Likewise.
32794 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32796 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32797 (ISA_AVOID_DIV_HILO): New macro.
32798 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32799 (ISA_HAS_DDIV): Likewise.
32801 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32803 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32805 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32807 * doc/invoke.texi (fabi-version): Spelling fix.
32809 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32812 * doc/invoke.texi (fabi-version): Mention mangling fix for
32815 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32818 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32820 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32823 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32824 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32826 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32828 (stack_protect_test): Ditto.
32829 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32830 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32832 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32833 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32834 -mstack-protector-guard-offset=.
32835 (RS/6000 and PowerPC Options): Ditto.
32837 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32839 * config/i386/i386.h (MASK_CLASS_P): New define.
32840 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32841 there are no registers from different register sets also when
32842 mask registers are used. Update function comment.
32843 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32844 to (*k/*r) and (*k/*km) alternatives.
32846 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32848 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32849 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32850 (EH_RETURN_HANDLER_RTX): New define.
32851 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32852 Force frame pointer in EH return functions.
32853 (aarch64_expand_epilogue): Add barrier for eh_return.
32854 (aarch64_final_eh_return_addr): Remove.
32855 (aarch64_eh_return_handler_rtx): New function.
32856 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32858 (aarch64_eh_return_handler_rtx): New prototype.
32860 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32862 * config/rs6000/altivec.h (vec_rlmi): New #define.
32863 (vec_vrlnm): Likewise.
32864 (vec_rlnm): Likewise.
32865 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32866 (UNSPEC_VRLNM): Likewise.
32867 (VIlong): New mode iterator.
32868 (altivec_vrl<VI_char>mi): New define_insn.
32869 (altivec_vrl<VI_char>nm): Likewise.
32870 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32872 (VRLDNM): Likewise.
32873 (RLNM): New polymorphic function entry.
32874 (VRLWMI): New monomorphic function entry.
32875 (VRLDMI): Likewise.
32876 (RLMI): New polymorphic function entry.
32877 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32878 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32879 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32882 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32885 * dwarf2out.c (field_byte_offset): Restore the
32886 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32887 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32888 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32891 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32894 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32896 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32899 * dwarf2out.c (add_data_member_location_attribute): For constant
32900 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32901 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32902 DW_AT_byte_size attributes.
32904 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32906 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32907 after forcing to constant memory when the code model is medium.
32909 2017-01-17 Julia Koval <julia.koval@intel.com>
32912 * config/i386/avx512fintrin.h
32913 (_mm512_i32gather_ps): Change __addr type to void const*.
32914 (_mm512_mask_i32gather_ps): Ditto.
32915 (_mm512_i32gather_pd): Ditto.
32916 (_mm512_mask_i32gather_pd): Ditto.
32917 (_mm512_i64gather_ps): Ditto.
32918 (_mm512_mask_i64gather_ps): Ditto.
32919 (_mm512_i64gather_pd): Ditto.
32920 (_mm512_mask_i64gather_pd): Ditto.
32921 (_mm512_i32gather_epi32): Ditto.
32922 (_mm512_mask_i32gather_epi32): Ditto.
32923 (_mm512_i32gather_epi64): Ditto.
32924 (_mm512_mask_i32gather_epi64): Ditto.
32925 (_mm512_i64gather_epi32): Ditto.
32926 (_mm512_mask_i64gather_epi32): Ditto.
32927 (_mm512_i64gather_epi64): Ditto.
32928 (_mm512_mask_i64gather_epi64): Ditto.
32929 (_mm512_i32scatter_ps): Change __addr type to void*.
32930 (_mm512_mask_i32scatter_ps): Ditto.
32931 (_mm512_i32scatter_pd): Ditto.
32932 (_mm512_mask_i32scatter_pd): Ditto.
32933 (_mm512_i64scatter_ps): Ditto.
32934 (_mm512_mask_i64scatter_ps): Ditto.
32935 (_mm512_i64scatter_pd): Ditto.
32936 (_mm512_mask_i64scatter_pd): Ditto.
32937 (_mm512_i32scatter_epi32): Ditto.
32938 (_mm512_mask_i32scatter_epi32): Ditto.
32939 (_mm512_i32scatter_epi64): Ditto.
32940 (_mm512_mask_i32scatter_epi64): Ditto.
32941 (_mm512_i64scatter_epi32): Ditto.
32942 (_mm512_mask_i64scatter_epi32): Ditto.
32943 (_mm512_i64scatter_epi64): Ditto.
32944 (_mm512_mask_i64scatter_epi64): Ditto.
32945 * config/i386/avx512pfintrin.h
32946 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32947 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32948 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32949 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32950 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32951 (_mm512_prefetch_i32scatter_ps): Ditto.
32952 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32953 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32954 (_mm512_prefetch_i64scatter_pd): Ditto.
32955 (_mm512_prefetch_i64scatter_ps): Ditto.
32956 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32957 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32958 * config/i386/avx512vlintrin.h
32959 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32960 (_mm_mmask_i32gather_ps): Ditto.
32961 (_mm256_mmask_i32gather_pd): Ditto.
32962 (_mm_mmask_i32gather_pd): Ditto.
32963 (_mm256_mmask_i64gather_ps): Ditto.
32964 (_mm_mmask_i64gather_ps): Ditto.
32965 (_mm256_mmask_i64gather_pd): Ditto.
32966 (_mm_mmask_i64gather_pd): Ditto.
32967 (_mm256_mmask_i32gather_epi32): Ditto.
32968 (_mm_mmask_i32gather_epi32): Ditto.
32969 (_mm256_mmask_i32gather_epi64): Ditto.
32970 (_mm_mmask_i32gather_epi64): Ditto.
32971 (_mm256_mmask_i64gather_epi32): Ditto.
32972 (_mm_mmask_i64gather_epi32): Ditto.
32973 (_mm256_mmask_i64gather_epi64): Ditto.
32974 (_mm_mmask_i64gather_epi64): Ditto.
32975 (_mm256_i32scatter_ps): Change __addr type to void*.
32976 (_mm256_mask_i32scatter_ps): Ditto.
32977 (_mm_i32scatter_ps): Ditto.
32978 (_mm_mask_i32scatter_ps): Ditto.
32979 (_mm256_i32scatter_pd): Ditto.
32980 (_mm256_mask_i32scatter_pd): Ditto.
32981 (_mm_i32scatter_pd): Ditto.
32982 (_mm_mask_i32scatter_pd): Ditto.
32983 (_mm256_i64scatter_ps): Ditto.
32984 (_mm256_mask_i64scatter_ps): Ditto.
32985 (_mm_i64scatter_ps): Ditto.
32986 (_mm_mask_i64scatter_ps): Ditto.
32987 (_mm256_i64scatter_pd): Ditto.
32988 (_mm256_mask_i64scatter_pd): Ditto.
32989 (_mm_i64scatter_pd): Ditto.
32990 (_mm_mask_i64scatter_pd): Ditto.
32991 (_mm256_i32scatter_epi32): Ditto.
32992 (_mm256_mask_i32scatter_epi32): Ditto.
32993 (_mm_i32scatter_epi32): Ditto.
32994 (_mm_mask_i32scatter_epi32): Ditto.
32995 (_mm256_i32scatter_epi64): Ditto.
32996 (_mm256_mask_i32scatter_epi64): Ditto.
32997 (_mm_i32scatter_epi64): Ditto.
32998 (_mm_mask_i32scatter_epi64): Ditto.
32999 (_mm256_i64scatter_epi32): Ditto.
33000 (_mm256_mask_i64scatter_epi32): Ditto.
33001 (_mm_i64scatter_epi32): Ditto.
33002 (_mm_mask_i64scatter_epi32): Ditto.
33003 (_mm256_i64scatter_epi64): Ditto.
33004 (_mm256_mask_i64scatter_epi64): Ditto.
33005 (_mm_i64scatter_epi64): Ditto.
33006 (_mm_mask_i64scatter_epi64): Ditto.
33007 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33008 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33009 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33010 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33011 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33012 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33013 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33014 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33015 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33016 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33017 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33018 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33019 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33020 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33021 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33022 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33023 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33024 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33025 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33026 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33027 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33028 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33029 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33030 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33031 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33032 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33033 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33034 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33035 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33036 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33037 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33038 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33039 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33040 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33041 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33042 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33043 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33044 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33045 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33046 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33047 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33048 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33049 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33050 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33051 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33052 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33053 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33054 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33055 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33056 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33057 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33058 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33059 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33060 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33061 definitions accordingly.
33063 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33064 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33067 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33070 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33073 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33074 hard regs for inner regno.
33076 2017-01-17 Martin Liska <mliska@suse.cz>
33079 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33080 assumption and add comment.
33082 2017-01-17 Nathan Sidwell <nathan@acm.org>
33084 * ipa-visibility.c (localize_node): New function, broken out of ...
33085 (function_and_variable_visibility): ... here. Call it.
33087 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33089 PR middle-end/77445
33090 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33091 correctly set frequency of oudgoing edge.
33092 (duplicate_thread_path): Fix profile updating.
33094 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33097 * configure.ac: Add GCC_BASE_VER.
33098 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33099 version from BASE-VER file.
33100 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33101 (gcc.o): Depend on $(BASEVER).
33102 * common.opt (dumpfullversion): New option.
33103 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33104 * doc/invoke.texi: Document -dumpfullversion.
33105 * doc/install.texi: Document --with-gcc-major-version-only.
33106 * configure: Regenerated.
33108 2017-01-17 Richard Biener <rguenther@suse.de>
33110 PR tree-optimization/71433
33111 * tree-vrp.c (register_new_assert_for): Merge same asserts
33112 on all incoming edges.
33113 (process_assert_insertions_for): Handle insertions at the
33116 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33118 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33119 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33121 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33124 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33127 2017-01-17 Alan Modra <amodra@gmail.com>
33130 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33131 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33132 symbolic stack limit when pic.
33134 2017-01-16 Martin Sebor <msebor@redhat.com>
33136 PR tree-optimization/78608
33137 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33139 2017-01-16 Jeff Law <law@redhat.com>
33142 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33143 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33144 for several include directories that may be relative to sysroot.
33145 * config/i386/x-mingw32 (gplus_includedir): Define.
33146 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33147 (native_system_includedir): Likewise.
33148 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33149 override if TARGET_SYSTEM_ROOT is defined.
33150 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33152 PR tree-optimization/79090
33153 PR tree-optimization/33562
33154 PR tree-optimization/61912
33155 PR tree-optimization/77485
33156 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33157 and computed trims into the dump file.
33159 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33161 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33163 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33166 * gimplify.c (gimplify_init_constructor): If want_value and
33167 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33171 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33172 sequence. Formatting fixes.
33173 (doloop_optimize): Formatting fixes.
33176 * gcc.c (debug_level_greater_than_spec_func): New function.
33177 (static_spec_functions): Add debug-level-gt spec function.
33178 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33180 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33181 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33182 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33183 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33184 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33185 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33187 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33189 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33190 QImode fixups to general and mask registers only.
33192 2017-01-16 Carl Love <cel@us.ibm.com>
33194 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33195 for built-in functions
33196 vector signed char vec_nabs (vector signed char)
33197 vector signed short vec_nabs (vector signed short)
33198 vector signed int vec_nabs (vector signed int)
33199 vector signed long long vec_nabs (vector signed long long)
33200 vector float vec_nabs (vector float)
33201 vector double vec_nabs (vector double)
33202 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33204 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33205 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33206 * doc/extend.texi: Update the documentation file for the new built-in
33209 2017-01-16 Martin Sebor <msebor@redhat.com>
33211 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33214 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33216 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33217 UNSPEC_VSX__XXSPLTD to require special splat handling.
33219 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33222 * system.h: Poison strndup.
33224 2017-01-16 Alan Modra <amodra@gmail.com>
33227 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33230 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33232 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33234 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33236 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33237 call recog here. Assert that INSN_CODE (insn) is non-negative.
33239 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33242 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33244 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33245 in the currently scheduled RTL fragment.
33247 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33249 PR rtl-optimization/78751
33250 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33253 2017-01-14 Jeff Law <law@redhat.com>
33255 PR tree-optimization/79090
33256 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33257 variable length stores.
33258 (compute_trims): Delete dead assignment to *trim_tail.
33259 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33262 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33264 PR rtl-optimization/78626
33265 PR rtl-optimization/78727
33266 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33267 of a block, and split such blocks after everything else is finished.
33269 2017-01-14 Alan Modra <amodra@gmail.com>
33272 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33273 target legitimate_combined_insn.
33274 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33275 (rs6000_legitimate_combined_insn): New function.
33276 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33278 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33279 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33280 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33282 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33284 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33286 2017-01-13 Jeff Law <law@redhat.com>
33288 PR tree-optimization/33562
33289 PR tree-optimization/61912
33290 PR tree-optimization/77485
33291 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33293 (delete_dead_assignment): Likewise.
33294 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33295 statement to delete_dead_call and delete_dead_assignment.
33297 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33300 * substring-locations.c (format_warning_va): Strengthen case 1 so
33301 that both endpoints of the substring must be within the format
33302 range for just the substring to be printed.
33304 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33306 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33307 * config/i386/i386.c (ix86_target_string): Add missing options
33308 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33309 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33310 flags_other and ix86_target_other to flags2_other. Display unknown
33312 (ix86_valid_target_attribute_inner_p): Add missing options and
33313 reorder options by implied ISAs, as in ix86_target_string.
33315 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33317 * hash-table.h (hash_table::too_empty_p): New function.
33318 (hash_table::expand): Use it.
33319 (hash_table::traverse): Likewise.
33320 (hash_table::empty_slot): Use sizeof (value_type) instead of
33321 sizeof (PTR) to convert bytes to elements. Shrink the table
33322 if the current size is excessive for the current number of
33325 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33327 * ira-costs.c (record_reg_classes): Break from the inner loop
33328 early once alt_fail is known to be true. Update outer loop
33329 handling accordingly.
33331 2017-01-13 Jeff Law <law@redhat.com>
33333 * tree-ssa-dse.c (decrement_count): New function.
33334 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33335 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33336 when we know the partially dead statement is a mem* function.
33338 PR tree-optimization/61912
33339 PR tree-optimization/77485
33340 * tree-ssa-dse.c: Include expr.h.
33341 (maybe_trim_constructor_store): New function.
33342 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33344 PR tree-optimization/33562
33345 PR tree-optimization/61912
33346 PR tree-optimization/77485
33347 * doc/invoke.texi: Document new dse-max-object-size param.
33348 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33349 * tree-ssa-dse.c: Include params.h.
33350 (dse_store_status): New enum.
33351 (initialize_ao_ref_for_dse): New, partially extracted from
33353 (valid_ao_ref_for_dse, normalize_ref): New.
33354 (setup_live_bytes_from_ref, compute_trims): Likewise.
33355 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33356 (maybe_trim_partially_dead_store): Likewise.
33357 (maybe_trim_complex_store): Likewise.
33358 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33359 Track what bytes live from the original store. Return tri-state
33360 for dead, partially dead or live.
33361 (dse_dom_walker): Add constructor, destructor and new private members.
33362 (delete_dead_call, delete_dead_assignment): New extracted from
33364 (dse_optimize_stmt): Make a member of dse_dom_walker.
33365 Use initialize_ao_ref_for_dse.
33367 PR tree-optimization/33562
33368 PR tree-optimization/61912
33369 PR tree-optimization/77485
33370 * sbitmap.h (bitmap_count_bits): Prototype.
33371 (bitmap_clear_range, bitmap_set_range): Likewise.
33372 * sbitmap.c (bitmap_clear_range): New function.
33373 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33375 2017-01-13 Martin Liska <mliska@suse.cz>
33378 * function.c (set_cfun): Add new argument force.
33379 * function.h (set_cfun): Likewise.
33380 * ipa-inline-transform.c (inline_call): Use the function when
33381 strict alising from is dropped for function we inline to.
33383 2017-01-13 Richard Biener <rguenther@suse.de>
33385 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33386 for dumping GIMPLE INTEGER_CSTs.
33388 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33390 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33391 to 201112L since C++17.
33393 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33396 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33397 if -fsanitize=kernel-address is present.
33399 2017-01-13 Richard Biener <rguenther@suse.de>
33401 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33402 as _Literal ( type ) number in case usual suffixes do not
33403 preserve all information.
33405 2017-01-13 Richard Biener <rguenther@suse.de>
33407 PR tree-optimization/77283
33408 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33409 and ssa-iterators.h.
33410 (is_feasible_trace): Implement a cost model based on joiner
33413 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33416 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33417 char or short to __float128/_Float128 directly.
33419 2017-01-12 Martin Sebor <msebor@redhat.com>
33421 to -Wformat-overflow.
33422 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33423 (min_bytes_remaining): Same.
33424 (get_string_length): Same.
33425 (format_string): Same.
33426 (format_directive): Same.
33428 (pass_sprintf_length::handle_gimple_call): Same.
33430 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33432 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33433 info.nowrite calls with no lhs that can't throw. Return bool
33434 whether gsi_remove has been called or not.
33435 (pass_sprintf_length::handle_gimple_call): Return bool whether
33436 try_substitute_return_value called gsi_remove. Formatting fix.
33437 (pass_sprintf_length::execute): Don't use gsi_remove if
33438 handle_gimple_call returned true.
33441 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33442 be removed due to side-effects, don't remove following barrier nor
33443 turn the successor edge into fallthru edge.
33445 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33448 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33449 element-reversing loads and stores as not swappable.
33451 2017-01-12 Nathan Sidwell <nathan@acm.org>
33452 Nicolai Stange <nicstange@gmail.com>
33454 * combine.c (try_combine): Don't ignore result of overlap checking
33455 loop. Combine overlap & asm check into single loop.
33457 2017-01-12 Richard Biener <rguenther@suse.de>
33459 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33460 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33462 2017-01-12 Richard Biener <rguenther@suse.de>
33464 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33465 and TS_TARGET_OPTION directly derive from TS_BASE.
33466 * tree-core.h (tree_optimization_option): Derive from tree_base.
33467 (tree_target_option): Likewise.
33469 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33471 * config/i386/i386.c (memory_address_length): Increase len
33472 only when rip_relative_addr_p returns false.
33474 2017-01-11 Julia Koval <julia.koval@intel.com>
33476 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33477 (OPTION_MASK_ISA_SGX_SET): New.
33478 (ix86_handle_option): Handle OPT_msgx.
33479 * config.gcc: Added sgxintrin.h.
33480 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33481 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33482 * config/i386/i386.c (ix86_target_string): Add -msgx.
33484 (ix86_option_override_internal): Handle new options.
33485 (ix86_valid_target_attribute_inner_p): Add sgx.
33486 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33487 * config/i386/i386.opt: Add msgx.
33488 * config/i386/sgxintrin.h: New file.
33489 * config/i386/x86intrin.h: Add sgxintrin.h.
33491 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33494 * fold-const.c (maybe_nonzero_address): Return 1 for function
33496 (tree_single_nonzero_warnv_p): Don't handle function local objects
33500 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33503 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33506 * opts.c: Include "spellcheck.h"
33507 (struct string_fragment): New struct.
33508 (struct edit_distance_traits<const string_fragment &>): New
33510 (get_closest_sanitizer_option): New function.
33511 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33513 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33515 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33517 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33518 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33519 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33520 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33521 for initial die_offset if dwarf_split_debug_info.
33522 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33523 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33525 (output_skeleton_debug_sections): Formatting fix. Use
33526 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33527 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33529 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33531 * config/arm/cortex-a53.md: Add bypasses for
33532 cortex_a53_r2f_cvt.
33533 (cortex_a53_r2f): Only use for transfers.
33534 (cortex_a53_f2r): Likewise.
33535 (cortex_a53_r2f_cvt): Add reservation for conversions.
33536 (cortex_a53_f2r_cvt): Likewise.
33538 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33540 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33541 to all inlined functions, change static to extern.
33543 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33546 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33548 (arm_assemble_integer): Likewise.
33550 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33552 * config.gcc: Use new awk script to check CPU, FPU and architecture
33553 parameters for --with-... options.
33554 * config/arm/parsecpu.awk: New file
33555 * config/arm/arm-cpus.in: New file.
33556 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33558 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33560 * config/arm/t-arm: Update dependency rules.
33561 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33562 of processing .def files.
33563 * config/arm/genopt.sh: Deleted.
33564 * config/arm/gentune.sh: Deleted.
33565 * config/arm/arm-cores.def: Deleted.
33566 * config/arm/arm-arches.def: Deleted.
33567 * config/arm/arm-fpus.def: Deleted.
33568 * config/arm/arm-tune.md: Regenerated.
33569 * config/arm/arm-tables.opt: Regenerated.
33570 * config/arm/arm-cpu.h: New generated file.
33571 * config/arm/arm-cpu-data.h: New generated file.
33572 * config/arm/arm-cpu-cdata.h: New generated file.
33574 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33577 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33579 (input_varpool_node): Unpack dynamically_initialized bit.
33581 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33583 PR rtl-optimization/79032
33584 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33585 the alignment of the adjusted memory reference against that of MODE,
33586 instead of the alignment of the original memory reference.
33588 2017-01-11 Martin Jambor <mjambor@suse.cz>
33590 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33592 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33593 decorated functions.
33595 2017-01-11 Richard Biener <rguenther@suse.de>
33597 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33598 set range/nonnull info for PHI results. Do not set it on
33599 stmts marked for removal.
33601 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33603 * expr.c (store_field): In the bitfield case, fetch the return value
33604 from the registers before applying a single big-endian adjustment.
33605 Always do a final load for a BLKmode value not larger than a word.
33607 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33610 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33611 that we correctly handle column numbers greater than
33612 LINE_MAP_MAX_COLUMN_NUMBER.
33614 2017-01-10 Martin Sebor <msebor@redhat.com>
33616 PR middle-end/78245
33617 * gimple-ssa-sprintf.c (get_destination_size): Call
33618 {init,fini}object_sizes.
33619 * tree-object-size.c (addr_object_size): Adjust.
33620 (pass_through_call): Adjust.
33621 (pass_object_sizes::execute): Adjust.
33622 * tree-object-size.h (fini_object_sizes): Declare.
33624 2017-01-10 Martin Sebor <msebor@redhat.com>
33626 PR tree-optimization/78775
33627 * builtins.c (get_size_range): Move...
33628 * calls.c: ...to here.
33629 (alloc_max_size): Accept zero argument.
33630 (operand_signed_p): Remove.
33631 (maybe_warn_alloc_args_overflow): Call get_size_range.
33632 * calls.h (get_size_range): Declare.
33634 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33636 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33637 from TI's devices.csv file as of September 2016.
33638 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33640 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33642 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33643 * doc/invoke.texi: Likewise.
33644 * doc/md.texi: Likewise.
33645 * doc/objc.texi: Likewise.
33647 2017-01-10 Joshua Conner <joshconner@google.com>
33649 * config/arm/fuchsia-elf.h: New file.
33650 * config/fuchsia.h: New file.
33651 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33652 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33654 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33656 2016-01-10 Richard Biener <rguenther@suse.de>
33658 PR tree-optimization/79034
33659 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33660 Propagate out degenerate PHIs in the joiner.
33662 2017-01-10 Martin Liska <mliska@suse.cz>
33664 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33665 (sort_congruence_classes_by_decl_uid): Likewise.
33666 (sort_congruence_class_groups_by_decl_uid): Likewise.
33667 (sem_item_optimizer::merge_classes): Sort class, groups in these
33668 classes and members in the groups by DECL_UID of declarations.
33669 This would make merge operations stable.
33671 2017-01-10 Martin Liska <mliska@suse.cz>
33673 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33674 usage of m_classes_vec.
33675 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33676 (sem_item_optimizer::get_group_by_hash): Likewise.
33677 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33678 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33679 (sem_item_optimizer::verify_classes): Likewise.
33680 (sem_item_optimizer::process_cong_reduction): Likewise.
33681 (sem_item_optimizer::dump_cong_classes): Likewise.
33682 (sem_item_optimizer::merge_classes): Likewise.
33683 * ipa-icf.h (congruence_class_hash): Rename from
33684 congruence_class_group_hash. Remove declaration of m_classes_vec.
33686 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33688 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33689 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33690 * config.gcc: Add avx512vpopcntdqintrin.h.
33691 * config/i386/avx512vpopcntdqintrin.h: New.
33692 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33693 * config/i386/i386-builtin-types.def: Add new types.
33694 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33695 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33696 __builtin_ia32_vpopcountq_v8di_mask): New.
33697 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33698 __AVX512VPOPCNTDQ__.
33699 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33700 (PTA_AVX512VPOPCNTDQ): Define.
33701 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33702 TARGET_AVX512VPOPCNTDQ_P): Define.
33703 * config/i386/i386.opt: Add mavx512vpopcntdq.
33704 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33705 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33707 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33709 PR middle-end/77484
33710 * predict.def (PRED_CALL): Set to 67.
33712 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33714 * expr.c (store_field): In the bitfield case, if the value comes from
33715 a function call and is of an aggregate type returned in registers, do
33716 not modify the field mode; extract the value in all cases if the mode
33717 is BLKmode and the size is not larger than a word.
33719 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33722 * config/i386/cpuid.h: Fix undefined behavior.
33724 2017-01-04 Jeff Law <law@redhat.com>
33726 PR tree-optimization/79007
33727 PR tree-optimization/67955
33728 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33729 conservative for pt.null when flag_non_call_exceptions is on.
33731 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33733 PR translation/79019
33734 PR translation/79020
33735 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33736 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33737 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33739 * config/avr/avr.opt (maccumulate-args): Likewise.
33740 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33741 * common.opt (freport-bug): Likewise.
33742 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33743 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33744 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33745 translatable string.
33746 * config/i386/i386.c (function_value_32): Likewise.
33747 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33748 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33750 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33751 * common/config/msp430/msp430-common.c (msp430_handle_option):
33753 * symtab.c (symtab_node::verify_base): Likewise.
33754 * opts.c (set_debug_level): Likewise.
33755 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33756 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33757 missing whitespace to translatable strings.
33758 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33759 * config/sh/superh.h: Likewise.
33760 * config/i386/xopintrin.h: Likewise.
33761 * config/i386/znver1.md: Likewise.
33762 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33763 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33764 * double-int.h (struct double_int): Likewise.
33765 * double-int.c (div_and_round_double): Likewise.
33766 * wide-int.cc: Likewise.
33767 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33768 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33769 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33770 (crossjumps_occurred): ... this.
33771 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33774 PR tree-optimization/78899
33775 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33776 returning bool return struct loop *, NULL for failure and the new
33778 (versionable_outer_loop_p): Don't version outer loop if it has
33779 dont_vectorized bit set.
33780 (tree_if_conversion): When versioning outer loop, ensure
33781 tree_if_conversion is performed also on the inner loop of the
33782 non-vectorizable outer loop copy.
33783 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33784 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33785 prevent vectorization of it.
33786 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33787 the outer loop vectorization of the non-scalar version is attempted
33788 before vectorization of the inner loop in scalar version. If
33789 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33790 vectorization of its inner loop.
33791 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33792 has 2 inner loops, rename also on edges from bb whose single pred
33793 is outer_loop->header. Fix typo in function comment.
33795 2017-01-09 Martin Sebor <msebor@redhat.com>
33798 * asan.c (asan_emit_stack_protection): Increase local buffer size
33799 to avoid snprintf truncation warning.
33801 2017-01-09 Andrew Pinski <apinski@cavium.com>
33803 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33804 to reference thunderx2t99 for the tuning structure
33805 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33807 (thunderx2t99_extra_costs): This.
33808 * config/aarch64/aarch64-tune.md: Regenerate.
33809 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33810 (vulcan_addrcost_table): This.
33811 (vulcan_regmove_cost): Rename to ...
33812 (thunderx2t99_regmove_cost): This.
33813 (vulcan_vector_cost): Rename to ...
33814 (thunderx2t99_vector_cost): this.
33815 (vulcan_branch_cost): Rename to ...
33816 (thunderx2t99_branch_cost): This.
33817 (vulcan_tunings): Rename to ...
33818 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33819 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33821 2017-01-09 Martin Jambor <mjambor@suse.cz>
33825 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33826 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33827 (propagate_vr_accross_jump_function): Use the above function for all
33828 value range computations for pass-through jump functions and type
33829 converasion from explicit value range values.
33830 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33831 parameters from TYPE_ARG_TYPES.
33832 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33833 (ipa_write_node_info): Stream type of the actual argument.
33834 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33836 2017-01-09 Martin Liska <mliska@suse.cz>
33839 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33840 (lookup_compiler): Do not show error message with have_E.
33842 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33844 PR tree-optimization/78938
33845 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33846 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33847 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33848 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33851 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33853 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33856 2017-01-09 Richard Biener <rguenther@suse.de>
33858 PR tree-optimization/78997
33859 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33860 name condition properly.
33862 2017-01-09 Richard Biener <rguenther@suse.de>
33865 * dwarf2out.c (is_cxx): New overload with context.
33866 (is_naming_typedef_decl): Use it.
33868 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33870 * invoke.texi (Option Summary): Correct spacing in option lists
33871 and add line breaks to fix over-long lines.
33873 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33875 PR middle-end/17660
33877 * extend.texi (Common Variable Attributes): Add xref to GCC
33878 Internals manual to explain mode attribute keywords.
33880 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33883 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33884 and Preprocessor Options.
33885 (Options for Linking): Document -pthread here....
33886 (RS/6000 and PowerPC Options): ...not here.
33887 (Solaris 2 Options): ...or here.
33888 * doc/cppopts.texi: Document -pthread.
33890 2017-01-08 Martin Sebor <msebor@redhat.com>
33892 PR middle-end/77708
33893 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33894 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33895 New member functions.
33896 (format_directive): Used them.
33898 (pass_sprintf_length::handle_gimple_call): Same.
33899 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33900 to avoid truncation for any argument.
33901 (extract_affine_mul): Same.
33902 * tree.c (get_file_function_name): Same.
33904 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33906 PR middle-end/77484
33907 * predict.def (PRED_INDIR_CALL): Set to 86.
33909 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33911 PR preprocessor/54124
33912 * doc/cppopts.texi: Reformat -d subtable to list the full name
33913 of the options. Add cross-reference to the docs for the general
33914 compiler -d options.
33915 * doc/invoke.texi (Developer Options): Add cross-reference to the
33916 preprocessor-specific -d option documentation.
33918 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33920 PR preprocessor/13498
33921 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33922 redudant material, and reflect new command-line options.
33923 (System Headers): Likewise.
33925 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33927 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33928 -isystem, and -idirafter. Copy-edit.
33929 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33930 default for -ftrack-macro-expansion. Delete obsolete and
33931 badly-formatted implementation details about -fdebug-cpp output.
33932 * doc/cppwarnopts.texi: Copy-edit.
33934 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33937 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33938 that the transition from a max line width >= 1<<10 to narrower
33939 lines works correctly.
33941 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33943 * doc/options.texi (PerFunction): New.
33944 * opt-functions.awk (switch_flags): Map both Optimization and
33945 PerFunction to CL_OPTIMIZATION.
33946 * opth-gen.awk: Test for PerFunction flag along with
33948 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33949 it only when the latter is present. Skip those that don't in
33950 the hash function generator.
33951 * common.opt (fvar-tracking): Mark as PerFunction instead of
33953 (fvar-tracking-assignments): Likewise.
33954 (fvar-tracking-assignments-toggle): Likewise.
33955 (fvar-tracking-uninit): Likewise.
33957 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33959 PR translation/79018
33960 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33963 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33966 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33967 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33968 TARGET_LONG_JUMP_TABLE_OFFSETS.
33969 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33970 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33971 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33972 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33973 * config/m68k/m68k.md (tablejump expander): Likewise.
33974 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33975 TARGET_LONG_JUMP_TABLE_OFFSETS.
33976 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33977 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33979 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33980 David Holsgrove <david.holsgrove@xilinx.com>
33982 * common/config/microblaze/microblaze-common.c
33983 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33984 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33986 * config/microblaze/microblaze.c (microblaze_must_save_register)
33987 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33989 (microblaze_eh_return): New function.
33990 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33991 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33992 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33993 * config/microblaze/microblaze.md (eh_return): New pattern.
33995 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33997 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33998 GCC_DIAGNOSTIC_STRINGIFY): Define.
34000 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34002 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34004 * config/arm/arm.md (<mcrr>): New.
34006 * config/arm/arm.c (arm_arch5te): New.
34007 (arm_option_override): Set arm_arch5te.
34008 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34010 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34011 (arm_mcrr_qualifiers): ... this. New.
34012 (MRRC_QUALIFIERS): Define to...
34013 (arm_mrrc_qualifiers): ... this. New.
34014 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34016 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34017 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34018 (MRRCI, mrrc, MRRC): New.
34019 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34020 VUNSPEC_MRRC2): New.
34022 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34024 * config/arm/arm.md (<mcr>): New.
34026 * config/arm/arm.c (arm_coproc_builtin_available): Add
34027 support for mcr, mrc, mcr2 and mrc2.
34028 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34029 (arm_mcr_qualifiers): ... this. New.
34030 (MRC_QUALIFIERS): Define to ...
34031 (arm_mrc_qualifiers): ... this. New.
34032 (MCR_QUALIFIERS): Define to ...
34033 (arm_mcr_qualifiers): ... this. New.
34034 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34036 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34037 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34038 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34039 VUNSPEC_MRC2): New.
34041 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34043 * config/arm/arm.md (*ldc): New.
34047 * config/arm/arm.c (arm_coproc_builtin_available): Add
34048 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34049 (arm_coproc_ldc_stc_legitimate_address): New.
34050 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34051 'qualifier_const_pointer'.
34052 (LDC_QUALIFIERS): Define to...
34053 (arm_ldc_qualifiers): ... this. New.
34054 (STC_QUALIFIERS): Define to...
34055 (arm_stc_qualifiers): ... this. New.
34056 * config/arm/arm-protos.h
34057 (arm_coproc_ldc_stc_legitimate_address): New.
34058 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34059 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34060 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34061 stc2, stcl, stc2l): New.
34062 * config/arm/constraints.md (Uz): New.
34063 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34064 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34065 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34066 VUNSPEC_STC2L): New.
34068 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34070 * config/arm/arm.md (<cdp>): New.
34071 * config/arm/arm.c (neon_const_bounds): Rename this ...
34072 (arm_const_bounds): ... this.
34073 (arm_coproc_builtin_available): New.
34074 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34075 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34076 (CDP_QUALIFIERS): Define to...
34077 (arm_cdp_qualifiers): ... this. New.
34079 (arm_expand_builtin_args): Add case for 6 arguments.
34080 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34081 (arm_const_bounds): ... this.
34082 (arm_coproc_builtin_available): New.
34083 * config/arm/arm_acle.h (__arm_cdp): New.
34085 * config/arm/arm_acle_builtins.def (cdp): New.
34087 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34088 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34089 'arm_const_bounds'.
34090 * config/arm/types.md (coproc): New.
34091 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34092 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34093 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34094 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34096 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34098 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34099 (UBINOP_QUALIFIERS): New.
34101 (acle_builtin_data): New. Change comment.
34102 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34103 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34104 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34105 arm_acle_builtins.def.
34106 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34107 (arm_init_acle_builtins): New.
34108 (CRC32_BUILTIN): Remove.
34109 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34110 crc32cb, crc32ch and crc32cw.
34111 (arm_init_crc32_builtins): Remove.
34112 (arm_init_builtins): Use arm_init_acle_builtins rather
34113 than arm_init_crc32_builtins.
34114 (arm_expand_acle_builtin): New.
34115 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34116 * config/arm/arm_acle_builtins.def: New.
34118 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34120 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34121 (arm_builtin_datum): ... this.
34122 (arm_init_neon_builtin): Rename to ...
34123 (arm_init_builtin): ... this. Add a new parameters PREFIX
34124 and USE_SIG_IN_NAME.
34125 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34126 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34127 'arm_builtin_datum'.
34128 (arm_init_vfp_builtins): Likewise.
34129 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34130 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34131 (arm_expand_neon_args): Rename to ...
34132 (arm_expand_builtin_args): ... this. Rename builtin_arg
34133 enum values and differentiate between ARG_BUILTIN_MEMORY
34134 and ARG_BUILTIN_NEON_MEMORY.
34135 (arm_expand_neon_builtin_1): Rename to ...
34136 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34137 values, arm_expand_builtin_args and add bool parameter NEON.
34138 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34139 (arm_expand_vfp_builtin): Likewise.
34140 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34142 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34144 PR middle-end/77484
34145 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34146 * predict.c (tree_estimate_probability_bb): Reverse direction of
34147 polymorphic call predictor.
34149 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34151 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34152 (determine_pass_name_match): ...this new function and...
34153 (should_skip_pass_p): ...this new function.
34155 2017-01-06 Nathan Sidwell <nathan@acm.org>
34157 * ipa-visibility.c (function_and_variable_visibility): Reformat
34158 comments and long lines. Remove extrneous if.
34159 * symtab.c (symtab_node::make_decl_local): Fix code format.
34160 (symtab_node::set_section_for_node): Fix comment typo.
34162 2017-01-06 Martin Liska <mliska@suse.cz>
34165 * lra-constraints.c: Rename invariant to lra_invariant.
34166 * predict.c (set_even_probabilities): Initialize e to NULL.
34168 2017-01-05 Martin Sebor <msebor@redhat.com>
34170 PR tree-optimization/78910
34171 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34172 (format_integer): Correct off-by-one error in the handling
34173 of precision with negative numbers in signed conversions..
34175 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34177 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34179 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34181 PR tree-optimization/71016
34182 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34183 factor_out_conditional_conversion. Formatting fix.
34184 (factor_out_conditional_conversion): Add cond_stmt argument.
34185 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34186 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34189 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34191 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34192 read-rtl-function.o, and selftest-rtl.o.
34193 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34194 (selftest::aarch64_test_loading_full_dump): New function.
34195 (selftest::aarch64_run_selftests): New function.
34196 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34197 selftest::aarch64_run_selftests.
34198 * config/i386/i386.c
34199 (selftest::ix86_test_loading_dump_fragment_1): New function.
34200 (selftest::ix86_test_loading_call_insn): New function.
34201 (selftest::ix86_test_loading_full_dump): New function.
34202 (selftest::ix86_test_loading_unspec): New function.
34203 (selftest::ix86_run_selftests): Call the new functions.
34204 * emit-rtl.c (maybe_set_max_label_num): New function.
34205 * emit-rtl.h (maybe_set_max_label_num): New decl.
34206 * function.c (instantiate_decls): Guard call to
34207 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34208 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34210 * gensupport.c (gen_reader::gen_reader): Pass "false"
34211 for new "compact" param of rtx_reader.
34212 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34213 rather than an empty string for NULL strings.
34214 * read-md.c: Potentially include config.h rather than bconfig.h.
34215 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34216 (have_error): New global, copied from errors.c.
34217 (md_reader::read_name): Rename to...
34218 (md_reader::read_name_1): ...this, adding "out_loc" param,
34219 and converting "missing name or number" to returning false, rather
34221 (md_reader::read_name): Reimplement in terms of read_name_1.
34222 (md_reader::read_name_or_nil): New function.
34223 (md_reader::read_string): Handle "(nil)" by returning NULL.
34224 (md_reader::md_reader): Add new param "compact".
34225 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34226 (md_reader::read_file): New method.
34227 * read-md.h (md_reader::md_reader): Add new param "compact".
34228 (md_reader::read_file): New method.
34229 (md_reader::is_compact): New accessor.
34230 (md_reader::read_name): Convert return type from void to file_location.
34231 (md_reader::read_name_or_nil): New decl.
34232 (md_reader::read_name_1): New decl.
34233 (md_reader::m_compact): New field.
34234 (noop_reader::noop_reader): Pass "false" for new "compact" param
34236 (rtx_reader::rtx_reader): Add new "compact" param.
34237 (rtx_reader::read_rtx_operand): Make virtual and convert return
34238 type from void to rtx.
34239 (rtx_reader::read_until): New decl.
34240 (rtx_reader::handle_any_trailing_information): New virtual function.
34241 (rtx_reader::postprocess): New virtual function.
34242 (rtx_reader::finalize_string): New virtual function.
34243 (rtx_reader::m_in_call_function_usage): New field.
34244 (rtx_reader::m_reuse_rtx_by_id): New field.
34245 * read-rtl-function.c: New file.
34246 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34247 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34248 (selftest::verify_three_block_rtl_cfg): New decl.
34249 * read-rtl-function.h: New file.
34250 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34251 For host, include function.h, memmodel.h, and emit-rtl.h.
34252 (one_time_initialization): New function.
34253 (struct compact_insn_name): New struct.
34254 (compact_insn_names): New array.
34255 (find_code): Handle insn codes in compact dumps.
34256 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34257 (bind_subst_iter_and_attr): Likewise.
34258 (add_condition_to_string): Likewise.
34259 (add_condition_to_rtx): Likewise.
34260 (apply_attribute_uses): Likewise.
34261 (add_current_iterators): Likewise.
34262 (apply_iterators): Likewise.
34263 (initialize_iterators): Guard usage of apply_subst_iterator with
34264 #ifdef GENERATOR_FILE.
34265 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34266 (md_reader::read_mapping): Likewise.
34267 (add_define_attr_for_define_subst): Likewise.
34268 (add_define_subst_attr): Likewise.
34269 (read_subst_mapping): Likewise.
34270 (check_code_iterator): Likewise.
34271 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34273 (one_time_initialization): New function.
34274 (rtx_reader::read_until): New method.
34275 (read_flags): New function.
34276 (parse_reg_note_name): New function.
34277 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34278 Handle reuse_rtx ids.
34279 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34280 Add parsing support for RTL dumps, mirroring the special-cases in
34281 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34282 values, and calling handle_any_trailing_information.
34283 (rtx_reader::read_rtx_operand): Convert return type from void
34284 to rtx, returning return_rtx. Handle case 'e'. Call
34285 finalize_string on XSTR and XTMPL fields.
34286 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34287 "(nil)" values were omitted. Call the postprocess vfunc on the
34289 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34290 class ctor. Initialize m_in_call_function_usage. Call
34291 one_time_initialization.
34292 * rtl-tests.c (selftest::test_uncond_jump): Call
34293 set_new_first_and_last_insn.
34294 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34295 * selftest-rtl.c: New file.
34296 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34297 (selftest::get_insn_by_uid): New decl.
34298 * selftest-run-tests.c (selftest::run_tests): Call
34299 read_rtl_function_c_tests.
34300 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34301 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34304 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34306 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34307 operands in a special way. Assert that pos+len <= mode precision.
34309 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34311 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34312 3 argument Alias with unlimited for the negative form.
34313 (fno-vect-cost-model): Removed.
34315 2017-01-05 Martin Liska <mliska@suse.cz>
34317 * hsa-gen.c (gen_hsa_divmod): New function.
34318 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34320 2017-01-05 Martin Liska <mliska@suse.cz>
34323 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34326 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34328 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34329 small constant length operands.
34331 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34333 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34334 between loop iterations.
34336 2017-01-05 Martin Liska <mliska@suse.cz>
34339 * gimplify.c (gimplify_decl_expr): Compare to
34340 asan_poisoned_variables instread of checking flags.
34341 (gimplify_target_expr): Likewise.
34342 (gimplify_expr): Likewise.
34343 (gimplify_function_tree): Conditionally initialize
34344 asan_poisoned_variables.
34346 2017-01-04 Jeff Law <law@redhat.com>
34348 PR tree-optimizatin/78812
34349 * rtl.h (contains_mem_rtx_p): Prototype.
34350 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34351 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34352 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34353 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34354 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34356 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34358 * input.c (assert_char_at_range): Default-initialize actual_range.
34360 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34362 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34363 to match the caller.
34365 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34367 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34368 insns after final jump in test to emit dummy move.
34370 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34372 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34373 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34375 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34377 * multiple_target.c (create_dispatcher_calls): Init e_next.
34378 * tree-ssa-loop-split.c (split_loop): Init border.
34379 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34382 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34387 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34388 (altivec_register_operand): Do not return true if the operand
34389 contains a SUBREG mixing SImode and SFmode.
34390 (vsx_register_operand): Likewise.
34391 (vsx_reg_sfsubreg_ok): New predicate.
34392 (vfloat_operand): Do not return true if the operand contains a
34393 SUBREG mixing SImode and SFmode.
34394 (vint_operand): Likewise.
34395 (vlogical_operand): Likewise.
34396 (gpc_reg_operand): Likewise.
34397 (int_reg_operand): Likewise.
34398 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34399 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34400 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34402 (rs6000_emit_move_si_sf_subreg): New helper function.
34403 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34404 fixup SUBREGs involving SImode and SFmode.
34405 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34406 numbers for the new peephole2 optimization.
34407 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34408 the GLIBC math library that do AND/IOR/XOR operations on single
34409 precision floating point.
34410 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34411 target macros to say whether we need to avoid SUBREGs mixing
34413 (TARGET_ALLOW_SF_SUBREG): Likewise.
34414 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34415 (UNSPEC_SI_FROM_SF): Likewise.
34416 (iorxor): Change spacing.
34417 (and_ior_xor): New iterator for AND, IOR, and XOR.
34418 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34419 (movdi_from_sf_zero_ext): Likewise.
34420 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34421 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34422 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34423 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34424 (fms<mode>4): Likewise.
34425 (fnma<mode>4): Likewise.
34426 (fnms<mode>4): Likewise.
34427 (nfma<mode>4): Likewise.
34428 (nfms<mode>4): Likewise.
34430 2017-01-04 Marek Polacek <polacek@redhat.com>
34433 * doc/invoke.texi: Document -Wpointer-compare.
34435 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34437 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34440 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34441 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34444 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34447 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34448 documentation of the powerpc_popcntb_ok attribute.
34449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34450 code to issue warning messages if a requested CPU configuration is
34451 not supported by the binary (assembler and loader) toolchain.
34452 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34453 made to define a built-in function that has been disabled.
34454 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34455 made to define a built-in function that has been disabled.
34456 (altivec_init_builtins): Add comment explaining why definition
34457 of the DST built-in functions is not preceded by an assertion
34458 check. Add assertions to prevent ICE if attempts are made to
34459 define an altivec predicate or an abs* built-in function that has
34461 (htm_init_builtins): Add comment explaining why definition of the
34462 htm built-in functions is not preceded by an assertion check.
34464 2017-01-04 Jeff Law <law@redhat.com>
34466 PR tree-optimizatin/67955
34467 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34468 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34469 the points-to solution does not include pt_null. Use DECL_PT_UID
34472 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34474 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34475 Use gen_int_mode instead of gen_lopwart for const_int operands.
34477 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34479 PR tree-optimization/71563
34480 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34481 out of range value - has low bits known to be zero.
34483 2017-01-04 Alan Modra <amodra@gmail.com>
34485 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34486 * configure: Regenerate.
34487 * config.in: Regenerate.
34489 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34492 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34493 a substring of the message, but strcmp with the whole message. Ifdef
34494 ENABLE_NLS, translate the message first using dgettext.
34496 2017-01-03 Jeff Law <law@redhat.com>
34498 PR tree-optimizatin/78856
34499 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34500 (mark_threaded_blocks): Remove code to truncate thread paths that
34501 cross multiple loop headers. Instead invalidate the cached loop
34502 iteration information and handle case of a thread path walking
34503 into an irreducible region.
34505 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34508 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34509 assertions. Add support for doing the signbit if the IEEE 128-bit
34510 floating point value is in a GPR.
34511 * config/rs6000/rs6000.md (Fsignbit): Delete.
34512 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34513 Update the length attribute if the value is in a GPR.
34514 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34515 the sign or zero extension instruction, since the value is always 0/1.
34516 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34519 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34520 extracting SImode to a GPR register so that we can generate a
34521 store, limit the vector to be in a traditional Altivec register
34522 for the vextuwrx instruction.
34524 2017-01-03 Ian Lance Taylor <iant@google.com>
34526 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34528 2017-01-03 Martin Sebor <msebor@redhat.com>
34530 PR tree-optimization/78696
34531 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34532 precision. Use MPFR for %f for greater fidelity. Correct handling
34534 (pass_sprintf_length::compute_format_length): Set width and precision
34535 specified by asrerisk to void_node for vararg functions.
34536 (try_substitute_return_value): Adjust dump output.
34538 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34540 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34542 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34544 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34545 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34546 -mlra/-mno-lra was passed to the compiler.
34548 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34550 PR rtl-optimization/65618
34551 * emit-rtl.c (try_split): Move initialization of "before" and
34552 "after" to just before the call to emit_insn_after_setloc.
34554 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34556 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34558 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34560 * dwarf2out.c (gen_enumeration_type_die): When
34561 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34563 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34565 PR tree-optimization/78965
34566 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34567 Change first argument from const call_info & to call_info &. For %n
34568 set info.nowrite to false.
34570 PR middle-end/78901
34571 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34572 possibly throwing calls.
34574 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34575 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34576 and fns handling, rather than in a separate case SSA_NAME.
34578 2017-01-02 Jeff Law <law@redhat.com>
34580 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34581 fixes for first_period and second_period variables.
34583 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34586 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34587 (*insvqi_1): New insn pattern.
34588 (*insvqi_1_mem_rex64): Ditto.
34589 (*insvqi_2): Ditto.
34590 (*insvqi_3): Rename from *insvqi.
34592 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34594 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34596 * doc/cfg.texi (Edges): Remove reference to Java.
34597 (Maintaining the CFG): Ditto.
34599 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34601 PR middle-end/77674
34602 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34603 transparent aliases.
34605 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34607 PR middle-end/77484
34608 * predict.def (PRED_CALL): Update hitrate.
34609 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34610 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34611 into direct/indirect/polymorphic variants.
34613 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34615 Update copyright years.
34617 * gcc.c (process_command): Update copyright notice dates.
34618 * gcov-dump.c (print_version): Ditto.
34619 * gcov.c (print_version): Ditto.
34620 * gcov-tool.c (print_version): Ditto.
34621 * gengtype.c (create_file): Ditto.
34622 * doc/cpp.texi: Bump @copying's copyright year.
34623 * doc/cppinternals.texi: Ditto.
34624 * doc/gcc.texi: Ditto.
34625 * doc/gccint.texi: Ditto.
34626 * doc/gcov.texi: Ditto.
34627 * doc/install.texi: Ditto.
34628 * doc/invoke.texi: Ditto.
34630 Copyright (C) 2017 Free Software Foundation, Inc.
34632 Copying and distribution of this file, with or without modification,
34633 are permitted in any medium without royalty provided the copyright
34634 notice and this notice are preserved.